diff -Nru slurm-llnl-1.3.13/auxdir/x_ac_affinity.m4 slurm-llnl-1.3.15/auxdir/x_ac_affinity.m4 --- slurm-llnl-1.3.13/auxdir/x_ac_affinity.m4 2007-03-21 21:59:01.000000000 +0100 +++ slurm-llnl-1.3.15/auxdir/x_ac_affinity.m4 2009-01-29 23:44:14.000000000 +0100 @@ -40,6 +40,7 @@ AM_CONDITIONAL(HAVE_NUMA, test "x$ac_have_numa" = "xyes") if test "x$ac_have_numa" = "xyes"; then AC_DEFINE(HAVE_NUMA, 1, [define if numa library installed]) + CFLAGS="-DNUMA_VERSION1_COMPATIBILITY $CFLAGS" else AC_MSG_WARN([Unable to locate NUMA memory affinity functions]) fi diff -Nru slurm-llnl-1.3.13/auxdir/x_ac_bluegene.m4 slurm-llnl-1.3.15/auxdir/x_ac_bluegene.m4 --- slurm-llnl-1.3.13/auxdir/x_ac_bluegene.m4 2009-01-07 21:57:41.000000000 +0100 +++ slurm-llnl-1.3.15/auxdir/x_ac_bluegene.m4 2009-02-26 20:49:53.000000000 +0100 @@ -1,5 +1,5 @@ ##***************************************************************************** -## $Id: x_ac_bluegene.m4 16156 2009-01-07 20:57:41Z jette $ +## $Id: x_ac_bluegene.m4 16697 2009-02-26 19:49:53Z da $ ##***************************************************************************** # AUTHOR: # Morris Jette @@ -80,13 +80,14 @@ # Test to make sure the api is good have_bg_files=yes saved_LDFLAGS="$LDFLAGS" - LDFLAGS="$saved_LDFLAGS $bg_ldflags" + LDFLAGS="$saved_LDFLAGS $bg_ldflags -m64" AC_LINK_IFELSE([AC_LANG_PROGRAM([[ int rm_set_serial(char *); ]], [[ rm_set_serial(""); ]])],[have_bg_files=yes],[AC_MSG_ERROR(There is a problem linking to the BG/L api.)]) LDFLAGS="$saved_LDFLAGS" fi if test ! -z "$have_bg_files" ; then BG_INCLUDES="$bg_includes" + CFLAGS="$CFLAGS -m64" AC_DEFINE(HAVE_3D, 1, [Define to 1 if 3-dimensional architecture]) AC_DEFINE(HAVE_BG, 1, [Define to 1 if emulating or running on Blue Gene system]) AC_DEFINE(HAVE_BGL, 1, [Define to 1 if emulating or running on Blue Gene/L system]) @@ -160,13 +161,14 @@ # ac_with_readline="no" # Test to make sure the api is good saved_LDFLAGS="$LDFLAGS" - LDFLAGS="$saved_LDFLAGS $bg_ldflags" + LDFLAGS="$saved_LDFLAGS $bg_ldflags -m64" AC_LINK_IFELSE([AC_LANG_PROGRAM([[ int rm_set_serial(char *); ]], [[ rm_set_serial(""); ]])],[have_bgp_files=yes],[AC_MSG_ERROR(There is a problem linking to the BG/P api.)]) LDFLAGS="$saved_LDFLAGS" fi if test ! -z "$have_bgp_files" ; then BG_INCLUDES="$bg_includes" + CFLAGS="$CFLAGS -m64" AC_DEFINE(HAVE_3D, 1, [Define to 1 if 3-dimensional architecture]) AC_DEFINE(HAVE_BG, 1, [Define to 1 if emulating or running on Blue Gene system]) AC_DEFINE(HAVE_BGP, 1, [Define to 1 if emulating or running on Blue Gene/P system]) diff -Nru slurm-llnl-1.3.13/auxdir/x_ac_gtk.m4 slurm-llnl-1.3.15/auxdir/x_ac_gtk.m4 --- slurm-llnl-1.3.13/auxdir/x_ac_gtk.m4 2008-08-15 21:08:09.000000000 +0200 +++ slurm-llnl-1.3.15/auxdir/x_ac_gtk.m4 2009-03-05 18:38:50.000000000 +0100 @@ -18,6 +18,12 @@ ac_have_gtk="yes" _x_ac_pkcfg_bin="no" + # use the correct libs if running on 64bit + if test -d "/usr/lib64/pkgconfig"; then + PKG_CONFIG_PATH="/usr/lib64/pkgconfig/" + fi + + ### Check for pkg-config program AC_ARG_WITH( [pkg-config], @@ -46,24 +52,20 @@ # fi -### Check for gtk2.7.1 package +### Check for min gtk package if test "$ac_have_gtk" == "yes" ; then $HAVEPKGCONFIG --exists gtk+-2.0 if ! test $? -eq 0 ; then AC_MSG_WARN([*** gtk+-2.0 is not available.]) ac_have_gtk="no" else - gtk_config_major_version=`$HAVEPKGCONFIG --modversion gtk+-2.0 | \ - sed 's/\([[0-9]]*\).\([[0-9]]*\).\([[0-9]]*\)/\1/'` - gtk_config_minor_version=`$HAVEPKGCONFIG --modversion gtk+-2.0 | \ - sed 's/\([[0-9]]*\).\([[0-9]]*\).\([[0-9]]*\)/\2/'` - gtk_config_micro_version=`$HAVEPKGCONFIG --modversion gtk+-2.0 | \ - sed 's/\([[0-9]]*\).\([[0-9]]*\).\([[0-9]]*\)/\3/'` - - if test $gtk_config_major_version -lt 2 || test $gtk_config_minor_version -lt 7 || test $gtk_config_micro_version -lt 1; then - AC_MSG_WARN([*** gtk+-$gtk_config_major_version.$gtk_config_minor_version.$gtk_config_micro_version available, we need >= gtk+-2.7.1 installed for sview.]) - ac_have_gtk="no" - fi + min_gtk_version="2.7.1" + $HAVEPKGCONFIG --atleast-version=$min_gtk_version gtk+-2.0 + if ! test $? -eq 0 ; then + gtk_config_version=`$HAVEPKGCONFIG --modversion gtk+-2.0` + AC_MSG_WARN([*** gtk+-$gtk_config_version available, we need >= gtk+-$min_gtk_version installed for sview.]) + ac_have_gtk="no" + fi fi fi diff -Nru slurm-llnl-1.3.13/configure slurm-llnl-1.3.15/configure --- slurm-llnl-1.3.13/configure 2009-01-07 21:57:41.000000000 +0100 +++ slurm-llnl-1.3.15/configure 2009-03-18 23:02:01.000000000 +0100 @@ -852,13 +852,6 @@ MAINTAINER_MODE_TRUE MAINTAINER_MODE_FALSE MAINT -CMD_LDFLAGS -LIB_LDFLAGS -SO_LDFLAGS -HAVE_AIX_TRUE -HAVE_AIX_FALSE -HAVE_AIX -PROCTRACKDIR CC CFLAGS LDFLAGS @@ -875,6 +868,17 @@ CCDEPMODE am__fastdepCC_TRUE am__fastdepCC_FALSE +BG_INCLUDES +BLUEGENE_LOADED_TRUE +BLUEGENE_LOADED_FALSE +BLUEGENE_LOADED +CMD_LDFLAGS +LIB_LDFLAGS +SO_LDFLAGS +HAVE_AIX_TRUE +HAVE_AIX_FALSE +HAVE_AIX +PROCTRACKDIR CPP GREP EGREP @@ -915,10 +919,6 @@ PTHREAD_CC PTHREAD_LIBS PTHREAD_CFLAGS -BG_INCLUDES -BLUEGENE_LOADED_TRUE -BLUEGENE_LOADED_FALSE -BLUEGENE_LOADED SEMAPHORE_SOURCES SEMAPHORE_LIBS NCURSES @@ -1566,8 +1566,12 @@ --enable-FEATURE[=ARG] include FEATURE [ARG=yes] --enable-maintainer-mode enable make rules and dependencies not useful (and sometimes confusing) to the casual installer + --enable-bluegene-emulation + deprecated use --enable-bgl-emulation + --enable-bgl-emulation Run SLURM in BGL mode on a non-bluegene system --disable-dependency-tracking speeds up one-time build --enable-dependency-tracking do not reject slow dependency extractors + --enable-bgp-emulation Run SLURM in BG/P mode on a non-bluegene system --disable-largefile omit support for large files --enable-shared[=PKGS] build shared libraries [default=yes] --enable-static[=PKGS] build static libraries [default=yes] @@ -1576,10 +1580,6 @@ --disable-libtool-lock avoid locking (might break parallel builds) --enable-pam enable PAM (Pluggable Authentication Modules) support - --enable-bluegene-emulation - deprecated use --enable-bgl-emulation - --enable-bgl-emulation Run SLURM in BGL mode on a non-bluegene system - --enable-bgp-emulation Run SLURM in BG/P mode on a non-bluegene system --enable-sun-const enable Sun Constellation system support --enable-debug enable debugging code for development --enable-memory-leak-debug @@ -1594,14 +1594,14 @@ Optional Packages: --with-PACKAGE[=ARG] use PACKAGE [ARG=yes] --without-PACKAGE do not use PACKAGE (same as --with-PACKAGE=no) + --with-db2-dir=PATH Specify path to parent directory of DB2 library + --with-bg-serial=NAME set BG_SERIAL value + --with-proctrack=PATH Specify path to proctrack sources --with-gnu-ld assume the C compiler uses GNU ld [default=no] --with-pic try to use only PIC/non-PIC objects [default=use both] --with-tags[=TAGS] include additional configurations [automatic] - --with-db2-dir=PATH Specify path to parent directory of DB2 library - --with-bg-serial=NAME set BG_SERIAL value - --with-xcpu=PATH specify path to XCPU directory --with-pkg-config=PATH Specify path to pkg-config binary --with-mysql_config=PATH @@ -3922,137 +3922,533 @@ -ac_ext=c -ac_cpp='$CPP $CPPFLAGS' -ac_compile='$CC -c $CFLAGS $CPPFLAGS conftest.$ac_ext >&5' -ac_link='$CC -o conftest$ac_exeext $CFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $LIBS >&5' -ac_compiler_gnu=$ac_cv_c_compiler_gnu -{ echo "$as_me:$LINENO: checking how to run the C preprocessor" >&5 -echo $ECHO_N "checking how to run the C preprocessor... $ECHO_C" >&6; } -# On Suns, sometimes $CPP names a directory. -if test -n "$CPP" && test -d "$CPP"; then - CPP= + + ac_bluegene_loaded=no + + +# Check whether --with-db2-dir was given. +if test "${with_db2_dir+set}" = set; then + withval=$with_db2_dir; trydb2dir=$withval fi -if test -z "$CPP"; then - if test "${ac_cv_prog_CPP+set}" = set; then - echo $ECHO_N "(cached) $ECHO_C" >&6 -else - # Double quotes because CPP needs to be expanded - for CPP in "$CC -E" "$CC -E -traditional-cpp" "/lib/cpp" - do - ac_preproc_ok=false -for ac_c_preproc_warn_flag in '' yes -do - # Use a header file that comes with gcc, so configuring glibc - # with a fresh cross-compiler works. - # Prefer to if __STDC__ is defined, since - # exists even on freestanding compilers. - # On the NeXT, cc -E runs the code through the compiler's parser, - # not just through cpp. "Syntax error" is here to catch this case. - cat >conftest.$ac_ext <<_ACEOF -/* confdefs.h. */ + + + # test for bluegene emulation mode + + # Check whether --enable-bluegene-emulation was given. +if test "${enable_bluegene_emulation+set}" = set; then + enableval=$enable_bluegene_emulation; case "$enableval" in + yes) bluegene_emulation=yes ;; + no) bluegene_emulation=no ;; + *) { { echo "$as_me:$LINENO: error: bad value \"$enableval\" for --enable-bluegene-emulation" >&5 +echo "$as_me: error: bad value \"$enableval\" for --enable-bluegene-emulation" >&2;} + { (exit 1); exit 1; }; } ;; + esac +fi + + + # Check whether --enable-bgl-emulation was given. +if test "${enable_bgl_emulation+set}" = set; then + enableval=$enable_bgl_emulation; case "$enableval" in + yes) bgl_emulation=yes ;; + no) bgl_emulation=no ;; + *) { { echo "$as_me:$LINENO: error: bad value \"$enableval\" for --enable-bgl-emulation" >&5 +echo "$as_me: error: bad value \"$enableval\" for --enable-bgl-emulation" >&2;} + { (exit 1); exit 1; }; } ;; + esac +fi + + + if test "x$bluegene_emulation" = "xyes" -o "x$bgl_emulation" = "xyes"; then + +cat >>confdefs.h <<\_ACEOF +#define HAVE_3D 1 _ACEOF -cat confdefs.h >>conftest.$ac_ext -cat >>conftest.$ac_ext <<_ACEOF -/* end confdefs.h. */ -#ifdef __STDC__ -# include -#else -# include -#endif - Syntax error + + +cat >>confdefs.h <<\_ACEOF +#define HAVE_BG 1 _ACEOF -if { (ac_try="$ac_cpp conftest.$ac_ext" -case "(($ac_try" in - *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;; - *) ac_try_echo=$ac_try;; -esac -eval "echo \"\$as_me:$LINENO: $ac_try_echo\"") >&5 - (eval "$ac_cpp conftest.$ac_ext") 2>conftest.er1 - ac_status=$? - grep -v '^ *+' conftest.er1 >conftest.err - rm -f conftest.er1 - cat conftest.err >&5 - echo "$as_me:$LINENO: \$? = $ac_status" >&5 - (exit $ac_status); } >/dev/null && { - test -z "$ac_c_preproc_warn_flag$ac_c_werror_flag" || - test ! -s conftest.err - }; then - : -else - echo "$as_me: failed program was:" >&5 -sed 's/^/| /' conftest.$ac_ext >&5 - # Broken: fails on valid input. -continue -fi -rm -f conftest.err conftest.$ac_ext +cat >>confdefs.h <<\_ACEOF +#define HAVE_BGL 1 +_ACEOF - # OK, works on sane cases. Now check whether nonexistent headers - # can be detected and how. - cat >conftest.$ac_ext <<_ACEOF + +cat >>confdefs.h <<\_ACEOF +#define HAVE_FRONT_END 1 +_ACEOF + + { echo "$as_me:$LINENO: Running in BG/L emulation mode" >&5 +echo "$as_me: Running in BG/L emulation mode" >&6;} + bg_default_dirs="" + #define ac_bluegene_loaded so we don't load another bluegene conf + ac_bluegene_loaded=yes + else + bg_default_dirs="/bgl/BlueLight/ppcfloor/bglsys /opt/IBM/db2/V8.1 /u/bgdb2cli/sqllib /home/bgdb2cli/sqllib" + fi + + for bg_dir in $trydb2dir "" $bg_default_dirs; do + # Skip directories that don't exist + if test ! -z "$bg_dir" -a ! -d "$bg_dir" ; then + continue; + fi + + # Search for required BG API libraries in the directory + if test -z "$have_bg_ar" -a -f "$bg_dir/lib64/libbglbridge.so" ; then + have_bg_ar=yes + bg_bridge_so="$bg_dir/lib64/libbglbridge.so" + bg_ldflags="$bg_ldflags -L$bg_dir/lib64 -L/usr/lib64 -Wl,--unresolved-symbols=ignore-in-shared-libs -lbglbridge -lbgldb -ltableapi -lbglmachine -lexpat -lsaymessage" + fi + + # Search for required DB2 library in the directory + if test -z "$have_db2" -a -f "$bg_dir/lib64/libdb2.so" ; then + have_db2=yes + bg_db2_so="$bg_dir/lib64/libdb2.so" + bg_ldflags="$bg_ldflags -L$bg_dir/lib64 -ldb2" + fi + + # Search for headers in the directory + if test -z "$have_bg_hdr" -a -f "$bg_dir/include/rm_api.h" ; then + have_bg_hdr=yes + bg_includes="-I$bg_dir/include" + fi + done + + if test ! -z "$have_bg_ar" -a ! -z "$have_bg_hdr" -a ! -z "$have_db2" ; then + # ac_with_readline="no" + # Test to make sure the api is good + have_bg_files=yes + saved_LDFLAGS="$LDFLAGS" + LDFLAGS="$saved_LDFLAGS $bg_ldflags -m64" + cat >conftest.$ac_ext <<_ACEOF /* confdefs.h. */ _ACEOF cat confdefs.h >>conftest.$ac_ext cat >>conftest.$ac_ext <<_ACEOF /* end confdefs.h. */ -#include + int rm_set_serial(char *); +int +main () +{ + rm_set_serial(""); + ; + return 0; +} _ACEOF -if { (ac_try="$ac_cpp conftest.$ac_ext" +rm -f conftest.$ac_objext conftest$ac_exeext +if { (ac_try="$ac_link" case "(($ac_try" in *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;; *) ac_try_echo=$ac_try;; esac eval "echo \"\$as_me:$LINENO: $ac_try_echo\"") >&5 - (eval "$ac_cpp conftest.$ac_ext") 2>conftest.er1 + (eval "$ac_link") 2>conftest.er1 ac_status=$? grep -v '^ *+' conftest.er1 >conftest.err rm -f conftest.er1 cat conftest.err >&5 echo "$as_me:$LINENO: \$? = $ac_status" >&5 - (exit $ac_status); } >/dev/null && { - test -z "$ac_c_preproc_warn_flag$ac_c_werror_flag" || + (exit $ac_status); } && { + test -z "$ac_c_werror_flag" || test ! -s conftest.err - }; then - # Broken: success on invalid input. -continue + } && test -s conftest$ac_exeext && + $as_test_x conftest$ac_exeext; then + have_bg_files=yes else echo "$as_me: failed program was:" >&5 sed 's/^/| /' conftest.$ac_ext >&5 - # Passes both tests. -ac_preproc_ok=: -break + { { echo "$as_me:$LINENO: error: There is a problem linking to the BG/L api." >&5 +echo "$as_me: error: There is a problem linking to the BG/L api." >&2;} + { (exit 1); exit 1; }; } fi -rm -f conftest.err conftest.$ac_ext +rm -f core conftest.err conftest.$ac_objext conftest_ipa8_conftest.oo \ + conftest$ac_exeext conftest.$ac_ext + LDFLAGS="$saved_LDFLAGS" + fi -done -# Because of `break', _AC_PREPROC_IFELSE's cleaning code was skipped. -rm -f conftest.err conftest.$ac_ext -if $ac_preproc_ok; then - break -fi + if test ! -z "$have_bg_files" ; then + BG_INCLUDES="$bg_includes" + CFLAGS="$CFLAGS -m64" - done - ac_cv_prog_CPP=$CPP +cat >>confdefs.h <<\_ACEOF +#define HAVE_3D 1 +_ACEOF -fi - CPP=$ac_cv_prog_CPP -else - ac_cv_prog_CPP=$CPP -fi -{ echo "$as_me:$LINENO: result: $CPP" >&5 -echo "${ECHO_T}$CPP" >&6; } -ac_preproc_ok=false -for ac_c_preproc_warn_flag in '' yes -do - # Use a header file that comes with gcc, so configuring glibc - # with a fresh cross-compiler works. - # Prefer to if __STDC__ is defined, since - # exists even on freestanding compilers. - # On the NeXT, cc -E runs the code through the compiler's parser, + +cat >>confdefs.h <<\_ACEOF +#define HAVE_BG 1 +_ACEOF + + +cat >>confdefs.h <<\_ACEOF +#define HAVE_BGL 1 +_ACEOF + + +cat >>confdefs.h <<\_ACEOF +#define HAVE_FRONT_END 1 +_ACEOF + + +cat >>confdefs.h <<\_ACEOF +#define HAVE_BG_FILES 1 +_ACEOF + + + +cat >>confdefs.h <<_ACEOF +#define BG_BRIDGE_SO "$bg_bridge_so" +_ACEOF + + +cat >>confdefs.h <<_ACEOF +#define BG_DB2_SO "$bg_db2_so" +_ACEOF + + { echo "$as_me:$LINENO: checking for BG serial value" >&5 +echo $ECHO_N "checking for BG serial value... $ECHO_C" >&6; } + bg_serial="BGL" + +# Check whether --with-bg-serial was given. +if test "${with_bg_serial+set}" = set; then + withval=$with_bg_serial; bg_serial="$withval" +fi + + { echo "$as_me:$LINENO: result: $bg_serial" >&5 +echo "${ECHO_T}$bg_serial" >&6; } + +cat >>confdefs.h <<_ACEOF +#define BG_SERIAL "$bg_serial" +_ACEOF + + #define ac_bluegene_loaded so we don't load another bluegene conf + ac_bluegene_loaded=yes + fi + + + + + # test for bluegene emulation mode + # Check whether --enable-bgp-emulation was given. +if test "${enable_bgp_emulation+set}" = set; then + enableval=$enable_bgp_emulation; case "$enableval" in + yes) bgp_emulation=yes ;; + no) bgp_emulation=no ;; + *) { { echo "$as_me:$LINENO: error: bad value \"$enableval\" for --enable-bgp-emulation" >&5 +echo "$as_me: error: bad value \"$enableval\" for --enable-bgp-emulation" >&2;} + { (exit 1); exit 1; }; } ;; + esac +fi + + + # Skip if already set + if test "x$ac_bluegene_loaded" = "xyes" ; then + bg_default_dirs="" + elif test "x$bgp_emulation" = "xyes"; then + +cat >>confdefs.h <<\_ACEOF +#define HAVE_3D 1 +_ACEOF + + +cat >>confdefs.h <<\_ACEOF +#define HAVE_BG 1 +_ACEOF + + +cat >>confdefs.h <<\_ACEOF +#define HAVE_BGP 1 +_ACEOF + + +cat >>confdefs.h <<\_ACEOF +#define HAVE_FRONT_END 1 +_ACEOF + + { echo "$as_me:$LINENO: Running in BG/P emulation mode" >&5 +echo "$as_me: Running in BG/P emulation mode" >&6;} + bg_default_dirs="" + #define ac_bluegene_loaded so we don't load another bluegene conf + ac_bluegene_loaded=yes + else + bg_default_dirs="/bgsys/drivers/ppcfloor" + fi + + libname=bgpbridge + + for bg_dir in $trydb2dir "" $bg_default_dirs; do + # Skip directories that don't exist + if test ! -z "$bg_dir" -a ! -d "$bg_dir" ; then + continue; + fi + + soloc=$bg_dir/lib64/lib$libname.so + # Search for required BG API libraries in the directory + if test -z "$have_bg_ar" -a -f "$soloc" ; then + have_bgp_ar=yes + bg_ldflags="$bg_ldflags -L$bg_dir/lib64 -L/usr/lib64 -Wl,--unresolved-symbols=ignore-in-shared-libs -l$libname" + fi + + # Search for headers in the directory + if test -z "$have_bg_hdr" -a -f "$bg_dir/include/rm_api.h" ; then + have_bgp_hdr=yes + bg_includes="-I$bg_dir/include" + fi + done + + if test ! -z "$have_bgp_ar" -a ! -z "$have_bgp_hdr" ; then + # ac_with_readline="no" + # Test to make sure the api is good + saved_LDFLAGS="$LDFLAGS" + LDFLAGS="$saved_LDFLAGS $bg_ldflags -m64" + cat >conftest.$ac_ext <<_ACEOF +/* confdefs.h. */ +_ACEOF +cat confdefs.h >>conftest.$ac_ext +cat >>conftest.$ac_ext <<_ACEOF +/* end confdefs.h. */ + int rm_set_serial(char *); +int +main () +{ + rm_set_serial(""); + ; + return 0; +} +_ACEOF +rm -f conftest.$ac_objext conftest$ac_exeext +if { (ac_try="$ac_link" +case "(($ac_try" in + *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;; + *) ac_try_echo=$ac_try;; +esac +eval "echo \"\$as_me:$LINENO: $ac_try_echo\"") >&5 + (eval "$ac_link") 2>conftest.er1 + ac_status=$? + grep -v '^ *+' conftest.er1 >conftest.err + rm -f conftest.er1 + cat conftest.err >&5 + echo "$as_me:$LINENO: \$? = $ac_status" >&5 + (exit $ac_status); } && { + test -z "$ac_c_werror_flag" || + test ! -s conftest.err + } && test -s conftest$ac_exeext && + $as_test_x conftest$ac_exeext; then + have_bgp_files=yes +else + echo "$as_me: failed program was:" >&5 +sed 's/^/| /' conftest.$ac_ext >&5 + + { { echo "$as_me:$LINENO: error: There is a problem linking to the BG/P api." >&5 +echo "$as_me: error: There is a problem linking to the BG/P api." >&2;} + { (exit 1); exit 1; }; } +fi + +rm -f core conftest.err conftest.$ac_objext conftest_ipa8_conftest.oo \ + conftest$ac_exeext conftest.$ac_ext + LDFLAGS="$saved_LDFLAGS" + fi + + if test ! -z "$have_bgp_files" ; then + BG_INCLUDES="$bg_includes" + CFLAGS="$CFLAGS -m64" + +cat >>confdefs.h <<\_ACEOF +#define HAVE_3D 1 +_ACEOF + + +cat >>confdefs.h <<\_ACEOF +#define HAVE_BG 1 +_ACEOF + + +cat >>confdefs.h <<\_ACEOF +#define HAVE_BGP 1 +_ACEOF + + +cat >>confdefs.h <<\_ACEOF +#define HAVE_FRONT_END 1 +_ACEOF + + +cat >>confdefs.h <<\_ACEOF +#define HAVE_BG_FILES 1 +_ACEOF + + +cat >>confdefs.h <<_ACEOF +#define BG_BRIDGE_SO "$soloc" +_ACEOF + + + { echo "$as_me:$LINENO: checking for BG serial value" >&5 +echo $ECHO_N "checking for BG serial value... $ECHO_C" >&6; } + bg_serial="BGP" + +# Check whether --with-bg-serial was given. +if test "${with_bg_serial+set}" = set; then + withval=$with_bg_serial; bg_serial="$withval" +fi + + { echo "$as_me:$LINENO: result: $bg_serial" >&5 +echo "${ECHO_T}$bg_serial" >&6; } + +cat >>confdefs.h <<_ACEOF +#define BG_SERIAL "$bg_serial" +_ACEOF + + #define ac_bluegene_loaded so we don't load another bluegene conf + ac_bluegene_loaded=yes + fi + + + + if test "x$ac_bluegene_loaded" = "xyes"; then + BLUEGENE_LOADED_TRUE= + BLUEGENE_LOADED_FALSE='#' +else + BLUEGENE_LOADED_TRUE='#' + BLUEGENE_LOADED_FALSE= +fi + + + +ac_ext=c +ac_cpp='$CPP $CPPFLAGS' +ac_compile='$CC -c $CFLAGS $CPPFLAGS conftest.$ac_ext >&5' +ac_link='$CC -o conftest$ac_exeext $CFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $LIBS >&5' +ac_compiler_gnu=$ac_cv_c_compiler_gnu +{ echo "$as_me:$LINENO: checking how to run the C preprocessor" >&5 +echo $ECHO_N "checking how to run the C preprocessor... $ECHO_C" >&6; } +# On Suns, sometimes $CPP names a directory. +if test -n "$CPP" && test -d "$CPP"; then + CPP= +fi +if test -z "$CPP"; then + if test "${ac_cv_prog_CPP+set}" = set; then + echo $ECHO_N "(cached) $ECHO_C" >&6 +else + # Double quotes because CPP needs to be expanded + for CPP in "$CC -E" "$CC -E -traditional-cpp" "/lib/cpp" + do + ac_preproc_ok=false +for ac_c_preproc_warn_flag in '' yes +do + # Use a header file that comes with gcc, so configuring glibc + # with a fresh cross-compiler works. + # Prefer to if __STDC__ is defined, since + # exists even on freestanding compilers. + # On the NeXT, cc -E runs the code through the compiler's parser, + # not just through cpp. "Syntax error" is here to catch this case. + cat >conftest.$ac_ext <<_ACEOF +/* confdefs.h. */ +_ACEOF +cat confdefs.h >>conftest.$ac_ext +cat >>conftest.$ac_ext <<_ACEOF +/* end confdefs.h. */ +#ifdef __STDC__ +# include +#else +# include +#endif + Syntax error +_ACEOF +if { (ac_try="$ac_cpp conftest.$ac_ext" +case "(($ac_try" in + *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;; + *) ac_try_echo=$ac_try;; +esac +eval "echo \"\$as_me:$LINENO: $ac_try_echo\"") >&5 + (eval "$ac_cpp conftest.$ac_ext") 2>conftest.er1 + ac_status=$? + grep -v '^ *+' conftest.er1 >conftest.err + rm -f conftest.er1 + cat conftest.err >&5 + echo "$as_me:$LINENO: \$? = $ac_status" >&5 + (exit $ac_status); } >/dev/null && { + test -z "$ac_c_preproc_warn_flag$ac_c_werror_flag" || + test ! -s conftest.err + }; then + : +else + echo "$as_me: failed program was:" >&5 +sed 's/^/| /' conftest.$ac_ext >&5 + + # Broken: fails on valid input. +continue +fi + +rm -f conftest.err conftest.$ac_ext + + # OK, works on sane cases. Now check whether nonexistent headers + # can be detected and how. + cat >conftest.$ac_ext <<_ACEOF +/* confdefs.h. */ +_ACEOF +cat confdefs.h >>conftest.$ac_ext +cat >>conftest.$ac_ext <<_ACEOF +/* end confdefs.h. */ +#include +_ACEOF +if { (ac_try="$ac_cpp conftest.$ac_ext" +case "(($ac_try" in + *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;; + *) ac_try_echo=$ac_try;; +esac +eval "echo \"\$as_me:$LINENO: $ac_try_echo\"") >&5 + (eval "$ac_cpp conftest.$ac_ext") 2>conftest.er1 + ac_status=$? + grep -v '^ *+' conftest.er1 >conftest.err + rm -f conftest.er1 + cat conftest.err >&5 + echo "$as_me:$LINENO: \$? = $ac_status" >&5 + (exit $ac_status); } >/dev/null && { + test -z "$ac_c_preproc_warn_flag$ac_c_werror_flag" || + test ! -s conftest.err + }; then + # Broken: success on invalid input. +continue +else + echo "$as_me: failed program was:" >&5 +sed 's/^/| /' conftest.$ac_ext >&5 + + # Passes both tests. +ac_preproc_ok=: +break +fi + +rm -f conftest.err conftest.$ac_ext + +done +# Because of `break', _AC_PREPROC_IFELSE's cleaning code was skipped. +rm -f conftest.err conftest.$ac_ext +if $ac_preproc_ok; then + break +fi + + done + ac_cv_prog_CPP=$CPP + +fi + CPP=$ac_cv_prog_CPP +else + ac_cv_prog_CPP=$CPP +fi +{ echo "$as_me:$LINENO: result: $CPP" >&5 +echo "${ECHO_T}$CPP" >&6; } +ac_preproc_ok=false +for ac_c_preproc_warn_flag in '' yes +do + # Use a header file that comes with gcc, so configuring glibc + # with a fresh cross-compiler works. + # Prefer to if __STDC__ is defined, since + # exists even on freestanding compilers. + # On the NeXT, cc -E runs the code through the compiler's parser, # not just through cpp. "Syntax error" is here to catch this case. cat >conftest.$ac_ext <<_ACEOF /* confdefs.h. */ @@ -7128,7 +7524,7 @@ ;; *-*-irix6*) # Find out which ABI we are using. - echo '#line 7131 "configure"' > conftest.$ac_ext + echo '#line 7527 "configure"' > conftest.$ac_ext if { (eval echo "$as_me:$LINENO: \"$ac_compile\"") >&5 (eval $ac_compile) 2>&5 ac_status=$? @@ -9234,11 +9630,11 @@ -e 's:.*FLAGS}\{0,1\} :&$lt_compiler_flag :; t' \ -e 's: [^ ]*conftest\.: $lt_compiler_flag&:; t' \ -e 's:$: $lt_compiler_flag:'` - (eval echo "\"\$as_me:9237: $lt_compile\"" >&5) + (eval echo "\"\$as_me:9633: $lt_compile\"" >&5) (eval "$lt_compile" 2>conftest.err) ac_status=$? cat conftest.err >&5 - echo "$as_me:9241: \$? = $ac_status" >&5 + echo "$as_me:9637: \$? = $ac_status" >&5 if (exit $ac_status) && test -s "$ac_outfile"; then # The compiler can only warn and ignore the option if not recognized # So say no if there are warnings other than the usual output. @@ -9524,11 +9920,11 @@ -e 's:.*FLAGS}\{0,1\} :&$lt_compiler_flag :; t' \ -e 's: [^ ]*conftest\.: $lt_compiler_flag&:; t' \ -e 's:$: $lt_compiler_flag:'` - (eval echo "\"\$as_me:9527: $lt_compile\"" >&5) + (eval echo "\"\$as_me:9923: $lt_compile\"" >&5) (eval "$lt_compile" 2>conftest.err) ac_status=$? cat conftest.err >&5 - echo "$as_me:9531: \$? = $ac_status" >&5 + echo "$as_me:9927: \$? = $ac_status" >&5 if (exit $ac_status) && test -s "$ac_outfile"; then # The compiler can only warn and ignore the option if not recognized # So say no if there are warnings other than the usual output. @@ -9628,11 +10024,11 @@ -e 's:.*FLAGS}\{0,1\} :&$lt_compiler_flag :; t' \ -e 's: [^ ]*conftest\.: $lt_compiler_flag&:; t' \ -e 's:$: $lt_compiler_flag:'` - (eval echo "\"\$as_me:9631: $lt_compile\"" >&5) + (eval echo "\"\$as_me:10027: $lt_compile\"" >&5) (eval "$lt_compile" 2>out/conftest.err) ac_status=$? cat out/conftest.err >&5 - echo "$as_me:9635: \$? = $ac_status" >&5 + echo "$as_me:10031: \$? = $ac_status" >&5 if (exit $ac_status) && test -s out/conftest2.$ac_objext then # The compiler can only warn and ignore the option if not recognized @@ -12005,7 +12401,7 @@ lt_dlunknown=0; lt_dlno_uscore=1; lt_dlneed_uscore=2 lt_status=$lt_dlunknown cat > conftest.$ac_ext < conftest.$ac_ext <&5) + (eval echo "\"\$as_me:14905: $lt_compile\"" >&5) (eval "$lt_compile" 2>conftest.err) ac_status=$? cat conftest.err >&5 - echo "$as_me:14513: \$? = $ac_status" >&5 + echo "$as_me:14909: \$? = $ac_status" >&5 if (exit $ac_status) && test -s "$ac_outfile"; then # The compiler can only warn and ignore the option if not recognized # So say no if there are warnings other than the usual output. @@ -14610,11 +15006,11 @@ -e 's:.*FLAGS}\{0,1\} :&$lt_compiler_flag :; t' \ -e 's: [^ ]*conftest\.: $lt_compiler_flag&:; t' \ -e 's:$: $lt_compiler_flag:'` - (eval echo "\"\$as_me:14613: $lt_compile\"" >&5) + (eval echo "\"\$as_me:15009: $lt_compile\"" >&5) (eval "$lt_compile" 2>out/conftest.err) ac_status=$? cat out/conftest.err >&5 - echo "$as_me:14617: \$? = $ac_status" >&5 + echo "$as_me:15013: \$? = $ac_status" >&5 if (exit $ac_status) && test -s out/conftest2.$ac_objext then # The compiler can only warn and ignore the option if not recognized @@ -16208,11 +16604,11 @@ -e 's:.*FLAGS}\{0,1\} :&$lt_compiler_flag :; t' \ -e 's: [^ ]*conftest\.: $lt_compiler_flag&:; t' \ -e 's:$: $lt_compiler_flag:'` - (eval echo "\"\$as_me:16211: $lt_compile\"" >&5) + (eval echo "\"\$as_me:16607: $lt_compile\"" >&5) (eval "$lt_compile" 2>conftest.err) ac_status=$? cat conftest.err >&5 - echo "$as_me:16215: \$? = $ac_status" >&5 + echo "$as_me:16611: \$? = $ac_status" >&5 if (exit $ac_status) && test -s "$ac_outfile"; then # The compiler can only warn and ignore the option if not recognized # So say no if there are warnings other than the usual output. @@ -16312,11 +16708,11 @@ -e 's:.*FLAGS}\{0,1\} :&$lt_compiler_flag :; t' \ -e 's: [^ ]*conftest\.: $lt_compiler_flag&:; t' \ -e 's:$: $lt_compiler_flag:'` - (eval echo "\"\$as_me:16315: $lt_compile\"" >&5) + (eval echo "\"\$as_me:16711: $lt_compile\"" >&5) (eval "$lt_compile" 2>out/conftest.err) ac_status=$? cat out/conftest.err >&5 - echo "$as_me:16319: \$? = $ac_status" >&5 + echo "$as_me:16715: \$? = $ac_status" >&5 if (exit $ac_status) && test -s out/conftest2.$ac_objext then # The compiler can only warn and ignore the option if not recognized @@ -18532,11 +18928,11 @@ -e 's:.*FLAGS}\{0,1\} :&$lt_compiler_flag :; t' \ -e 's: [^ ]*conftest\.: $lt_compiler_flag&:; t' \ -e 's:$: $lt_compiler_flag:'` - (eval echo "\"\$as_me:18535: $lt_compile\"" >&5) + (eval echo "\"\$as_me:18931: $lt_compile\"" >&5) (eval "$lt_compile" 2>conftest.err) ac_status=$? cat conftest.err >&5 - echo "$as_me:18539: \$? = $ac_status" >&5 + echo "$as_me:18935: \$? = $ac_status" >&5 if (exit $ac_status) && test -s "$ac_outfile"; then # The compiler can only warn and ignore the option if not recognized # So say no if there are warnings other than the usual output. @@ -18822,11 +19218,11 @@ -e 's:.*FLAGS}\{0,1\} :&$lt_compiler_flag :; t' \ -e 's: [^ ]*conftest\.: $lt_compiler_flag&:; t' \ -e 's:$: $lt_compiler_flag:'` - (eval echo "\"\$as_me:18825: $lt_compile\"" >&5) + (eval echo "\"\$as_me:19221: $lt_compile\"" >&5) (eval "$lt_compile" 2>conftest.err) ac_status=$? cat conftest.err >&5 - echo "$as_me:18829: \$? = $ac_status" >&5 + echo "$as_me:19225: \$? = $ac_status" >&5 if (exit $ac_status) && test -s "$ac_outfile"; then # The compiler can only warn and ignore the option if not recognized # So say no if there are warnings other than the usual output. @@ -18926,11 +19322,11 @@ -e 's:.*FLAGS}\{0,1\} :&$lt_compiler_flag :; t' \ -e 's: [^ ]*conftest\.: $lt_compiler_flag&:; t' \ -e 's:$: $lt_compiler_flag:'` - (eval echo "\"\$as_me:18929: $lt_compile\"" >&5) + (eval echo "\"\$as_me:19325: $lt_compile\"" >&5) (eval "$lt_compile" 2>out/conftest.err) ac_status=$? cat out/conftest.err >&5 - echo "$as_me:18933: \$? = $ac_status" >&5 + echo "$as_me:19329: \$? = $ac_status" >&5 if (exit $ac_status) && test -s out/conftest2.$ac_objext then # The compiler can only warn and ignore the option if not recognized @@ -22688,6 +23084,7 @@ #define HAVE_NUMA 1 _ACEOF + CFLAGS="-DNUMA_VERSION1_COMPATIBILITY $CFLAGS" else { echo "$as_me:$LINENO: WARNING: Unable to locate NUMA memory affinity functions" >&5 echo "$as_me: WARNING: Unable to locate NUMA memory affinity functions" >&2;} @@ -23137,296 +23534,7 @@ fi -rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext -else - cat >conftest.$ac_ext <<_ACEOF -/* confdefs.h. */ -_ACEOF -cat confdefs.h >>conftest.$ac_ext -cat >>conftest.$ac_ext <<_ACEOF -/* end confdefs.h. */ -$ac_includes_default -int -main () -{ - - /* Are we little or big endian? From Harbison&Steele. */ - union - { - long int l; - char c[sizeof (long int)]; - } u; - u.l = 1; - return u.c[sizeof (long int) - 1] == 1; - - ; - return 0; -} -_ACEOF -rm -f conftest$ac_exeext -if { (ac_try="$ac_link" -case "(($ac_try" in - *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;; - *) ac_try_echo=$ac_try;; -esac -eval "echo \"\$as_me:$LINENO: $ac_try_echo\"") >&5 - (eval "$ac_link") 2>&5 - ac_status=$? - echo "$as_me:$LINENO: \$? = $ac_status" >&5 - (exit $ac_status); } && { ac_try='./conftest$ac_exeext' - { (case "(($ac_try" in - *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;; - *) ac_try_echo=$ac_try;; -esac -eval "echo \"\$as_me:$LINENO: $ac_try_echo\"") >&5 - (eval "$ac_try") 2>&5 - ac_status=$? - echo "$as_me:$LINENO: \$? = $ac_status" >&5 - (exit $ac_status); }; }; then - ac_cv_c_bigendian=no -else - echo "$as_me: program exited with status $ac_status" >&5 -echo "$as_me: failed program was:" >&5 -sed 's/^/| /' conftest.$ac_ext >&5 - -( exit $ac_status ) -ac_cv_c_bigendian=yes -fi -rm -f core *.core core.conftest.* gmon.out bb.out conftest$ac_exeext conftest.$ac_objext conftest.$ac_ext -fi - - -fi - -rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext -fi -{ echo "$as_me:$LINENO: result: $ac_cv_c_bigendian" >&5 -echo "${ECHO_T}$ac_cv_c_bigendian" >&6; } -case $ac_cv_c_bigendian in - yes) - -cat >>confdefs.h <<\_ACEOF -#define WORDS_BIGENDIAN 1 -_ACEOF - ;; - no) - ;; - *) - { { echo "$as_me:$LINENO: error: unknown endianness -presetting ac_cv_c_bigendian=no (or yes) will help" >&5 -echo "$as_me: error: unknown endianness -presetting ac_cv_c_bigendian=no (or yes) will help" >&2;} - { (exit 1); exit 1; }; } ;; -esac - - if test "x$ac_cv_c_bigendian" = "xyes"; then - -cat >>confdefs.h <<\_ACEOF -#define SLURM_BIGENDIAN 1 -_ACEOF - - fi - - -if test $ac_cv_c_compiler_gnu = yes; then - { echo "$as_me:$LINENO: checking whether $CC needs -traditional" >&5 -echo $ECHO_N "checking whether $CC needs -traditional... $ECHO_C" >&6; } -if test "${ac_cv_prog_gcc_traditional+set}" = set; then - echo $ECHO_N "(cached) $ECHO_C" >&6 -else - ac_pattern="Autoconf.*'x'" - cat >conftest.$ac_ext <<_ACEOF -/* confdefs.h. */ -_ACEOF -cat confdefs.h >>conftest.$ac_ext -cat >>conftest.$ac_ext <<_ACEOF -/* end confdefs.h. */ -#include -Autoconf TIOCGETP -_ACEOF -if (eval "$ac_cpp conftest.$ac_ext") 2>&5 | - $EGREP "$ac_pattern" >/dev/null 2>&1; then - ac_cv_prog_gcc_traditional=yes -else - ac_cv_prog_gcc_traditional=no -fi -rm -f conftest* - - - if test $ac_cv_prog_gcc_traditional = no; then - cat >conftest.$ac_ext <<_ACEOF -/* confdefs.h. */ -_ACEOF -cat confdefs.h >>conftest.$ac_ext -cat >>conftest.$ac_ext <<_ACEOF -/* end confdefs.h. */ -#include -Autoconf TCGETA -_ACEOF -if (eval "$ac_cpp conftest.$ac_ext") 2>&5 | - $EGREP "$ac_pattern" >/dev/null 2>&1; then - ac_cv_prog_gcc_traditional=yes -fi -rm -f conftest* - - fi -fi -{ echo "$as_me:$LINENO: result: $ac_cv_prog_gcc_traditional" >&5 -echo "${ECHO_T}$ac_cv_prog_gcc_traditional" >&6; } - if test $ac_cv_prog_gcc_traditional = yes; then - CC="$CC -traditional" - fi -fi - - - - -for ac_header in stdlib.h -do -as_ac_Header=`echo "ac_cv_header_$ac_header" | $as_tr_sh` -if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then - { echo "$as_me:$LINENO: checking for $ac_header" >&5 -echo $ECHO_N "checking for $ac_header... $ECHO_C" >&6; } -if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then - echo $ECHO_N "(cached) $ECHO_C" >&6 -fi -ac_res=`eval echo '${'$as_ac_Header'}'` - { echo "$as_me:$LINENO: result: $ac_res" >&5 -echo "${ECHO_T}$ac_res" >&6; } -else - # Is the header compilable? -{ echo "$as_me:$LINENO: checking $ac_header usability" >&5 -echo $ECHO_N "checking $ac_header usability... $ECHO_C" >&6; } -cat >conftest.$ac_ext <<_ACEOF -/* confdefs.h. */ -_ACEOF -cat confdefs.h >>conftest.$ac_ext -cat >>conftest.$ac_ext <<_ACEOF -/* end confdefs.h. */ -$ac_includes_default -#include <$ac_header> -_ACEOF -rm -f conftest.$ac_objext -if { (ac_try="$ac_compile" -case "(($ac_try" in - *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;; - *) ac_try_echo=$ac_try;; -esac -eval "echo \"\$as_me:$LINENO: $ac_try_echo\"") >&5 - (eval "$ac_compile") 2>conftest.er1 - ac_status=$? - grep -v '^ *+' conftest.er1 >conftest.err - rm -f conftest.er1 - cat conftest.err >&5 - echo "$as_me:$LINENO: \$? = $ac_status" >&5 - (exit $ac_status); } && { - test -z "$ac_c_werror_flag" || - test ! -s conftest.err - } && test -s conftest.$ac_objext; then - ac_header_compiler=yes -else - echo "$as_me: failed program was:" >&5 -sed 's/^/| /' conftest.$ac_ext >&5 - - ac_header_compiler=no -fi - -rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext -{ echo "$as_me:$LINENO: result: $ac_header_compiler" >&5 -echo "${ECHO_T}$ac_header_compiler" >&6; } - -# Is the header present? -{ echo "$as_me:$LINENO: checking $ac_header presence" >&5 -echo $ECHO_N "checking $ac_header presence... $ECHO_C" >&6; } -cat >conftest.$ac_ext <<_ACEOF -/* confdefs.h. */ -_ACEOF -cat confdefs.h >>conftest.$ac_ext -cat >>conftest.$ac_ext <<_ACEOF -/* end confdefs.h. */ -#include <$ac_header> -_ACEOF -if { (ac_try="$ac_cpp conftest.$ac_ext" -case "(($ac_try" in - *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;; - *) ac_try_echo=$ac_try;; -esac -eval "echo \"\$as_me:$LINENO: $ac_try_echo\"") >&5 - (eval "$ac_cpp conftest.$ac_ext") 2>conftest.er1 - ac_status=$? - grep -v '^ *+' conftest.er1 >conftest.err - rm -f conftest.er1 - cat conftest.err >&5 - echo "$as_me:$LINENO: \$? = $ac_status" >&5 - (exit $ac_status); } >/dev/null && { - test -z "$ac_c_preproc_warn_flag$ac_c_werror_flag" || - test ! -s conftest.err - }; then - ac_header_preproc=yes -else - echo "$as_me: failed program was:" >&5 -sed 's/^/| /' conftest.$ac_ext >&5 - - ac_header_preproc=no -fi - -rm -f conftest.err conftest.$ac_ext -{ echo "$as_me:$LINENO: result: $ac_header_preproc" >&5 -echo "${ECHO_T}$ac_header_preproc" >&6; } - -# So? What about this header? -case $ac_header_compiler:$ac_header_preproc:$ac_c_preproc_warn_flag in - yes:no: ) - { echo "$as_me:$LINENO: WARNING: $ac_header: accepted by the compiler, rejected by the preprocessor!" >&5 -echo "$as_me: WARNING: $ac_header: accepted by the compiler, rejected by the preprocessor!" >&2;} - { echo "$as_me:$LINENO: WARNING: $ac_header: proceeding with the compiler's result" >&5 -echo "$as_me: WARNING: $ac_header: proceeding with the compiler's result" >&2;} - ac_header_preproc=yes - ;; - no:yes:* ) - { echo "$as_me:$LINENO: WARNING: $ac_header: present but cannot be compiled" >&5 -echo "$as_me: WARNING: $ac_header: present but cannot be compiled" >&2;} - { echo "$as_me:$LINENO: WARNING: $ac_header: check for missing prerequisite headers?" >&5 -echo "$as_me: WARNING: $ac_header: check for missing prerequisite headers?" >&2;} - { echo "$as_me:$LINENO: WARNING: $ac_header: see the Autoconf documentation" >&5 -echo "$as_me: WARNING: $ac_header: see the Autoconf documentation" >&2;} - { echo "$as_me:$LINENO: WARNING: $ac_header: section \"Present But Cannot Be Compiled\"" >&5 -echo "$as_me: WARNING: $ac_header: section \"Present But Cannot Be Compiled\"" >&2;} - { echo "$as_me:$LINENO: WARNING: $ac_header: proceeding with the preprocessor's result" >&5 -echo "$as_me: WARNING: $ac_header: proceeding with the preprocessor's result" >&2;} - { echo "$as_me:$LINENO: WARNING: $ac_header: in the future, the compiler will take precedence" >&5 -echo "$as_me: WARNING: $ac_header: in the future, the compiler will take precedence" >&2;} - - ;; -esac -{ echo "$as_me:$LINENO: checking for $ac_header" >&5 -echo $ECHO_N "checking for $ac_header... $ECHO_C" >&6; } -if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then - echo $ECHO_N "(cached) $ECHO_C" >&6 -else - eval "$as_ac_Header=\$ac_header_preproc" -fi -ac_res=`eval echo '${'$as_ac_Header'}'` - { echo "$as_me:$LINENO: result: $ac_res" >&5 -echo "${ECHO_T}$ac_res" >&6; } - -fi -if test `eval echo '${'$as_ac_Header'}'` = yes; then - cat >>confdefs.h <<_ACEOF -#define `echo "HAVE_$ac_header" | $as_tr_cpp` 1 -_ACEOF - -fi - -done - -{ echo "$as_me:$LINENO: checking for GNU libc compatible malloc" >&5 -echo $ECHO_N "checking for GNU libc compatible malloc... $ECHO_C" >&6; } -if test "${ac_cv_func_malloc_0_nonnull+set}" = set; then - echo $ECHO_N "(cached) $ECHO_C" >&6 -else - if test "$cross_compiling" = yes; then - ac_cv_func_malloc_0_nonnull=no +rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext else cat >conftest.$ac_ext <<_ACEOF /* confdefs.h. */ @@ -23434,16 +23542,20 @@ cat confdefs.h >>conftest.$ac_ext cat >>conftest.$ac_ext <<_ACEOF /* end confdefs.h. */ -#if defined STDC_HEADERS || defined HAVE_STDLIB_H -# include -#else -char *malloc (); -#endif - +$ac_includes_default int main () { -return ! malloc (0); + + /* Are we little or big endian? From Harbison&Steele. */ + union + { + long int l; + char c[sizeof (long int)]; + } u; + u.l = 1; + return u.c[sizeof (long int) - 1] == 1; + ; return 0; } @@ -23468,178 +23580,137 @@ ac_status=$? echo "$as_me:$LINENO: \$? = $ac_status" >&5 (exit $ac_status); }; }; then - ac_cv_func_malloc_0_nonnull=yes + ac_cv_c_bigendian=no else echo "$as_me: program exited with status $ac_status" >&5 echo "$as_me: failed program was:" >&5 sed 's/^/| /' conftest.$ac_ext >&5 ( exit $ac_status ) -ac_cv_func_malloc_0_nonnull=no +ac_cv_c_bigendian=yes fi rm -f core *.core core.conftest.* gmon.out bb.out conftest$ac_exeext conftest.$ac_objext conftest.$ac_ext fi fi -{ echo "$as_me:$LINENO: result: $ac_cv_func_malloc_0_nonnull" >&5 -echo "${ECHO_T}$ac_cv_func_malloc_0_nonnull" >&6; } -if test $ac_cv_func_malloc_0_nonnull = yes; then -cat >>confdefs.h <<\_ACEOF -#define HAVE_MALLOC 1 -_ACEOF +rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext +fi +{ echo "$as_me:$LINENO: result: $ac_cv_c_bigendian" >&5 +echo "${ECHO_T}$ac_cv_c_bigendian" >&6; } +case $ac_cv_c_bigendian in + yes) -else - cat >>confdefs.h <<\_ACEOF -#define HAVE_MALLOC 0 +cat >>confdefs.h <<\_ACEOF +#define WORDS_BIGENDIAN 1 _ACEOF - - case " $LIBOBJS " in - *" malloc.$ac_objext "* ) ;; - *) LIBOBJS="$LIBOBJS malloc.$ac_objext" ;; + no) + ;; + *) + { { echo "$as_me:$LINENO: error: unknown endianness +presetting ac_cv_c_bigendian=no (or yes) will help" >&5 +echo "$as_me: error: unknown endianness +presetting ac_cv_c_bigendian=no (or yes) will help" >&2;} + { (exit 1); exit 1; }; } ;; esac + if test "x$ac_cv_c_bigendian" = "xyes"; then cat >>confdefs.h <<\_ACEOF -#define malloc rpl_malloc +#define SLURM_BIGENDIAN 1 _ACEOF -fi - + fi -{ echo "$as_me:$LINENO: checking whether strerror_r is declared" >&5 -echo $ECHO_N "checking whether strerror_r is declared... $ECHO_C" >&6; } -if test "${ac_cv_have_decl_strerror_r+set}" = set; then +if test $ac_cv_c_compiler_gnu = yes; then + { echo "$as_me:$LINENO: checking whether $CC needs -traditional" >&5 +echo $ECHO_N "checking whether $CC needs -traditional... $ECHO_C" >&6; } +if test "${ac_cv_prog_gcc_traditional+set}" = set; then echo $ECHO_N "(cached) $ECHO_C" >&6 else + ac_pattern="Autoconf.*'x'" cat >conftest.$ac_ext <<_ACEOF /* confdefs.h. */ _ACEOF cat confdefs.h >>conftest.$ac_ext cat >>conftest.$ac_ext <<_ACEOF /* end confdefs.h. */ -$ac_includes_default -int -main () -{ -#ifndef strerror_r - (void) strerror_r; -#endif - - ; - return 0; -} -_ACEOF -rm -f conftest.$ac_objext -if { (ac_try="$ac_compile" -case "(($ac_try" in - *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;; - *) ac_try_echo=$ac_try;; -esac -eval "echo \"\$as_me:$LINENO: $ac_try_echo\"") >&5 - (eval "$ac_compile") 2>conftest.er1 - ac_status=$? - grep -v '^ *+' conftest.er1 >conftest.err - rm -f conftest.er1 - cat conftest.err >&5 - echo "$as_me:$LINENO: \$? = $ac_status" >&5 - (exit $ac_status); } && { - test -z "$ac_c_werror_flag" || - test ! -s conftest.err - } && test -s conftest.$ac_objext; then - ac_cv_have_decl_strerror_r=yes -else - echo "$as_me: failed program was:" >&5 -sed 's/^/| /' conftest.$ac_ext >&5 - - ac_cv_have_decl_strerror_r=no -fi - -rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext -fi -{ echo "$as_me:$LINENO: result: $ac_cv_have_decl_strerror_r" >&5 -echo "${ECHO_T}$ac_cv_have_decl_strerror_r" >&6; } -if test $ac_cv_have_decl_strerror_r = yes; then - -cat >>confdefs.h <<_ACEOF -#define HAVE_DECL_STRERROR_R 1 +#include +Autoconf TIOCGETP _ACEOF - - +if (eval "$ac_cpp conftest.$ac_ext") 2>&5 | + $EGREP "$ac_pattern" >/dev/null 2>&1; then + ac_cv_prog_gcc_traditional=yes else - cat >>confdefs.h <<_ACEOF -#define HAVE_DECL_STRERROR_R 0 -_ACEOF - - + ac_cv_prog_gcc_traditional=no fi +rm -f conftest* - -for ac_func in strerror_r -do -as_ac_var=`echo "ac_cv_func_$ac_func" | $as_tr_sh` -{ echo "$as_me:$LINENO: checking for $ac_func" >&5 -echo $ECHO_N "checking for $ac_func... $ECHO_C" >&6; } -if { as_var=$as_ac_var; eval "test \"\${$as_var+set}\" = set"; }; then - echo $ECHO_N "(cached) $ECHO_C" >&6 -else - cat >conftest.$ac_ext <<_ACEOF + if test $ac_cv_prog_gcc_traditional = no; then + cat >conftest.$ac_ext <<_ACEOF /* confdefs.h. */ _ACEOF cat confdefs.h >>conftest.$ac_ext cat >>conftest.$ac_ext <<_ACEOF /* end confdefs.h. */ -/* Define $ac_func to an innocuous variant, in case declares $ac_func. - For example, HP-UX 11i declares gettimeofday. */ -#define $ac_func innocuous_$ac_func +#include +Autoconf TCGETA +_ACEOF +if (eval "$ac_cpp conftest.$ac_ext") 2>&5 | + $EGREP "$ac_pattern" >/dev/null 2>&1; then + ac_cv_prog_gcc_traditional=yes +fi +rm -f conftest* -/* System header to define __stub macros and hopefully few prototypes, - which can conflict with char $ac_func (); below. - Prefer to if __STDC__ is defined, since - exists even on freestanding compilers. */ + fi +fi +{ echo "$as_me:$LINENO: result: $ac_cv_prog_gcc_traditional" >&5 +echo "${ECHO_T}$ac_cv_prog_gcc_traditional" >&6; } + if test $ac_cv_prog_gcc_traditional = yes; then + CC="$CC -traditional" + fi +fi -#ifdef __STDC__ -# include -#else -# include -#endif -#undef $ac_func -/* Override any GCC internal prototype to avoid an error. - Use char because int might match the return type of a GCC - builtin and then its argument prototype would still apply. */ -#ifdef __cplusplus -extern "C" -#endif -char $ac_func (); -/* The GNU C library defines this for functions which it implements - to always fail with ENOSYS. Some functions are actually named - something starting with __ and the normal name is an alias. */ -#if defined __stub_$ac_func || defined __stub___$ac_func -choke me -#endif -int -main () -{ -return $ac_func (); - ; - return 0; -} +for ac_header in stdlib.h +do +as_ac_Header=`echo "ac_cv_header_$ac_header" | $as_tr_sh` +if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then + { echo "$as_me:$LINENO: checking for $ac_header" >&5 +echo $ECHO_N "checking for $ac_header... $ECHO_C" >&6; } +if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then + echo $ECHO_N "(cached) $ECHO_C" >&6 +fi +ac_res=`eval echo '${'$as_ac_Header'}'` + { echo "$as_me:$LINENO: result: $ac_res" >&5 +echo "${ECHO_T}$ac_res" >&6; } +else + # Is the header compilable? +{ echo "$as_me:$LINENO: checking $ac_header usability" >&5 +echo $ECHO_N "checking $ac_header usability... $ECHO_C" >&6; } +cat >conftest.$ac_ext <<_ACEOF +/* confdefs.h. */ _ACEOF -rm -f conftest.$ac_objext conftest$ac_exeext -if { (ac_try="$ac_link" +cat confdefs.h >>conftest.$ac_ext +cat >>conftest.$ac_ext <<_ACEOF +/* end confdefs.h. */ +$ac_includes_default +#include <$ac_header> +_ACEOF +rm -f conftest.$ac_objext +if { (ac_try="$ac_compile" case "(($ac_try" in *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;; *) ac_try_echo=$ac_try;; esac eval "echo \"\$as_me:$LINENO: $ac_try_echo\"") >&5 - (eval "$ac_link") 2>conftest.er1 + (eval "$ac_compile") 2>conftest.er1 ac_status=$? grep -v '^ *+' conftest.er1 >conftest.err rm -f conftest.er1 @@ -23648,93 +23719,111 @@ (exit $ac_status); } && { test -z "$ac_c_werror_flag" || test ! -s conftest.err - } && test -s conftest$ac_exeext && - $as_test_x conftest$ac_exeext; then - eval "$as_ac_var=yes" + } && test -s conftest.$ac_objext; then + ac_header_compiler=yes else echo "$as_me: failed program was:" >&5 sed 's/^/| /' conftest.$ac_ext >&5 - eval "$as_ac_var=no" -fi - -rm -f core conftest.err conftest.$ac_objext conftest_ipa8_conftest.oo \ - conftest$ac_exeext conftest.$ac_ext -fi -ac_res=`eval echo '${'$as_ac_var'}'` - { echo "$as_me:$LINENO: result: $ac_res" >&5 -echo "${ECHO_T}$ac_res" >&6; } -if test `eval echo '${'$as_ac_var'}'` = yes; then - cat >>confdefs.h <<_ACEOF -#define `echo "HAVE_$ac_func" | $as_tr_cpp` 1 -_ACEOF - + ac_header_compiler=no fi -done -{ echo "$as_me:$LINENO: checking whether strerror_r returns char *" >&5 -echo $ECHO_N "checking whether strerror_r returns char *... $ECHO_C" >&6; } -if test "${ac_cv_func_strerror_r_char_p+set}" = set; then - echo $ECHO_N "(cached) $ECHO_C" >&6 -else +rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext +{ echo "$as_me:$LINENO: result: $ac_header_compiler" >&5 +echo "${ECHO_T}$ac_header_compiler" >&6; } - ac_cv_func_strerror_r_char_p=no - if test $ac_cv_have_decl_strerror_r = yes; then - cat >conftest.$ac_ext <<_ACEOF +# Is the header present? +{ echo "$as_me:$LINENO: checking $ac_header presence" >&5 +echo $ECHO_N "checking $ac_header presence... $ECHO_C" >&6; } +cat >conftest.$ac_ext <<_ACEOF /* confdefs.h. */ _ACEOF cat confdefs.h >>conftest.$ac_ext cat >>conftest.$ac_ext <<_ACEOF /* end confdefs.h. */ -$ac_includes_default -int -main () -{ - - char buf[100]; - char x = *strerror_r (0, buf, sizeof buf); - char *p = strerror_r (0, buf, sizeof buf); - return !p || x; - - ; - return 0; -} +#include <$ac_header> _ACEOF -rm -f conftest.$ac_objext -if { (ac_try="$ac_compile" +if { (ac_try="$ac_cpp conftest.$ac_ext" case "(($ac_try" in *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;; *) ac_try_echo=$ac_try;; esac eval "echo \"\$as_me:$LINENO: $ac_try_echo\"") >&5 - (eval "$ac_compile") 2>conftest.er1 + (eval "$ac_cpp conftest.$ac_ext") 2>conftest.er1 ac_status=$? grep -v '^ *+' conftest.er1 >conftest.err rm -f conftest.er1 cat conftest.err >&5 echo "$as_me:$LINENO: \$? = $ac_status" >&5 - (exit $ac_status); } && { - test -z "$ac_c_werror_flag" || + (exit $ac_status); } >/dev/null && { + test -z "$ac_c_preproc_warn_flag$ac_c_werror_flag" || test ! -s conftest.err - } && test -s conftest.$ac_objext; then - ac_cv_func_strerror_r_char_p=yes + }; then + ac_header_preproc=yes else echo "$as_me: failed program was:" >&5 sed 's/^/| /' conftest.$ac_ext >&5 + ac_header_preproc=no +fi + +rm -f conftest.err conftest.$ac_ext +{ echo "$as_me:$LINENO: result: $ac_header_preproc" >&5 +echo "${ECHO_T}$ac_header_preproc" >&6; } + +# So? What about this header? +case $ac_header_compiler:$ac_header_preproc:$ac_c_preproc_warn_flag in + yes:no: ) + { echo "$as_me:$LINENO: WARNING: $ac_header: accepted by the compiler, rejected by the preprocessor!" >&5 +echo "$as_me: WARNING: $ac_header: accepted by the compiler, rejected by the preprocessor!" >&2;} + { echo "$as_me:$LINENO: WARNING: $ac_header: proceeding with the compiler's result" >&5 +echo "$as_me: WARNING: $ac_header: proceeding with the compiler's result" >&2;} + ac_header_preproc=yes + ;; + no:yes:* ) + { echo "$as_me:$LINENO: WARNING: $ac_header: present but cannot be compiled" >&5 +echo "$as_me: WARNING: $ac_header: present but cannot be compiled" >&2;} + { echo "$as_me:$LINENO: WARNING: $ac_header: check for missing prerequisite headers?" >&5 +echo "$as_me: WARNING: $ac_header: check for missing prerequisite headers?" >&2;} + { echo "$as_me:$LINENO: WARNING: $ac_header: see the Autoconf documentation" >&5 +echo "$as_me: WARNING: $ac_header: see the Autoconf documentation" >&2;} + { echo "$as_me:$LINENO: WARNING: $ac_header: section \"Present But Cannot Be Compiled\"" >&5 +echo "$as_me: WARNING: $ac_header: section \"Present But Cannot Be Compiled\"" >&2;} + { echo "$as_me:$LINENO: WARNING: $ac_header: proceeding with the preprocessor's result" >&5 +echo "$as_me: WARNING: $ac_header: proceeding with the preprocessor's result" >&2;} + { echo "$as_me:$LINENO: WARNING: $ac_header: in the future, the compiler will take precedence" >&5 +echo "$as_me: WARNING: $ac_header: in the future, the compiler will take precedence" >&2;} + ;; +esac +{ echo "$as_me:$LINENO: checking for $ac_header" >&5 +echo $ECHO_N "checking for $ac_header... $ECHO_C" >&6; } +if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then + echo $ECHO_N "(cached) $ECHO_C" >&6 +else + eval "$as_ac_Header=\$ac_header_preproc" fi +ac_res=`eval echo '${'$as_ac_Header'}'` + { echo "$as_me:$LINENO: result: $ac_res" >&5 +echo "${ECHO_T}$ac_res" >&6; } -rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext - else - # strerror_r is not declared. Choose between - # systems that have relatively inaccessible declarations for the - # function. BeOS and DEC UNIX 4.0 fall in this category, but the - # former has a strerror_r that returns char*, while the latter - # has a strerror_r that returns `int'. - # This test should segfault on the DEC system. - if test "$cross_compiling" = yes; then - : +fi +if test `eval echo '${'$as_ac_Header'}'` = yes; then + cat >>confdefs.h <<_ACEOF +#define `echo "HAVE_$ac_header" | $as_tr_cpp` 1 +_ACEOF + +fi + +done + +{ echo "$as_me:$LINENO: checking for GNU libc compatible malloc" >&5 +echo $ECHO_N "checking for GNU libc compatible malloc... $ECHO_C" >&6; } +if test "${ac_cv_func_malloc_0_nonnull+set}" = set; then + echo $ECHO_N "(cached) $ECHO_C" >&6 +else + if test "$cross_compiling" = yes; then + ac_cv_func_malloc_0_nonnull=no else cat >conftest.$ac_ext <<_ACEOF /* confdefs.h. */ @@ -23742,14 +23831,16 @@ cat confdefs.h >>conftest.$ac_ext cat >>conftest.$ac_ext <<_ACEOF /* end confdefs.h. */ -$ac_includes_default - extern char *strerror_r (); +#if defined STDC_HEADERS || defined HAVE_STDLIB_H +# include +#else +char *malloc (); +#endif + int main () { -char buf[100]; - char x = *strerror_r (0, buf, sizeof buf); - return ! isalpha (x); +return ! malloc (0); ; return 0; } @@ -23774,57 +23865,118 @@ ac_status=$? echo "$as_me:$LINENO: \$? = $ac_status" >&5 (exit $ac_status); }; }; then - ac_cv_func_strerror_r_char_p=yes + ac_cv_func_malloc_0_nonnull=yes else echo "$as_me: program exited with status $ac_status" >&5 echo "$as_me: failed program was:" >&5 sed 's/^/| /' conftest.$ac_ext >&5 +( exit $ac_status ) +ac_cv_func_malloc_0_nonnull=no fi rm -f core *.core core.conftest.* gmon.out bb.out conftest$ac_exeext conftest.$ac_objext conftest.$ac_ext fi - fi - fi -{ echo "$as_me:$LINENO: result: $ac_cv_func_strerror_r_char_p" >&5 -echo "${ECHO_T}$ac_cv_func_strerror_r_char_p" >&6; } -if test $ac_cv_func_strerror_r_char_p = yes; then +{ echo "$as_me:$LINENO: result: $ac_cv_func_malloc_0_nonnull" >&5 +echo "${ECHO_T}$ac_cv_func_malloc_0_nonnull" >&6; } +if test $ac_cv_func_malloc_0_nonnull = yes; then cat >>confdefs.h <<\_ACEOF -#define STRERROR_R_CHAR_P 1 +#define HAVE_MALLOC 1 _ACEOF -fi - - - +else + cat >>confdefs.h <<\_ACEOF +#define HAVE_MALLOC 0 +_ACEOF + case " $LIBOBJS " in + *" malloc.$ac_objext "* ) ;; + *) LIBOBJS="$LIBOBJS malloc.$ac_objext" + ;; +esac +cat >>confdefs.h <<\_ACEOF +#define malloc rpl_malloc +_ACEOF +fi +{ echo "$as_me:$LINENO: checking whether strerror_r is declared" >&5 +echo $ECHO_N "checking whether strerror_r is declared... $ECHO_C" >&6; } +if test "${ac_cv_have_decl_strerror_r+set}" = set; then + echo $ECHO_N "(cached) $ECHO_C" >&6 +else + cat >conftest.$ac_ext <<_ACEOF +/* confdefs.h. */ +_ACEOF +cat confdefs.h >>conftest.$ac_ext +cat >>conftest.$ac_ext <<_ACEOF +/* end confdefs.h. */ +$ac_includes_default +int +main () +{ +#ifndef strerror_r + (void) strerror_r; +#endif + ; + return 0; +} +_ACEOF +rm -f conftest.$ac_objext +if { (ac_try="$ac_compile" +case "(($ac_try" in + *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;; + *) ac_try_echo=$ac_try;; +esac +eval "echo \"\$as_me:$LINENO: $ac_try_echo\"") >&5 + (eval "$ac_compile") 2>conftest.er1 + ac_status=$? + grep -v '^ *+' conftest.er1 >conftest.err + rm -f conftest.er1 + cat conftest.err >&5 + echo "$as_me:$LINENO: \$? = $ac_status" >&5 + (exit $ac_status); } && { + test -z "$ac_c_werror_flag" || + test ! -s conftest.err + } && test -s conftest.$ac_objext; then + ac_cv_have_decl_strerror_r=yes +else + echo "$as_me: failed program was:" >&5 +sed 's/^/| /' conftest.$ac_ext >&5 + ac_cv_have_decl_strerror_r=no +fi +rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext +fi +{ echo "$as_me:$LINENO: result: $ac_cv_have_decl_strerror_r" >&5 +echo "${ECHO_T}$ac_cv_have_decl_strerror_r" >&6; } +if test $ac_cv_have_decl_strerror_r = yes; then -for ac_func in \ - fdatasync \ - hstrerror \ - strerror \ - mtrace \ - strndup \ - strlcpy \ - strsignal \ - inet_aton \ - inet_ntop \ - inet_pton \ - setproctitle \ - sysctlbyname \ +cat >>confdefs.h <<_ACEOF +#define HAVE_DECL_STRERROR_R 1 +_ACEOF + +else + cat >>confdefs.h <<_ACEOF +#define HAVE_DECL_STRERROR_R 0 +_ACEOF + + +fi + + + +for ac_func in strerror_r do as_ac_var=`echo "ac_cv_func_$ac_func" | $as_tr_sh` { echo "$as_me:$LINENO: checking for $ac_func" >&5 @@ -23917,13 +24069,15 @@ fi done - -{ echo "$as_me:$LINENO: checking whether hstrerror is declared" >&5 -echo $ECHO_N "checking whether hstrerror is declared... $ECHO_C" >&6; } -if test "${ac_cv_have_decl_hstrerror+set}" = set; then +{ echo "$as_me:$LINENO: checking whether strerror_r returns char *" >&5 +echo $ECHO_N "checking whether strerror_r returns char *... $ECHO_C" >&6; } +if test "${ac_cv_func_strerror_r_char_p+set}" = set; then echo $ECHO_N "(cached) $ECHO_C" >&6 else - cat >conftest.$ac_ext <<_ACEOF + + ac_cv_func_strerror_r_char_p=no + if test $ac_cv_have_decl_strerror_r = yes; then + cat >conftest.$ac_ext <<_ACEOF /* confdefs.h. */ _ACEOF cat confdefs.h >>conftest.$ac_ext @@ -23933,9 +24087,11 @@ int main () { -#ifndef hstrerror - (void) hstrerror; -#endif + + char buf[100]; + char x = *strerror_r (0, buf, sizeof buf); + char *p = strerror_r (0, buf, sizeof buf); + return !p || x; ; return 0; @@ -23958,36 +24114,24 @@ test -z "$ac_c_werror_flag" || test ! -s conftest.err } && test -s conftest.$ac_objext; then - ac_cv_have_decl_hstrerror=yes + ac_cv_func_strerror_r_char_p=yes else echo "$as_me: failed program was:" >&5 sed 's/^/| /' conftest.$ac_ext >&5 - ac_cv_have_decl_hstrerror=no -fi -rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext fi -{ echo "$as_me:$LINENO: result: $ac_cv_have_decl_hstrerror" >&5 -echo "${ECHO_T}$ac_cv_have_decl_hstrerror" >&6; } -if test $ac_cv_have_decl_hstrerror = yes; then - -cat >>confdefs.h <<_ACEOF -#define HAVE_DECL_HSTRERROR 1 -_ACEOF - - -else - cat >>confdefs.h <<_ACEOF -#define HAVE_DECL_HSTRERROR 0 -_ACEOF - -fi -{ echo "$as_me:$LINENO: checking whether strsignal is declared" >&5 -echo $ECHO_N "checking whether strsignal is declared... $ECHO_C" >&6; } -if test "${ac_cv_have_decl_strsignal+set}" = set; then - echo $ECHO_N "(cached) $ECHO_C" >&6 +rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext + else + # strerror_r is not declared. Choose between + # systems that have relatively inaccessible declarations for the + # function. BeOS and DEC UNIX 4.0 fall in this category, but the + # former has a strerror_r that returns char*, while the latter + # has a strerror_r that returns `int'. + # This test should segfault on the DEC system. + if test "$cross_compiling" = yes; then + : else cat >conftest.$ac_ext <<_ACEOF /* confdefs.h. */ @@ -23996,131 +24140,88 @@ cat >>conftest.$ac_ext <<_ACEOF /* end confdefs.h. */ $ac_includes_default + extern char *strerror_r (); int main () { -#ifndef strsignal - (void) strsignal; -#endif - +char buf[100]; + char x = *strerror_r (0, buf, sizeof buf); + return ! isalpha (x); ; return 0; } _ACEOF -rm -f conftest.$ac_objext -if { (ac_try="$ac_compile" +rm -f conftest$ac_exeext +if { (ac_try="$ac_link" case "(($ac_try" in *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;; *) ac_try_echo=$ac_try;; esac eval "echo \"\$as_me:$LINENO: $ac_try_echo\"") >&5 - (eval "$ac_compile") 2>conftest.er1 + (eval "$ac_link") 2>&5 ac_status=$? - grep -v '^ *+' conftest.er1 >conftest.err - rm -f conftest.er1 - cat conftest.err >&5 echo "$as_me:$LINENO: \$? = $ac_status" >&5 - (exit $ac_status); } && { - test -z "$ac_c_werror_flag" || - test ! -s conftest.err - } && test -s conftest.$ac_objext; then - ac_cv_have_decl_strsignal=yes + (exit $ac_status); } && { ac_try='./conftest$ac_exeext' + { (case "(($ac_try" in + *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;; + *) ac_try_echo=$ac_try;; +esac +eval "echo \"\$as_me:$LINENO: $ac_try_echo\"") >&5 + (eval "$ac_try") 2>&5 + ac_status=$? + echo "$as_me:$LINENO: \$? = $ac_status" >&5 + (exit $ac_status); }; }; then + ac_cv_func_strerror_r_char_p=yes else - echo "$as_me: failed program was:" >&5 + echo "$as_me: program exited with status $ac_status" >&5 +echo "$as_me: failed program was:" >&5 sed 's/^/| /' conftest.$ac_ext >&5 - ac_cv_have_decl_strsignal=no +fi +rm -f core *.core core.conftest.* gmon.out bb.out conftest$ac_exeext conftest.$ac_objext conftest.$ac_ext fi -rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext + + fi + fi -{ echo "$as_me:$LINENO: result: $ac_cv_have_decl_strsignal" >&5 -echo "${ECHO_T}$ac_cv_have_decl_strsignal" >&6; } -if test $ac_cv_have_decl_strsignal = yes; then +{ echo "$as_me:$LINENO: result: $ac_cv_func_strerror_r_char_p" >&5 +echo "${ECHO_T}$ac_cv_func_strerror_r_char_p" >&6; } +if test $ac_cv_func_strerror_r_char_p = yes; then -cat >>confdefs.h <<_ACEOF -#define HAVE_DECL_STRSIGNAL 1 +cat >>confdefs.h <<\_ACEOF +#define STRERROR_R_CHAR_P 1 _ACEOF +fi -else - cat >>confdefs.h <<_ACEOF -#define HAVE_DECL_STRSIGNAL 0 -_ACEOF -fi -{ echo "$as_me:$LINENO: checking whether sys_siglist is declared" >&5 -echo $ECHO_N "checking whether sys_siglist is declared... $ECHO_C" >&6; } -if test "${ac_cv_have_decl_sys_siglist+set}" = set; then - echo $ECHO_N "(cached) $ECHO_C" >&6 -else - cat >conftest.$ac_ext <<_ACEOF -/* confdefs.h. */ -_ACEOF -cat confdefs.h >>conftest.$ac_ext -cat >>conftest.$ac_ext <<_ACEOF -/* end confdefs.h. */ -$ac_includes_default -int -main () -{ -#ifndef sys_siglist - (void) sys_siglist; -#endif - ; - return 0; -} -_ACEOF -rm -f conftest.$ac_objext -if { (ac_try="$ac_compile" -case "(($ac_try" in - *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;; - *) ac_try_echo=$ac_try;; -esac -eval "echo \"\$as_me:$LINENO: $ac_try_echo\"") >&5 - (eval "$ac_compile") 2>conftest.er1 - ac_status=$? - grep -v '^ *+' conftest.er1 >conftest.err - rm -f conftest.er1 - cat conftest.err >&5 - echo "$as_me:$LINENO: \$? = $ac_status" >&5 - (exit $ac_status); } && { - test -z "$ac_c_werror_flag" || - test ! -s conftest.err - } && test -s conftest.$ac_objext; then - ac_cv_have_decl_sys_siglist=yes -else - echo "$as_me: failed program was:" >&5 -sed 's/^/| /' conftest.$ac_ext >&5 - ac_cv_have_decl_sys_siglist=no -fi -rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext -fi -{ echo "$as_me:$LINENO: result: $ac_cv_have_decl_sys_siglist" >&5 -echo "${ECHO_T}$ac_cv_have_decl_sys_siglist" >&6; } -if test $ac_cv_have_decl_sys_siglist = yes; then -cat >>confdefs.h <<_ACEOF -#define HAVE_DECL_SYS_SIGLIST 1 -_ACEOF -else - cat >>confdefs.h <<_ACEOF -#define HAVE_DECL_SYS_SIGLIST 0 -_ACEOF -fi +for ac_func in \ + fdatasync \ + hstrerror \ + strerror \ + mtrace \ + strndup \ + strlcpy \ + strsignal \ + inet_aton \ + inet_ntop \ + inet_pton \ + setproctitle \ + sysctlbyname \ -for ac_func in unsetenv do as_ac_var=`echo "ac_cv_func_$ac_func" | $as_tr_sh` { echo "$as_me:$LINENO: checking for $ac_func" >&5 @@ -24209,74 +24310,42 @@ cat >>confdefs.h <<_ACEOF #define `echo "HAVE_$ac_func" | $as_tr_cpp` 1 _ACEOF - have_unsetenv=yes -fi -done - if test "x$have_unsetenv" = "xyes"; then - HAVE_UNSETENV_TRUE= - HAVE_UNSETENV_FALSE='#' -else - HAVE_UNSETENV_TRUE='#' - HAVE_UNSETENV_FALSE= fi +done - - - -ac_ext=c -ac_cpp='$CPP $CPPFLAGS' -ac_compile='$CC -c $CFLAGS $CPPFLAGS conftest.$ac_ext >&5' -ac_link='$CC -o conftest$ac_exeext $CFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $LIBS >&5' -ac_compiler_gnu=$ac_cv_c_compiler_gnu - -acx_pthread_ok=no - -# We used to check for pthread.h first, but this fails if pthread.h -# requires special compiler flags (e.g. on True64 or Sequent). -# It gets checked for in the link test anyway. - -# First of all, check if the user has set any of the PTHREAD_LIBS, -# etcetera environment variables, and if threads linking works using -# them: -if test x"$PTHREAD_LIBS$PTHREAD_CFLAGS" != x; then - save_CFLAGS="$CFLAGS" - CFLAGS="$CFLAGS $PTHREAD_CFLAGS" - save_LIBS="$LIBS" - LIBS="$PTHREAD_LIBS $LIBS" - { echo "$as_me:$LINENO: checking for pthread_join in LIBS=$PTHREAD_LIBS with CFLAGS=$PTHREAD_CFLAGS" >&5 -echo $ECHO_N "checking for pthread_join in LIBS=$PTHREAD_LIBS with CFLAGS=$PTHREAD_CFLAGS... $ECHO_C" >&6; } - cat >conftest.$ac_ext <<_ACEOF +{ echo "$as_me:$LINENO: checking whether hstrerror is declared" >&5 +echo $ECHO_N "checking whether hstrerror is declared... $ECHO_C" >&6; } +if test "${ac_cv_have_decl_hstrerror+set}" = set; then + echo $ECHO_N "(cached) $ECHO_C" >&6 +else + cat >conftest.$ac_ext <<_ACEOF /* confdefs.h. */ _ACEOF cat confdefs.h >>conftest.$ac_ext cat >>conftest.$ac_ext <<_ACEOF /* end confdefs.h. */ - -/* Override any GCC internal prototype to avoid an error. - Use char because int might match the return type of a GCC - builtin and then its argument prototype would still apply. */ -#ifdef __cplusplus -extern "C" -#endif -char pthread_join (); +$ac_includes_default int main () { -return pthread_join (); +#ifndef hstrerror + (void) hstrerror; +#endif + ; return 0; } -_ACEOF -rm -f conftest.$ac_objext conftest$ac_exeext -if { (ac_try="$ac_link" +_ACEOF +rm -f conftest.$ac_objext +if { (ac_try="$ac_compile" case "(($ac_try" in *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;; *) ac_try_echo=$ac_try;; esac eval "echo \"\$as_me:$LINENO: $ac_try_echo\"") >&5 - (eval "$ac_link") 2>conftest.er1 + (eval "$ac_compile") 2>conftest.er1 ac_status=$? grep -v '^ *+' conftest.er1 >conftest.err rm -f conftest.er1 @@ -24285,133 +24354,64 @@ (exit $ac_status); } && { test -z "$ac_c_werror_flag" || test ! -s conftest.err - } && test -s conftest$ac_exeext && - $as_test_x conftest$ac_exeext; then - acx_pthread_ok=yes + } && test -s conftest.$ac_objext; then + ac_cv_have_decl_hstrerror=yes else echo "$as_me: failed program was:" >&5 sed 's/^/| /' conftest.$ac_ext >&5 - + ac_cv_have_decl_hstrerror=no fi -rm -f core conftest.err conftest.$ac_objext conftest_ipa8_conftest.oo \ - conftest$ac_exeext conftest.$ac_ext - { echo "$as_me:$LINENO: result: $acx_pthread_ok" >&5 -echo "${ECHO_T}$acx_pthread_ok" >&6; } - if test x"$acx_pthread_ok" = xno; then - PTHREAD_LIBS="" - PTHREAD_CFLAGS="" - fi - LIBS="$save_LIBS" - CFLAGS="$save_CFLAGS" +rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext fi +{ echo "$as_me:$LINENO: result: $ac_cv_have_decl_hstrerror" >&5 +echo "${ECHO_T}$ac_cv_have_decl_hstrerror" >&6; } +if test $ac_cv_have_decl_hstrerror = yes; then -# We must check for the threads library under a number of different -# names; the ordering is very important because some systems -# (e.g. DEC) have both -lpthread and -lpthreads, where one of the -# libraries is broken (non-POSIX). - -# Create a list of thread flags to try. Items starting with a "-" are -# C compiler flags, and other items are library names, except for "none" -# which indicates that we try without any flags at all. - -acx_pthread_flags="pthreads none -Kthread -kthread lthread -pthread -pthreads -mthreads pthread --thread-safe -mt" - -# The ordering *is* (sometimes) important. Some notes on the -# individual items follow: - -# pthreads: AIX (must check this before -lpthread) -# none: in case threads are in libc; should be tried before -Kthread and -# other compiler flags to prevent continual compiler warnings -# -Kthread: Sequent (threads in libc, but -Kthread needed for pthread.h) -# -kthread: FreeBSD kernel threads (preferred to -pthread since SMP-able) -# lthread: LinuxThreads port on FreeBSD (also preferred to -pthread) -# -pthread: Linux/gcc (kernel threads), BSD/gcc (userland threads) -# -pthreads: Solaris/gcc -# -mthreads: Mingw32/gcc, Lynx/gcc -# -mt: Sun Workshop C (may only link SunOS threads [-lthread], but it -# doesn't hurt to check since this sometimes defines pthreads too; -# also defines -D_REENTRANT) -# pthread: Linux, etcetera -# --thread-safe: KAI C++ - -case "${host_cpu}-${host_os}" in - *solaris*) - - # On Solaris (at least, for some versions), libc contains stubbed - # (non-functional) versions of the pthreads routines, so link-based - # tests will erroneously succeed. (We need to link with -pthread or - # -lpthread.) (The stubs are missing pthread_cleanup_push, or rather - # a function called by this macro, so we could check for that, but - # who knows whether they'll stub that too in a future libc.) So, - # we'll just look for -pthreads and -lpthread first: - - acx_pthread_flags="-pthread -pthreads pthread -mt $acx_pthread_flags" - ;; -esac - -if test x"$acx_pthread_ok" = xno; then -for flag in $acx_pthread_flags; do - - case $flag in - none) - { echo "$as_me:$LINENO: checking whether pthreads work without any flags" >&5 -echo $ECHO_N "checking whether pthreads work without any flags... $ECHO_C" >&6; } - ;; +cat >>confdefs.h <<_ACEOF +#define HAVE_DECL_HSTRERROR 1 +_ACEOF - -*) - { echo "$as_me:$LINENO: checking whether pthreads work with $flag" >&5 -echo $ECHO_N "checking whether pthreads work with $flag... $ECHO_C" >&6; } - PTHREAD_CFLAGS="$flag" - ;; - *) - { echo "$as_me:$LINENO: checking for the pthreads library -l$flag" >&5 -echo $ECHO_N "checking for the pthreads library -l$flag... $ECHO_C" >&6; } - PTHREAD_LIBS="-l$flag" - ;; - esac +else + cat >>confdefs.h <<_ACEOF +#define HAVE_DECL_HSTRERROR 0 +_ACEOF - save_LIBS="$LIBS" - save_CFLAGS="$CFLAGS" - LIBS="$PTHREAD_LIBS $LIBS" - CFLAGS="$CFLAGS $PTHREAD_CFLAGS" - # Check for various functions. We must include pthread.h, - # since some functions may be macros. (On the Sequent, we - # need a special flag -Kthread to make this header compile.) - # We check for pthread_join because it is in -lpthread on IRIX - # while pthread_create is in libc. We check for pthread_attr_init - # due to DEC craziness with -lpthreads. We check for - # pthread_cleanup_push because it is one of the few pthread - # functions on Solaris that doesn't have a non-functional libc stub. - # We try pthread_create on general principles. - cat >conftest.$ac_ext <<_ACEOF +fi +{ echo "$as_me:$LINENO: checking whether strsignal is declared" >&5 +echo $ECHO_N "checking whether strsignal is declared... $ECHO_C" >&6; } +if test "${ac_cv_have_decl_strsignal+set}" = set; then + echo $ECHO_N "(cached) $ECHO_C" >&6 +else + cat >conftest.$ac_ext <<_ACEOF /* confdefs.h. */ _ACEOF cat confdefs.h >>conftest.$ac_ext cat >>conftest.$ac_ext <<_ACEOF /* end confdefs.h. */ -#include +$ac_includes_default int main () { -pthread_t th; pthread_join(th, 0); - pthread_attr_init(0); pthread_cleanup_push(0, 0); - pthread_create(0,0,0,0); pthread_cleanup_pop(0); +#ifndef strsignal + (void) strsignal; +#endif + ; return 0; } _ACEOF -rm -f conftest.$ac_objext conftest$ac_exeext -if { (ac_try="$ac_link" +rm -f conftest.$ac_objext +if { (ac_try="$ac_compile" case "(($ac_try" in *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;; *) ac_try_echo=$ac_try;; esac eval "echo \"\$as_me:$LINENO: $ac_try_echo\"") >&5 - (eval "$ac_link") 2>conftest.er1 + (eval "$ac_compile") 2>conftest.er1 ac_status=$? grep -v '^ *+' conftest.er1 >conftest.err rm -f conftest.er1 @@ -24420,67 +24420,64 @@ (exit $ac_status); } && { test -z "$ac_c_werror_flag" || test ! -s conftest.err - } && test -s conftest$ac_exeext && - $as_test_x conftest$ac_exeext; then - acx_pthread_ok=yes + } && test -s conftest.$ac_objext; then + ac_cv_have_decl_strsignal=yes else echo "$as_me: failed program was:" >&5 sed 's/^/| /' conftest.$ac_ext >&5 - + ac_cv_have_decl_strsignal=no fi -rm -f core conftest.err conftest.$ac_objext conftest_ipa8_conftest.oo \ - conftest$ac_exeext conftest.$ac_ext +rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext +fi +{ echo "$as_me:$LINENO: result: $ac_cv_have_decl_strsignal" >&5 +echo "${ECHO_T}$ac_cv_have_decl_strsignal" >&6; } +if test $ac_cv_have_decl_strsignal = yes; then - LIBS="$save_LIBS" - CFLAGS="$save_CFLAGS" +cat >>confdefs.h <<_ACEOF +#define HAVE_DECL_STRSIGNAL 1 +_ACEOF - { echo "$as_me:$LINENO: result: $acx_pthread_ok" >&5 -echo "${ECHO_T}$acx_pthread_ok" >&6; } - if test "x$acx_pthread_ok" = xyes; then - break; - fi - PTHREAD_LIBS="" - PTHREAD_CFLAGS="" -done -fi +else + cat >>confdefs.h <<_ACEOF +#define HAVE_DECL_STRSIGNAL 0 +_ACEOF -# Various other checks: -if test "x$acx_pthread_ok" = xyes; then - save_LIBS="$LIBS" - LIBS="$PTHREAD_LIBS $LIBS" - save_CFLAGS="$CFLAGS" - CFLAGS="$CFLAGS $PTHREAD_CFLAGS" - # Detect AIX lossage: threads are created detached by default - # and the JOINABLE attribute has a nonstandard name (UNDETACHED). - { echo "$as_me:$LINENO: checking for joinable pthread attribute" >&5 -echo $ECHO_N "checking for joinable pthread attribute... $ECHO_C" >&6; } - cat >conftest.$ac_ext <<_ACEOF +fi +{ echo "$as_me:$LINENO: checking whether sys_siglist is declared" >&5 +echo $ECHO_N "checking whether sys_siglist is declared... $ECHO_C" >&6; } +if test "${ac_cv_have_decl_sys_siglist+set}" = set; then + echo $ECHO_N "(cached) $ECHO_C" >&6 +else + cat >conftest.$ac_ext <<_ACEOF /* confdefs.h. */ _ACEOF cat confdefs.h >>conftest.$ac_ext cat >>conftest.$ac_ext <<_ACEOF /* end confdefs.h. */ -#include +$ac_includes_default int main () { -int attr=PTHREAD_CREATE_JOINABLE; +#ifndef sys_siglist + (void) sys_siglist; +#endif + ; return 0; } _ACEOF -rm -f conftest.$ac_objext conftest$ac_exeext -if { (ac_try="$ac_link" +rm -f conftest.$ac_objext +if { (ac_try="$ac_compile" case "(($ac_try" in *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;; *) ac_try_echo=$ac_try;; esac eval "echo \"\$as_me:$LINENO: $ac_try_echo\"") >&5 - (eval "$ac_link") 2>conftest.er1 + (eval "$ac_compile") 2>conftest.er1 ac_status=$? grep -v '^ *+' conftest.er1 >conftest.err rm -f conftest.er1 @@ -24489,30 +24486,86 @@ (exit $ac_status); } && { test -z "$ac_c_werror_flag" || test ! -s conftest.err - } && test -s conftest$ac_exeext && - $as_test_x conftest$ac_exeext; then - ok=PTHREAD_CREATE_JOINABLE + } && test -s conftest.$ac_objext; then + ac_cv_have_decl_sys_siglist=yes else echo "$as_me: failed program was:" >&5 sed 's/^/| /' conftest.$ac_ext >&5 - ok=unknown -fi + ac_cv_have_decl_sys_siglist=no +fi + +rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext +fi +{ echo "$as_me:$LINENO: result: $ac_cv_have_decl_sys_siglist" >&5 +echo "${ECHO_T}$ac_cv_have_decl_sys_siglist" >&6; } +if test $ac_cv_have_decl_sys_siglist = yes; then + +cat >>confdefs.h <<_ACEOF +#define HAVE_DECL_SYS_SIGLIST 1 +_ACEOF + + +else + cat >>confdefs.h <<_ACEOF +#define HAVE_DECL_SYS_SIGLIST 0 +_ACEOF + + +fi + + + + +for ac_func in unsetenv +do +as_ac_var=`echo "ac_cv_func_$ac_func" | $as_tr_sh` +{ echo "$as_me:$LINENO: checking for $ac_func" >&5 +echo $ECHO_N "checking for $ac_func... $ECHO_C" >&6; } +if { as_var=$as_ac_var; eval "test \"\${$as_var+set}\" = set"; }; then + echo $ECHO_N "(cached) $ECHO_C" >&6 +else + cat >conftest.$ac_ext <<_ACEOF +/* confdefs.h. */ +_ACEOF +cat confdefs.h >>conftest.$ac_ext +cat >>conftest.$ac_ext <<_ACEOF +/* end confdefs.h. */ +/* Define $ac_func to an innocuous variant, in case declares $ac_func. + For example, HP-UX 11i declares gettimeofday. */ +#define $ac_func innocuous_$ac_func + +/* System header to define __stub macros and hopefully few prototypes, + which can conflict with char $ac_func (); below. + Prefer to if __STDC__ is defined, since + exists even on freestanding compilers. */ + +#ifdef __STDC__ +# include +#else +# include +#endif + +#undef $ac_func + +/* Override any GCC internal prototype to avoid an error. + Use char because int might match the return type of a GCC + builtin and then its argument prototype would still apply. */ +#ifdef __cplusplus +extern "C" +#endif +char $ac_func (); +/* The GNU C library defines this for functions which it implements + to always fail with ENOSYS. Some functions are actually named + something starting with __ and the normal name is an alias. */ +#if defined __stub_$ac_func || defined __stub___$ac_func +choke me +#endif -rm -f core conftest.err conftest.$ac_objext conftest_ipa8_conftest.oo \ - conftest$ac_exeext conftest.$ac_ext - if test x"$ok" = xunknown; then - cat >conftest.$ac_ext <<_ACEOF -/* confdefs.h. */ -_ACEOF -cat confdefs.h >>conftest.$ac_ext -cat >>conftest.$ac_ext <<_ACEOF -/* end confdefs.h. */ -#include int main () { -int attr=PTHREAD_CREATE_UNDETACHED; +return $ac_func (); ; return 0; } @@ -24535,236 +24588,80 @@ test ! -s conftest.err } && test -s conftest$ac_exeext && $as_test_x conftest$ac_exeext; then - ok=PTHREAD_CREATE_UNDETACHED + eval "$as_ac_var=yes" else echo "$as_me: failed program was:" >&5 sed 's/^/| /' conftest.$ac_ext >&5 - ok=unknown + eval "$as_ac_var=no" fi rm -f core conftest.err conftest.$ac_objext conftest_ipa8_conftest.oo \ conftest$ac_exeext conftest.$ac_ext - fi - if test x"$ok" != xPTHREAD_CREATE_JOINABLE; then - -cat >>confdefs.h <<\_ACEOF -#define PTHREAD_CREATE_JOINABLE $ok -_ACEOF - - fi - { echo "$as_me:$LINENO: result: ${ok}" >&5 -echo "${ECHO_T}${ok}" >&6; } - if test x"$ok" = xunknown; then - { echo "$as_me:$LINENO: WARNING: we do not know how to create joinable pthreads" >&5 -echo "$as_me: WARNING: we do not know how to create joinable pthreads" >&2;} - fi - - { echo "$as_me:$LINENO: checking if more special flags are required for pthreads" >&5 -echo $ECHO_N "checking if more special flags are required for pthreads... $ECHO_C" >&6; } - flag=no - case "${host_cpu}-${host_os}" in - *-aix* | *-freebsd*) flag="-D_THREAD_SAFE";; - *solaris* | alpha*-osf*) flag="-D_REENTRANT";; - esac - { echo "$as_me:$LINENO: result: ${flag}" >&5 -echo "${ECHO_T}${flag}" >&6; } - if test "x$flag" != xno; then - PTHREAD_CFLAGS="$flag $PTHREAD_CFLAGS" - fi - - LIBS="$save_LIBS" - CFLAGS="$save_CFLAGS" - - # More AIX lossage: must compile with cc_r - # Extract the first word of "cc_r", so it can be a program name with args. -set dummy cc_r; ac_word=$2 -{ echo "$as_me:$LINENO: checking for $ac_word" >&5 -echo $ECHO_N "checking for $ac_word... $ECHO_C" >&6; } -if test "${ac_cv_prog_PTHREAD_CC+set}" = set; then - echo $ECHO_N "(cached) $ECHO_C" >&6 -else - if test -n "$PTHREAD_CC"; then - ac_cv_prog_PTHREAD_CC="$PTHREAD_CC" # Let the user override the test. -else -as_save_IFS=$IFS; IFS=$PATH_SEPARATOR -for as_dir in $PATH -do - IFS=$as_save_IFS - test -z "$as_dir" && as_dir=. - for ac_exec_ext in '' $ac_executable_extensions; do - if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then - ac_cv_prog_PTHREAD_CC="cc_r" - echo "$as_me:$LINENO: found $as_dir/$ac_word$ac_exec_ext" >&5 - break 2 - fi -done -done -IFS=$as_save_IFS - - test -z "$ac_cv_prog_PTHREAD_CC" && ac_cv_prog_PTHREAD_CC="${CC}" -fi fi -PTHREAD_CC=$ac_cv_prog_PTHREAD_CC -if test -n "$PTHREAD_CC"; then - { echo "$as_me:$LINENO: result: $PTHREAD_CC" >&5 -echo "${ECHO_T}$PTHREAD_CC" >&6; } -else - { echo "$as_me:$LINENO: result: no" >&5 -echo "${ECHO_T}no" >&6; } +ac_res=`eval echo '${'$as_ac_var'}'` + { echo "$as_me:$LINENO: result: $ac_res" >&5 +echo "${ECHO_T}$ac_res" >&6; } +if test `eval echo '${'$as_ac_var'}'` = yes; then + cat >>confdefs.h <<_ACEOF +#define `echo "HAVE_$ac_func" | $as_tr_cpp` 1 +_ACEOF + have_unsetenv=yes fi +done - + if test "x$have_unsetenv" = "xyes"; then + HAVE_UNSETENV_TRUE= + HAVE_UNSETENV_FALSE='#' else - PTHREAD_CC="$CC" + HAVE_UNSETENV_TRUE='#' + HAVE_UNSETENV_FALSE= fi -# Finally, execute ACTION-IF-FOUND/ACTION-IF-NOT-FOUND: -if test x"$acx_pthread_ok" = xyes; then - -cat >>confdefs.h <<\_ACEOF -#define HAVE_PTHREAD 1 -_ACEOF - - : -else - acx_pthread_ok=no - { { echo "$as_me:$LINENO: error: Error: Cannot figure out how to use pthreads!" >&5 -echo "$as_me: error: Error: Cannot figure out how to use pthreads!" >&2;} - { (exit 1); exit 1; }; } -fi ac_ext=c ac_cpp='$CPP $CPPFLAGS' ac_compile='$CC -c $CFLAGS $CPPFLAGS conftest.$ac_ext >&5' ac_link='$CC -o conftest$ac_exeext $CFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $LIBS >&5' ac_compiler_gnu=$ac_cv_c_compiler_gnu +acx_pthread_ok=no +# We used to check for pthread.h first, but this fails if pthread.h +# requires special compiler flags (e.g. on True64 or Sequent). +# It gets checked for in the link test anyway. -# Always define WITH_PTHREADS if we make it this far - -cat >>confdefs.h <<\_ACEOF -#define WITH_PTHREADS 1 -_ACEOF - -LDFLAGS="$LDFLAGS " -CFLAGS="$CFLAGS $PTHREAD_CFLAGS" -LIBS="$PTHREAD_LIBS $LIBS" - - - ac_bluegene_loaded=no - - -# Check whether --with-db2-dir was given. -if test "${with_db2_dir+set}" = set; then - withval=$with_db2_dir; trydb2dir=$withval -fi - - - # test for bluegene emulation mode - - # Check whether --enable-bluegene-emulation was given. -if test "${enable_bluegene_emulation+set}" = set; then - enableval=$enable_bluegene_emulation; case "$enableval" in - yes) bluegene_emulation=yes ;; - no) bluegene_emulation=no ;; - *) { { echo "$as_me:$LINENO: error: bad value \"$enableval\" for --enable-bluegene-emulation" >&5 -echo "$as_me: error: bad value \"$enableval\" for --enable-bluegene-emulation" >&2;} - { (exit 1); exit 1; }; } ;; - esac -fi - - - # Check whether --enable-bgl-emulation was given. -if test "${enable_bgl_emulation+set}" = set; then - enableval=$enable_bgl_emulation; case "$enableval" in - yes) bgl_emulation=yes ;; - no) bgl_emulation=no ;; - *) { { echo "$as_me:$LINENO: error: bad value \"$enableval\" for --enable-bgl-emulation" >&5 -echo "$as_me: error: bad value \"$enableval\" for --enable-bgl-emulation" >&2;} - { (exit 1); exit 1; }; } ;; - esac -fi - - - if test "x$bluegene_emulation" = "xyes" -o "x$bgl_emulation" = "xyes"; then - -cat >>confdefs.h <<\_ACEOF -#define HAVE_3D 1 -_ACEOF - - -cat >>confdefs.h <<\_ACEOF -#define HAVE_BG 1 -_ACEOF - - -cat >>confdefs.h <<\_ACEOF -#define HAVE_BGL 1 -_ACEOF - - -cat >>confdefs.h <<\_ACEOF -#define HAVE_FRONT_END 1 -_ACEOF - - { echo "$as_me:$LINENO: Running in BG/L emulation mode" >&5 -echo "$as_me: Running in BG/L emulation mode" >&6;} - bg_default_dirs="" - #define ac_bluegene_loaded so we don't load another bluegene conf - ac_bluegene_loaded=yes - else - bg_default_dirs="/bgl/BlueLight/ppcfloor/bglsys /opt/IBM/db2/V8.1 /u/bgdb2cli/sqllib /home/bgdb2cli/sqllib" - fi - - for bg_dir in $trydb2dir "" $bg_default_dirs; do - # Skip directories that don't exist - if test ! -z "$bg_dir" -a ! -d "$bg_dir" ; then - continue; - fi - - # Search for required BG API libraries in the directory - if test -z "$have_bg_ar" -a -f "$bg_dir/lib64/libbglbridge.so" ; then - have_bg_ar=yes - bg_bridge_so="$bg_dir/lib64/libbglbridge.so" - bg_ldflags="$bg_ldflags -L$bg_dir/lib64 -L/usr/lib64 -Wl,--unresolved-symbols=ignore-in-shared-libs -lbglbridge -lbgldb -ltableapi -lbglmachine -lexpat -lsaymessage" - fi - - # Search for required DB2 library in the directory - if test -z "$have_db2" -a -f "$bg_dir/lib64/libdb2.so" ; then - have_db2=yes - bg_db2_so="$bg_dir/lib64/libdb2.so" - bg_ldflags="$bg_ldflags -L$bg_dir/lib64 -ldb2" - fi - - # Search for headers in the directory - if test -z "$have_bg_hdr" -a -f "$bg_dir/include/rm_api.h" ; then - have_bg_hdr=yes - bg_includes="-I$bg_dir/include" - fi - done - - if test ! -z "$have_bg_ar" -a ! -z "$have_bg_hdr" -a ! -z "$have_db2" ; then - # ac_with_readline="no" - # Test to make sure the api is good - have_bg_files=yes - saved_LDFLAGS="$LDFLAGS" - LDFLAGS="$saved_LDFLAGS $bg_ldflags" - cat >conftest.$ac_ext <<_ACEOF +# First of all, check if the user has set any of the PTHREAD_LIBS, +# etcetera environment variables, and if threads linking works using +# them: +if test x"$PTHREAD_LIBS$PTHREAD_CFLAGS" != x; then + save_CFLAGS="$CFLAGS" + CFLAGS="$CFLAGS $PTHREAD_CFLAGS" + save_LIBS="$LIBS" + LIBS="$PTHREAD_LIBS $LIBS" + { echo "$as_me:$LINENO: checking for pthread_join in LIBS=$PTHREAD_LIBS with CFLAGS=$PTHREAD_CFLAGS" >&5 +echo $ECHO_N "checking for pthread_join in LIBS=$PTHREAD_LIBS with CFLAGS=$PTHREAD_CFLAGS... $ECHO_C" >&6; } + cat >conftest.$ac_ext <<_ACEOF /* confdefs.h. */ _ACEOF cat confdefs.h >>conftest.$ac_ext cat >>conftest.$ac_ext <<_ACEOF /* end confdefs.h. */ - int rm_set_serial(char *); + +/* Override any GCC internal prototype to avoid an error. + Use char because int might match the return type of a GCC + builtin and then its argument prototype would still apply. */ +#ifdef __cplusplus +extern "C" +#endif +char pthread_join (); int main () { - rm_set_serial(""); +return pthread_join (); ; return 0; } @@ -24787,166 +24684,188 @@ test ! -s conftest.err } && test -s conftest$ac_exeext && $as_test_x conftest$ac_exeext; then - have_bg_files=yes + acx_pthread_ok=yes else echo "$as_me: failed program was:" >&5 sed 's/^/| /' conftest.$ac_ext >&5 - { { echo "$as_me:$LINENO: error: There is a problem linking to the BG/L api." >&5 -echo "$as_me: error: There is a problem linking to the BG/L api." >&2;} - { (exit 1); exit 1; }; } + fi rm -f core conftest.err conftest.$ac_objext conftest_ipa8_conftest.oo \ conftest$ac_exeext conftest.$ac_ext - LDFLAGS="$saved_LDFLAGS" - fi - - if test ! -z "$have_bg_files" ; then - BG_INCLUDES="$bg_includes" + { echo "$as_me:$LINENO: result: $acx_pthread_ok" >&5 +echo "${ECHO_T}$acx_pthread_ok" >&6; } + if test x"$acx_pthread_ok" = xno; then + PTHREAD_LIBS="" + PTHREAD_CFLAGS="" + fi + LIBS="$save_LIBS" + CFLAGS="$save_CFLAGS" +fi -cat >>confdefs.h <<\_ACEOF -#define HAVE_3D 1 -_ACEOF +# We must check for the threads library under a number of different +# names; the ordering is very important because some systems +# (e.g. DEC) have both -lpthread and -lpthreads, where one of the +# libraries is broken (non-POSIX). +# Create a list of thread flags to try. Items starting with a "-" are +# C compiler flags, and other items are library names, except for "none" +# which indicates that we try without any flags at all. -cat >>confdefs.h <<\_ACEOF -#define HAVE_BG 1 -_ACEOF +acx_pthread_flags="pthreads none -Kthread -kthread lthread -pthread -pthreads -mthreads pthread --thread-safe -mt" +# The ordering *is* (sometimes) important. Some notes on the +# individual items follow: -cat >>confdefs.h <<\_ACEOF -#define HAVE_BGL 1 -_ACEOF +# pthreads: AIX (must check this before -lpthread) +# none: in case threads are in libc; should be tried before -Kthread and +# other compiler flags to prevent continual compiler warnings +# -Kthread: Sequent (threads in libc, but -Kthread needed for pthread.h) +# -kthread: FreeBSD kernel threads (preferred to -pthread since SMP-able) +# lthread: LinuxThreads port on FreeBSD (also preferred to -pthread) +# -pthread: Linux/gcc (kernel threads), BSD/gcc (userland threads) +# -pthreads: Solaris/gcc +# -mthreads: Mingw32/gcc, Lynx/gcc +# -mt: Sun Workshop C (may only link SunOS threads [-lthread], but it +# doesn't hurt to check since this sometimes defines pthreads too; +# also defines -D_REENTRANT) +# pthread: Linux, etcetera +# --thread-safe: KAI C++ +case "${host_cpu}-${host_os}" in + *solaris*) -cat >>confdefs.h <<\_ACEOF -#define HAVE_FRONT_END 1 -_ACEOF + # On Solaris (at least, for some versions), libc contains stubbed + # (non-functional) versions of the pthreads routines, so link-based + # tests will erroneously succeed. (We need to link with -pthread or + # -lpthread.) (The stubs are missing pthread_cleanup_push, or rather + # a function called by this macro, so we could check for that, but + # who knows whether they'll stub that too in a future libc.) So, + # we'll just look for -pthreads and -lpthread first: + acx_pthread_flags="-pthread -pthreads pthread -mt $acx_pthread_flags" + ;; +esac -cat >>confdefs.h <<\_ACEOF -#define HAVE_BG_FILES 1 -_ACEOF +if test x"$acx_pthread_ok" = xno; then +for flag in $acx_pthread_flags; do + case $flag in + none) + { echo "$as_me:$LINENO: checking whether pthreads work without any flags" >&5 +echo $ECHO_N "checking whether pthreads work without any flags... $ECHO_C" >&6; } + ;; + -*) + { echo "$as_me:$LINENO: checking whether pthreads work with $flag" >&5 +echo $ECHO_N "checking whether pthreads work with $flag... $ECHO_C" >&6; } + PTHREAD_CFLAGS="$flag" + ;; -cat >>confdefs.h <<_ACEOF -#define BG_BRIDGE_SO "$bg_bridge_so" -_ACEOF + *) + { echo "$as_me:$LINENO: checking for the pthreads library -l$flag" >&5 +echo $ECHO_N "checking for the pthreads library -l$flag... $ECHO_C" >&6; } + PTHREAD_LIBS="-l$flag" + ;; + esac + save_LIBS="$LIBS" + save_CFLAGS="$CFLAGS" + LIBS="$PTHREAD_LIBS $LIBS" + CFLAGS="$CFLAGS $PTHREAD_CFLAGS" -cat >>confdefs.h <<_ACEOF -#define BG_DB2_SO "$bg_db2_so" + # Check for various functions. We must include pthread.h, + # since some functions may be macros. (On the Sequent, we + # need a special flag -Kthread to make this header compile.) + # We check for pthread_join because it is in -lpthread on IRIX + # while pthread_create is in libc. We check for pthread_attr_init + # due to DEC craziness with -lpthreads. We check for + # pthread_cleanup_push because it is one of the few pthread + # functions on Solaris that doesn't have a non-functional libc stub. + # We try pthread_create on general principles. + cat >conftest.$ac_ext <<_ACEOF +/* confdefs.h. */ _ACEOF - - { echo "$as_me:$LINENO: checking for BG serial value" >&5 -echo $ECHO_N "checking for BG serial value... $ECHO_C" >&6; } - bg_serial="BGL" - -# Check whether --with-bg-serial was given. -if test "${with_bg_serial+set}" = set; then - withval=$with_bg_serial; bg_serial="$withval" -fi - - { echo "$as_me:$LINENO: result: $bg_serial" >&5 -echo "${ECHO_T}$bg_serial" >&6; } - -cat >>confdefs.h <<_ACEOF -#define BG_SERIAL "$bg_serial" +cat confdefs.h >>conftest.$ac_ext +cat >>conftest.$ac_ext <<_ACEOF +/* end confdefs.h. */ +#include +int +main () +{ +pthread_t th; pthread_join(th, 0); + pthread_attr_init(0); pthread_cleanup_push(0, 0); + pthread_create(0,0,0,0); pthread_cleanup_pop(0); + ; + return 0; +} _ACEOF - - #define ac_bluegene_loaded so we don't load another bluegene conf - ac_bluegene_loaded=yes - fi - - +rm -f conftest.$ac_objext conftest$ac_exeext +if { (ac_try="$ac_link" +case "(($ac_try" in + *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;; + *) ac_try_echo=$ac_try;; +esac +eval "echo \"\$as_me:$LINENO: $ac_try_echo\"") >&5 + (eval "$ac_link") 2>conftest.er1 + ac_status=$? + grep -v '^ *+' conftest.er1 >conftest.err + rm -f conftest.er1 + cat conftest.err >&5 + echo "$as_me:$LINENO: \$? = $ac_status" >&5 + (exit $ac_status); } && { + test -z "$ac_c_werror_flag" || + test ! -s conftest.err + } && test -s conftest$ac_exeext && + $as_test_x conftest$ac_exeext; then + acx_pthread_ok=yes +else + echo "$as_me: failed program was:" >&5 +sed 's/^/| /' conftest.$ac_ext >&5 - # test for bluegene emulation mode - # Check whether --enable-bgp-emulation was given. -if test "${enable_bgp_emulation+set}" = set; then - enableval=$enable_bgp_emulation; case "$enableval" in - yes) bgp_emulation=yes ;; - no) bgp_emulation=no ;; - *) { { echo "$as_me:$LINENO: error: bad value \"$enableval\" for --enable-bgp-emulation" >&5 -echo "$as_me: error: bad value \"$enableval\" for --enable-bgp-emulation" >&2;} - { (exit 1); exit 1; }; } ;; - esac fi +rm -f core conftest.err conftest.$ac_objext conftest_ipa8_conftest.oo \ + conftest$ac_exeext conftest.$ac_ext - # Skip if already set - if test "x$ac_bluegene_loaded" = "xyes" ; then - bg_default_dirs="" - elif test "x$bgp_emulation" = "xyes"; then - -cat >>confdefs.h <<\_ACEOF -#define HAVE_3D 1 -_ACEOF - - -cat >>confdefs.h <<\_ACEOF -#define HAVE_BG 1 -_ACEOF - - -cat >>confdefs.h <<\_ACEOF -#define HAVE_BGP 1 -_ACEOF - - -cat >>confdefs.h <<\_ACEOF -#define HAVE_FRONT_END 1 -_ACEOF - - { echo "$as_me:$LINENO: Running in BG/P emulation mode" >&5 -echo "$as_me: Running in BG/P emulation mode" >&6;} - bg_default_dirs="" - #define ac_bluegene_loaded so we don't load another bluegene conf - ac_bluegene_loaded=yes - else - bg_default_dirs="/bgsys/drivers/ppcfloor" - fi - - libname=bgpbridge - - for bg_dir in $trydb2dir "" $bg_default_dirs; do - # Skip directories that don't exist - if test ! -z "$bg_dir" -a ! -d "$bg_dir" ; then - continue; - fi + LIBS="$save_LIBS" + CFLAGS="$save_CFLAGS" - soloc=$bg_dir/lib64/lib$libname.so - # Search for required BG API libraries in the directory - if test -z "$have_bg_ar" -a -f "$soloc" ; then - have_bgp_ar=yes - bg_ldflags="$bg_ldflags -L$bg_dir/lib64 -L/usr/lib64 -Wl,--unresolved-symbols=ignore-in-shared-libs -l$libname" - fi + { echo "$as_me:$LINENO: result: $acx_pthread_ok" >&5 +echo "${ECHO_T}$acx_pthread_ok" >&6; } + if test "x$acx_pthread_ok" = xyes; then + break; + fi - # Search for headers in the directory - if test -z "$have_bg_hdr" -a -f "$bg_dir/include/rm_api.h" ; then - have_bgp_hdr=yes - bg_includes="-I$bg_dir/include" - fi - done + PTHREAD_LIBS="" + PTHREAD_CFLAGS="" +done +fi - if test ! -z "$have_bgp_ar" -a ! -z "$have_bgp_hdr" ; then - # ac_with_readline="no" - # Test to make sure the api is good - saved_LDFLAGS="$LDFLAGS" - LDFLAGS="$saved_LDFLAGS $bg_ldflags" - cat >conftest.$ac_ext <<_ACEOF +# Various other checks: +if test "x$acx_pthread_ok" = xyes; then + save_LIBS="$LIBS" + LIBS="$PTHREAD_LIBS $LIBS" + save_CFLAGS="$CFLAGS" + CFLAGS="$CFLAGS $PTHREAD_CFLAGS" + + # Detect AIX lossage: threads are created detached by default + # and the JOINABLE attribute has a nonstandard name (UNDETACHED). + { echo "$as_me:$LINENO: checking for joinable pthread attribute" >&5 +echo $ECHO_N "checking for joinable pthread attribute... $ECHO_C" >&6; } + cat >conftest.$ac_ext <<_ACEOF /* confdefs.h. */ _ACEOF cat confdefs.h >>conftest.$ac_ext cat >>conftest.$ac_ext <<_ACEOF /* end confdefs.h. */ - int rm_set_serial(char *); +#include int main () { - rm_set_serial(""); +int attr=PTHREAD_CREATE_JOINABLE; ; return 0; } @@ -24969,85 +24888,169 @@ test ! -s conftest.err } && test -s conftest$ac_exeext && $as_test_x conftest$ac_exeext; then - have_bgp_files=yes + ok=PTHREAD_CREATE_JOINABLE else echo "$as_me: failed program was:" >&5 sed 's/^/| /' conftest.$ac_ext >&5 - { { echo "$as_me:$LINENO: error: There is a problem linking to the BG/P api." >&5 -echo "$as_me: error: There is a problem linking to the BG/P api." >&2;} - { (exit 1); exit 1; }; } + ok=unknown fi rm -f core conftest.err conftest.$ac_objext conftest_ipa8_conftest.oo \ conftest$ac_exeext conftest.$ac_ext - LDFLAGS="$saved_LDFLAGS" - fi - - if test ! -z "$have_bgp_files" ; then - BG_INCLUDES="$bg_includes" - -cat >>confdefs.h <<\_ACEOF -#define HAVE_3D 1 + if test x"$ok" = xunknown; then + cat >conftest.$ac_ext <<_ACEOF +/* confdefs.h. */ _ACEOF - - -cat >>confdefs.h <<\_ACEOF -#define HAVE_BG 1 +cat confdefs.h >>conftest.$ac_ext +cat >>conftest.$ac_ext <<_ACEOF +/* end confdefs.h. */ +#include +int +main () +{ +int attr=PTHREAD_CREATE_UNDETACHED; + ; + return 0; +} _ACEOF +rm -f conftest.$ac_objext conftest$ac_exeext +if { (ac_try="$ac_link" +case "(($ac_try" in + *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;; + *) ac_try_echo=$ac_try;; +esac +eval "echo \"\$as_me:$LINENO: $ac_try_echo\"") >&5 + (eval "$ac_link") 2>conftest.er1 + ac_status=$? + grep -v '^ *+' conftest.er1 >conftest.err + rm -f conftest.er1 + cat conftest.err >&5 + echo "$as_me:$LINENO: \$? = $ac_status" >&5 + (exit $ac_status); } && { + test -z "$ac_c_werror_flag" || + test ! -s conftest.err + } && test -s conftest$ac_exeext && + $as_test_x conftest$ac_exeext; then + ok=PTHREAD_CREATE_UNDETACHED +else + echo "$as_me: failed program was:" >&5 +sed 's/^/| /' conftest.$ac_ext >&5 + ok=unknown +fi -cat >>confdefs.h <<\_ACEOF -#define HAVE_BGP 1 -_ACEOF - +rm -f core conftest.err conftest.$ac_objext conftest_ipa8_conftest.oo \ + conftest$ac_exeext conftest.$ac_ext + fi + if test x"$ok" != xPTHREAD_CREATE_JOINABLE; then cat >>confdefs.h <<\_ACEOF -#define HAVE_FRONT_END 1 +#define PTHREAD_CREATE_JOINABLE $ok _ACEOF + fi + { echo "$as_me:$LINENO: result: ${ok}" >&5 +echo "${ECHO_T}${ok}" >&6; } + if test x"$ok" = xunknown; then + { echo "$as_me:$LINENO: WARNING: we do not know how to create joinable pthreads" >&5 +echo "$as_me: WARNING: we do not know how to create joinable pthreads" >&2;} + fi -cat >>confdefs.h <<\_ACEOF -#define HAVE_BG_FILES 1 -_ACEOF + { echo "$as_me:$LINENO: checking if more special flags are required for pthreads" >&5 +echo $ECHO_N "checking if more special flags are required for pthreads... $ECHO_C" >&6; } + flag=no + case "${host_cpu}-${host_os}" in + *-aix* | *-freebsd*) flag="-D_THREAD_SAFE";; + *solaris* | alpha*-osf*) flag="-D_REENTRANT";; + esac + { echo "$as_me:$LINENO: result: ${flag}" >&5 +echo "${ECHO_T}${flag}" >&6; } + if test "x$flag" != xno; then + PTHREAD_CFLAGS="$flag $PTHREAD_CFLAGS" + fi + LIBS="$save_LIBS" + CFLAGS="$save_CFLAGS" -cat >>confdefs.h <<_ACEOF -#define BG_BRIDGE_SO "$soloc" -_ACEOF + # More AIX lossage: must compile with cc_r + # Extract the first word of "cc_r", so it can be a program name with args. +set dummy cc_r; ac_word=$2 +{ echo "$as_me:$LINENO: checking for $ac_word" >&5 +echo $ECHO_N "checking for $ac_word... $ECHO_C" >&6; } +if test "${ac_cv_prog_PTHREAD_CC+set}" = set; then + echo $ECHO_N "(cached) $ECHO_C" >&6 +else + if test -n "$PTHREAD_CC"; then + ac_cv_prog_PTHREAD_CC="$PTHREAD_CC" # Let the user override the test. +else +as_save_IFS=$IFS; IFS=$PATH_SEPARATOR +for as_dir in $PATH +do + IFS=$as_save_IFS + test -z "$as_dir" && as_dir=. + for ac_exec_ext in '' $ac_executable_extensions; do + if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then + ac_cv_prog_PTHREAD_CC="cc_r" + echo "$as_me:$LINENO: found $as_dir/$ac_word$ac_exec_ext" >&5 + break 2 + fi +done +done +IFS=$as_save_IFS + test -z "$ac_cv_prog_PTHREAD_CC" && ac_cv_prog_PTHREAD_CC="${CC}" +fi +fi +PTHREAD_CC=$ac_cv_prog_PTHREAD_CC +if test -n "$PTHREAD_CC"; then + { echo "$as_me:$LINENO: result: $PTHREAD_CC" >&5 +echo "${ECHO_T}$PTHREAD_CC" >&6; } +else + { echo "$as_me:$LINENO: result: no" >&5 +echo "${ECHO_T}no" >&6; } +fi - { echo "$as_me:$LINENO: checking for BG serial value" >&5 -echo $ECHO_N "checking for BG serial value... $ECHO_C" >&6; } - bg_serial="BGP" -# Check whether --with-bg-serial was given. -if test "${with_bg_serial+set}" = set; then - withval=$with_bg_serial; bg_serial="$withval" +else + PTHREAD_CC="$CC" fi - { echo "$as_me:$LINENO: result: $bg_serial" >&5 -echo "${ECHO_T}$bg_serial" >&6; } -cat >>confdefs.h <<_ACEOF -#define BG_SERIAL "$bg_serial" -_ACEOF - #define ac_bluegene_loaded so we don't load another bluegene conf - ac_bluegene_loaded=yes - fi +# Finally, execute ACTION-IF-FOUND/ACTION-IF-NOT-FOUND: +if test x"$acx_pthread_ok" = xyes; then - if test "x$ac_bluegene_loaded" = "xyes"; then - BLUEGENE_LOADED_TRUE= - BLUEGENE_LOADED_FALSE='#' +cat >>confdefs.h <<\_ACEOF +#define HAVE_PTHREAD 1 +_ACEOF + + : else - BLUEGENE_LOADED_TRUE='#' - BLUEGENE_LOADED_FALSE= + acx_pthread_ok=no + { { echo "$as_me:$LINENO: error: Error: Cannot figure out how to use pthreads!" >&5 +echo "$as_me: error: Error: Cannot figure out how to use pthreads!" >&2;} + { (exit 1); exit 1; }; } fi +ac_ext=c +ac_cpp='$CPP $CPPFLAGS' +ac_compile='$CC -c $CFLAGS $CPPFLAGS conftest.$ac_ext >&5' +ac_link='$CC -o conftest$ac_exeext $CFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $LIBS >&5' +ac_compiler_gnu=$ac_cv_c_compiler_gnu + + + +# Always define WITH_PTHREADS if we make it this far +cat >>confdefs.h <<\_ACEOF +#define WITH_PTHREADS 1 +_ACEOF +LDFLAGS="$LDFLAGS " +CFLAGS="$CFLAGS $PTHREAD_CFLAGS" +LIBS="$PTHREAD_LIBS $LIBS" { echo "$as_me:$LINENO: checking for Sun Constellation system" >&5 @@ -25436,6 +25439,12 @@ ac_have_gtk="yes" _x_ac_pkcfg_bin="no" + # use the correct libs if running on 64bit + if test -d "/usr/lib64/pkgconfig"; then + PKG_CONFIG_PATH="/usr/lib64/pkgconfig/" + fi + + ### Check for pkg-config program # Check whether --with-pkg-config was given. @@ -25546,7 +25555,7 @@ # fi -### Check for gtk2.7.1 package +### Check for min gtk package if test "$ac_have_gtk" == "yes" ; then $HAVEPKGCONFIG --exists gtk+-2.0 if ! test $? -eq 0 ; then @@ -25554,18 +25563,14 @@ echo "$as_me: WARNING: *** gtk+-2.0 is not available." >&2;} ac_have_gtk="no" else - gtk_config_major_version=`$HAVEPKGCONFIG --modversion gtk+-2.0 | \ - sed 's/\([0-9]*\).\([0-9]*\).\([0-9]*\)/\1/'` - gtk_config_minor_version=`$HAVEPKGCONFIG --modversion gtk+-2.0 | \ - sed 's/\([0-9]*\).\([0-9]*\).\([0-9]*\)/\2/'` - gtk_config_micro_version=`$HAVEPKGCONFIG --modversion gtk+-2.0 | \ - sed 's/\([0-9]*\).\([0-9]*\).\([0-9]*\)/\3/'` - - if test $gtk_config_major_version -lt 2 || test $gtk_config_minor_version -lt 7 || test $gtk_config_micro_version -lt 1; then - { echo "$as_me:$LINENO: WARNING: *** gtk+-$gtk_config_major_version.$gtk_config_minor_version.$gtk_config_micro_version available, we need >= gtk+-2.7.1 installed for sview." >&5 -echo "$as_me: WARNING: *** gtk+-$gtk_config_major_version.$gtk_config_minor_version.$gtk_config_micro_version available, we need >= gtk+-2.7.1 installed for sview." >&2;} - ac_have_gtk="no" - fi + min_gtk_version="2.7.1" + $HAVEPKGCONFIG --atleast-version=$min_gtk_version gtk+-2.0 + if ! test $? -eq 0 ; then + gtk_config_version=`$HAVEPKGCONFIG --modversion gtk+-2.0` + { echo "$as_me:$LINENO: WARNING: *** gtk+-$gtk_config_version available, we need >= gtk+-$min_gtk_version installed for sview." >&5 +echo "$as_me: WARNING: *** gtk+-$gtk_config_version available, we need >= gtk+-$min_gtk_version installed for sview." >&2;} + ac_have_gtk="no" + fi fi fi @@ -27518,7 +27523,7 @@ -ac_config_files="$ac_config_files Makefile config.xml auxdir/Makefile contribs/Makefile contribs/perlapi/Makefile contribs/perlapi/libslurm-perl/Makefile.PL contribs/torque/Makefile contribs/phpext/Makefile contribs/phpext/slurm_php/config.m4 contribs/python/Makefile contribs/python/hostlist/Makefile contribs/python/hostlist/test/Makefile src/Makefile src/api/Makefile src/common/Makefile src/database/Makefile src/sacct/Makefile src/sacctmgr/Makefile src/sreport/Makefile src/sstat/Makefile src/salloc/Makefile src/sbatch/Makefile src/sattach/Makefile src/srun/Makefile src/slurmd/Makefile src/slurmd/slurmd/Makefile src/slurmd/slurmstepd/Makefile src/slurmdbd/Makefile src/slurmctld/Makefile src/sbcast/Makefile src/scontrol/Makefile src/scancel/Makefile src/squeue/Makefile src/sinfo/Makefile src/smap/Makefile src/strigger/Makefile src/sview/Makefile src/plugins/Makefile src/plugins/accounting_storage/Makefile src/plugins/accounting_storage/filetxt/Makefile src/plugins/accounting_storage/mysql/Makefile src/plugins/accounting_storage/pgsql/Makefile src/plugins/accounting_storage/none/Makefile src/plugins/accounting_storage/slurmdbd/Makefile src/plugins/auth/Makefile src/plugins/auth/authd/Makefile src/plugins/auth/munge/Makefile src/plugins/auth/none/Makefile src/plugins/checkpoint/Makefile src/plugins/checkpoint/aix/Makefile src/plugins/checkpoint/none/Makefile src/plugins/checkpoint/ompi/Makefile src/plugins/checkpoint/xlch/Makefile src/plugins/crypto/Makefile src/plugins/crypto/munge/Makefile src/plugins/crypto/openssl/Makefile src/plugins/jobacct_gather/Makefile src/plugins/jobacct_gather/linux/Makefile src/plugins/jobacct_gather/aix/Makefile src/plugins/jobacct_gather/none/Makefile src/plugins/jobcomp/Makefile src/plugins/jobcomp/filetxt/Makefile src/plugins/jobcomp/none/Makefile src/plugins/jobcomp/script/Makefile src/plugins/jobcomp/mysql/Makefile src/plugins/jobcomp/pgsql/Makefile src/plugins/proctrack/Makefile src/plugins/proctrack/aix/Makefile src/plugins/proctrack/pgid/Makefile src/plugins/proctrack/linuxproc/Makefile src/plugins/proctrack/rms/Makefile src/plugins/proctrack/sgi_job/Makefile src/plugins/sched/Makefile src/plugins/sched/backfill/Makefile src/plugins/sched/builtin/Makefile src/plugins/sched/gang/Makefile src/plugins/sched/hold/Makefile src/plugins/sched/wiki/Makefile src/plugins/sched/wiki2/Makefile src/plugins/select/Makefile src/plugins/select/bluegene/Makefile src/plugins/select/bluegene/block_allocator/Makefile src/plugins/select/bluegene/plugin/Makefile src/plugins/select/linear/Makefile src/plugins/select/cons_res/Makefile src/plugins/switch/Makefile src/plugins/switch/elan/Makefile src/plugins/switch/none/Makefile src/plugins/switch/federation/Makefile src/plugins/mpi/Makefile src/plugins/mpi/mpich1_p4/Makefile src/plugins/mpi/mpich1_shmem/Makefile src/plugins/mpi/mpichgm/Makefile src/plugins/mpi/mpichmx/Makefile src/plugins/mpi/mvapich/Makefile src/plugins/mpi/lam/Makefile src/plugins/mpi/none/Makefile src/plugins/mpi/openmpi/Makefile src/plugins/task/Makefile src/plugins/task/affinity/Makefile src/plugins/task/none/Makefile doc/Makefile doc/man/Makefile doc/html/Makefile doc/html/configurator.html testsuite/Makefile testsuite/expect/Makefile testsuite/slurm_unit/Makefile testsuite/slurm_unit/api/Makefile testsuite/slurm_unit/api/manual/Makefile testsuite/slurm_unit/common/Makefile testsuite/slurm_unit/slurmctld/Makefile testsuite/slurm_unit/slurmd/Makefile testsuite/slurm_unit/slurmdbd/Makefile" +ac_config_files="$ac_config_files Makefile config.xml auxdir/Makefile contribs/Makefile contribs/perlapi/Makefile contribs/perlapi/libslurm-perl/Makefile.PL contribs/torque/Makefile contribs/phpext/Makefile contribs/phpext/slurm_php/config.m4 contribs/python/Makefile contribs/python/hostlist/Makefile contribs/python/hostlist/test/Makefile contribs/slurmdb-direct/Makefile src/Makefile src/api/Makefile src/common/Makefile src/database/Makefile src/sacct/Makefile src/sacctmgr/Makefile src/sreport/Makefile src/sstat/Makefile src/salloc/Makefile src/sbatch/Makefile src/sattach/Makefile src/srun/Makefile src/slurmd/Makefile src/slurmd/slurmd/Makefile src/slurmd/slurmstepd/Makefile src/slurmdbd/Makefile src/slurmctld/Makefile src/sbcast/Makefile src/scontrol/Makefile src/scancel/Makefile src/squeue/Makefile src/sinfo/Makefile src/smap/Makefile src/strigger/Makefile src/sview/Makefile src/plugins/Makefile src/plugins/accounting_storage/Makefile src/plugins/accounting_storage/filetxt/Makefile src/plugins/accounting_storage/mysql/Makefile src/plugins/accounting_storage/pgsql/Makefile src/plugins/accounting_storage/none/Makefile src/plugins/accounting_storage/slurmdbd/Makefile src/plugins/auth/Makefile src/plugins/auth/authd/Makefile src/plugins/auth/munge/Makefile src/plugins/auth/none/Makefile src/plugins/checkpoint/Makefile src/plugins/checkpoint/aix/Makefile src/plugins/checkpoint/none/Makefile src/plugins/checkpoint/ompi/Makefile src/plugins/checkpoint/xlch/Makefile src/plugins/crypto/Makefile src/plugins/crypto/munge/Makefile src/plugins/crypto/openssl/Makefile src/plugins/jobacct_gather/Makefile src/plugins/jobacct_gather/linux/Makefile src/plugins/jobacct_gather/aix/Makefile src/plugins/jobacct_gather/none/Makefile src/plugins/jobcomp/Makefile src/plugins/jobcomp/filetxt/Makefile src/plugins/jobcomp/none/Makefile src/plugins/jobcomp/script/Makefile src/plugins/jobcomp/mysql/Makefile src/plugins/jobcomp/pgsql/Makefile src/plugins/proctrack/Makefile src/plugins/proctrack/aix/Makefile src/plugins/proctrack/pgid/Makefile src/plugins/proctrack/linuxproc/Makefile src/plugins/proctrack/rms/Makefile src/plugins/proctrack/sgi_job/Makefile src/plugins/sched/Makefile src/plugins/sched/backfill/Makefile src/plugins/sched/builtin/Makefile src/plugins/sched/gang/Makefile src/plugins/sched/hold/Makefile src/plugins/sched/wiki/Makefile src/plugins/sched/wiki2/Makefile src/plugins/select/Makefile src/plugins/select/bluegene/Makefile src/plugins/select/bluegene/block_allocator/Makefile src/plugins/select/bluegene/plugin/Makefile src/plugins/select/linear/Makefile src/plugins/select/cons_res/Makefile src/plugins/switch/Makefile src/plugins/switch/elan/Makefile src/plugins/switch/none/Makefile src/plugins/switch/federation/Makefile src/plugins/mpi/Makefile src/plugins/mpi/mpich1_p4/Makefile src/plugins/mpi/mpich1_shmem/Makefile src/plugins/mpi/mpichgm/Makefile src/plugins/mpi/mpichmx/Makefile src/plugins/mpi/mvapich/Makefile src/plugins/mpi/lam/Makefile src/plugins/mpi/none/Makefile src/plugins/mpi/openmpi/Makefile src/plugins/task/Makefile src/plugins/task/affinity/Makefile src/plugins/task/none/Makefile doc/Makefile doc/man/Makefile doc/html/Makefile doc/html/configurator.html testsuite/Makefile testsuite/expect/Makefile testsuite/slurm_unit/Makefile testsuite/slurm_unit/api/Makefile testsuite/slurm_unit/api/manual/Makefile testsuite/slurm_unit/common/Makefile testsuite/slurm_unit/slurmctld/Makefile testsuite/slurm_unit/slurmd/Makefile testsuite/slurm_unit/slurmdbd/Makefile" cat >confcache <<\_ACEOF @@ -27624,13 +27629,6 @@ Usually this means the macro was only invoked conditionally." >&2;} { (exit 1); exit 1; }; } fi -if test -z "${HAVE_AIX_TRUE}" && test -z "${HAVE_AIX_FALSE}"; then - { { echo "$as_me:$LINENO: error: conditional \"HAVE_AIX\" was never defined. -Usually this means the macro was only invoked conditionally." >&5 -echo "$as_me: error: conditional \"HAVE_AIX\" was never defined. -Usually this means the macro was only invoked conditionally." >&2;} - { (exit 1); exit 1; }; } -fi if test -z "${AMDEP_TRUE}" && test -z "${AMDEP_FALSE}"; then { { echo "$as_me:$LINENO: error: conditional \"AMDEP\" was never defined. Usually this means the macro was only invoked conditionally." >&5 @@ -27645,6 +27643,20 @@ Usually this means the macro was only invoked conditionally." >&2;} { (exit 1); exit 1; }; } fi +if test -z "${BLUEGENE_LOADED_TRUE}" && test -z "${BLUEGENE_LOADED_FALSE}"; then + { { echo "$as_me:$LINENO: error: conditional \"BLUEGENE_LOADED\" was never defined. +Usually this means the macro was only invoked conditionally." >&5 +echo "$as_me: error: conditional \"BLUEGENE_LOADED\" was never defined. +Usually this means the macro was only invoked conditionally." >&2;} + { (exit 1); exit 1; }; } +fi +if test -z "${HAVE_AIX_TRUE}" && test -z "${HAVE_AIX_FALSE}"; then + { { echo "$as_me:$LINENO: error: conditional \"HAVE_AIX\" was never defined. +Usually this means the macro was only invoked conditionally." >&5 +echo "$as_me: error: conditional \"HAVE_AIX\" was never defined. +Usually this means the macro was only invoked conditionally." >&2;} + { (exit 1); exit 1; }; } +fi if test -z "${HAVE_AIX_PROCTRACK_TRUE}" && test -z "${HAVE_AIX_PROCTRACK_FALSE}"; then { { echo "$as_me:$LINENO: error: conditional \"HAVE_AIX_PROCTRACK\" was never defined. Usually this means the macro was only invoked conditionally." >&5 @@ -27701,13 +27713,6 @@ Usually this means the macro was only invoked conditionally." >&2;} { (exit 1); exit 1; }; } fi -if test -z "${BLUEGENE_LOADED_TRUE}" && test -z "${BLUEGENE_LOADED_FALSE}"; then - { { echo "$as_me:$LINENO: error: conditional \"BLUEGENE_LOADED\" was never defined. -Usually this means the macro was only invoked conditionally." >&5 -echo "$as_me: error: conditional \"BLUEGENE_LOADED\" was never defined. -Usually this means the macro was only invoked conditionally." >&2;} - { (exit 1); exit 1; }; } -fi if test -z "${HAVE_SOME_CURSES_TRUE}" && test -z "${HAVE_SOME_CURSES_FALSE}"; then { { echo "$as_me:$LINENO: error: conditional \"HAVE_SOME_CURSES\" was never defined. Usually this means the macro was only invoked conditionally." >&5 @@ -28253,6 +28258,7 @@ "contribs/python/Makefile") CONFIG_FILES="$CONFIG_FILES contribs/python/Makefile" ;; "contribs/python/hostlist/Makefile") CONFIG_FILES="$CONFIG_FILES contribs/python/hostlist/Makefile" ;; "contribs/python/hostlist/test/Makefile") CONFIG_FILES="$CONFIG_FILES contribs/python/hostlist/test/Makefile" ;; + "contribs/slurmdb-direct/Makefile") CONFIG_FILES="$CONFIG_FILES contribs/slurmdb-direct/Makefile" ;; "src/Makefile") CONFIG_FILES="$CONFIG_FILES src/Makefile" ;; "src/api/Makefile") CONFIG_FILES="$CONFIG_FILES src/api/Makefile" ;; "src/common/Makefile") CONFIG_FILES="$CONFIG_FILES src/common/Makefile" ;; @@ -28502,18 +28508,18 @@ MAINTAINER_MODE_TRUE!$MAINTAINER_MODE_TRUE$ac_delim MAINTAINER_MODE_FALSE!$MAINTAINER_MODE_FALSE$ac_delim MAINT!$MAINT$ac_delim -CMD_LDFLAGS!$CMD_LDFLAGS$ac_delim -LIB_LDFLAGS!$LIB_LDFLAGS$ac_delim -SO_LDFLAGS!$SO_LDFLAGS$ac_delim -HAVE_AIX_TRUE!$HAVE_AIX_TRUE$ac_delim -HAVE_AIX_FALSE!$HAVE_AIX_FALSE$ac_delim -HAVE_AIX!$HAVE_AIX$ac_delim -PROCTRACKDIR!$PROCTRACKDIR$ac_delim CC!$CC$ac_delim CFLAGS!$CFLAGS$ac_delim LDFLAGS!$LDFLAGS$ac_delim CPPFLAGS!$CPPFLAGS$ac_delim ac_ct_CC!$ac_ct_CC$ac_delim +EXEEXT!$EXEEXT$ac_delim +OBJEXT!$OBJEXT$ac_delim +DEPDIR!$DEPDIR$ac_delim +am__include!$am__include$ac_delim +am__quote!$am__quote$ac_delim +AMDEP_TRUE!$AMDEP_TRUE$ac_delim +AMDEP_FALSE!$AMDEP_FALSE$ac_delim _ACEOF if test `sed -n "s/.*$ac_delim\$/X/p" conf$$subs.sed | grep -c X` = 97; then @@ -28555,17 +28561,21 @@ ac_delim='%!_!# ' for ac_last_try in false false false false false :; do cat >conf$$subs.sed <<_ACEOF -EXEEXT!$EXEEXT$ac_delim -OBJEXT!$OBJEXT$ac_delim -DEPDIR!$DEPDIR$ac_delim -am__include!$am__include$ac_delim -am__quote!$am__quote$ac_delim -AMDEP_TRUE!$AMDEP_TRUE$ac_delim -AMDEP_FALSE!$AMDEP_FALSE$ac_delim AMDEPBACKSLASH!$AMDEPBACKSLASH$ac_delim CCDEPMODE!$CCDEPMODE$ac_delim am__fastdepCC_TRUE!$am__fastdepCC_TRUE$ac_delim am__fastdepCC_FALSE!$am__fastdepCC_FALSE$ac_delim +BG_INCLUDES!$BG_INCLUDES$ac_delim +BLUEGENE_LOADED_TRUE!$BLUEGENE_LOADED_TRUE$ac_delim +BLUEGENE_LOADED_FALSE!$BLUEGENE_LOADED_FALSE$ac_delim +BLUEGENE_LOADED!$BLUEGENE_LOADED$ac_delim +CMD_LDFLAGS!$CMD_LDFLAGS$ac_delim +LIB_LDFLAGS!$LIB_LDFLAGS$ac_delim +SO_LDFLAGS!$SO_LDFLAGS$ac_delim +HAVE_AIX_TRUE!$HAVE_AIX_TRUE$ac_delim +HAVE_AIX_FALSE!$HAVE_AIX_FALSE$ac_delim +HAVE_AIX!$HAVE_AIX$ac_delim +PROCTRACKDIR!$PROCTRACKDIR$ac_delim CPP!$CPP$ac_delim GREP!$GREP$ac_delim EGREP!$EGREP$ac_delim @@ -28606,10 +28616,6 @@ PTHREAD_CC!$PTHREAD_CC$ac_delim PTHREAD_LIBS!$PTHREAD_LIBS$ac_delim PTHREAD_CFLAGS!$PTHREAD_CFLAGS$ac_delim -BG_INCLUDES!$BG_INCLUDES$ac_delim -BLUEGENE_LOADED_TRUE!$BLUEGENE_LOADED_TRUE$ac_delim -BLUEGENE_LOADED_FALSE!$BLUEGENE_LOADED_FALSE$ac_delim -BLUEGENE_LOADED!$BLUEGENE_LOADED$ac_delim SEMAPHORE_SOURCES!$SEMAPHORE_SOURCES$ac_delim SEMAPHORE_LIBS!$SEMAPHORE_LIBS$ac_delim NCURSES!$NCURSES$ac_delim diff -Nru slurm-llnl-1.3.13/configure.ac slurm-llnl-1.3.15/configure.ac --- slurm-llnl-1.3.13/configure.ac 2008-12-29 22:56:17.000000000 +0100 +++ slurm-llnl-1.3.15/configure.ac 2009-03-18 23:02:01.000000000 +0100 @@ -1,4 +1,4 @@ -# $Id: configure.ac 16088 2008-12-29 21:56:17Z jette $ +# $Id: configure.ac 16936 2009-03-18 22:02:01Z da $ # This file is to be processed with autoconf to generate a configure script dnl Prologue @@ -24,6 +24,12 @@ AC_CONFIG_HEADERS([config.h]) AC_CONFIG_HEADERS([slurm/slurm.h]) +dnl This needs to be close to the front to set CFLAGS=-m64 +X_AC_BGL +X_AC_BGP +AM_CONDITIONAL(BLUEGENE_LOADED, test "x$ac_bluegene_loaded" = "xyes") +AC_SUBST(BLUEGENE_LOADED) + X_AC_AIX dnl @@ -133,11 +139,6 @@ CFLAGS="$CFLAGS $PTHREAD_CFLAGS" LIBS="$PTHREAD_LIBS $LIBS" -X_AC_BGL -X_AC_BGP -AM_CONDITIONAL(BLUEGENE_LOADED, test "x$ac_bluegene_loaded" = "xyes") -AC_SUBST(BLUEGENE_LOADED) - X_AC_SUN_CONST X_AC_CFLAGS @@ -296,6 +297,7 @@ contribs/python/Makefile contribs/python/hostlist/Makefile contribs/python/hostlist/test/Makefile + contribs/slurmdb-direct/Makefile src/Makefile src/api/Makefile src/common/Makefile diff -Nru slurm-llnl-1.3.13/contribs/Makefile.am slurm-llnl-1.3.15/contribs/Makefile.am --- slurm-llnl-1.3.13/contribs/Makefile.am 2008-07-28 22:32:12.000000000 +0200 +++ slurm-llnl-1.3.15/contribs/Makefile.am 2009-03-18 23:02:01.000000000 +0100 @@ -1,4 +1,4 @@ -SUBDIRS = perlapi python torque +SUBDIRS = perlapi python torque slurmdb-direct EXTRA_DIST = \ env_cache_builder.c \ diff -Nru slurm-llnl-1.3.13/contribs/Makefile.in slurm-llnl-1.3.15/contribs/Makefile.in --- slurm-llnl-1.3.13/contribs/Makefile.in 2008-12-29 22:56:17.000000000 +0100 +++ slurm-llnl-1.3.15/contribs/Makefile.in 2009-03-18 23:02:01.000000000 +0100 @@ -248,7 +248,7 @@ target_vendor = @target_vendor@ top_builddir = @top_builddir@ top_srcdir = @top_srcdir@ -SUBDIRS = perlapi python torque +SUBDIRS = perlapi python torque slurmdb-direct EXTRA_DIST = \ env_cache_builder.c \ make.slurm.patch \ diff -Nru slurm-llnl-1.3.13/contribs/make.slurm.patch slurm-llnl-1.3.15/contribs/make.slurm.patch --- slurm-llnl-1.3.13/contribs/make.slurm.patch 2007-08-10 22:36:26.000000000 +0200 +++ slurm-llnl-1.3.15/contribs/make.slurm.patch 2009-01-20 17:43:05.000000000 +0100 @@ -30,7 +30,7 @@ child_execute_job (int stdin_fd, int stdout_fd, char **argv, char **envp) { +/* PARALLEL JOB LAUNCH VIA SLURM */ -+ if (getenv("SLURM_JOBID") { ++ if (getenv("SLURM_JOBID")) { + int i; + static char *argx[128]; + argx[0] = "srun"; diff -Nru slurm-llnl-1.3.13/contribs/perlapi/libslurm-perl/Slurm.xs slurm-llnl-1.3.15/contribs/perlapi/libslurm-perl/Slurm.xs --- slurm-llnl-1.3.13/contribs/perlapi/libslurm-perl/Slurm.xs 2008-04-23 19:13:05.000000000 +0200 +++ slurm-llnl-1.3.15/contribs/perlapi/libslurm-perl/Slurm.xs 2009-03-12 23:03:32.000000000 +0100 @@ -25,6 +25,7 @@ extern int slurm_hostlist_push_host(hostlist_t hl, const char *host); extern int slurm_hostlist_find(hostlist_t hl, const char *hostname); extern size_t slurm_hostlist_ranged_string(hostlist_t hl, size_t n, char *buf); +extern void slurm_hostlist_uniq(hostlist_t hl); struct slurm { node_info_msg_t *node_info_msg; @@ -805,6 +806,11 @@ RETVAL void +slurm_hostlist_uniq(hostlist_t hl = NULL) + CODE: + slurm_hostlist_uniq(hl); + +void DESTROY(hl) hostlist_t hl=NULL; CODE: diff -Nru slurm-llnl-1.3.13/contribs/python/hostlist/CHANGES slurm-llnl-1.3.15/contribs/python/hostlist/CHANGES --- slurm-llnl-1.3.13/contribs/python/hostlist/CHANGES 2008-09-30 18:28:41.000000000 +0200 +++ slurm-llnl-1.3.15/contribs/python/hostlist/CHANGES 2009-02-23 23:53:55.000000000 +0100 @@ -1,3 +1,22 @@ +Version 1.5 (2009-02-22) + + Make each "-" on the command line count as one hostlist argument. + If multiple hostslists are given on stdin they are combined to a + union hostlist before being used in the way requested by the + options. + + Add hostgrep utility to search for lines matching a hostlist. + + Make the build system (used when building tar.gz and RPMs from the + source code held in git) smarter. + +Version 1.4 (2008-12-28) + + Support Python 3. + + Import reduce from functools if possible. + Use Python 2/3 installation trick from .../Demo/distutils/test2to3 + Version 1.3 (2008-09-30) Add -s/--separator, -p/--prepend, -a/--append and --version diff -Nru slurm-llnl-1.3.13/contribs/python/hostlist/hostlist.py slurm-llnl-1.3.15/contribs/python/hostlist/hostlist.py --- slurm-llnl-1.3.13/contribs/python/hostlist/hostlist.py 2008-09-30 18:28:41.000000000 +0200 +++ slurm-llnl-1.3.15/contribs/python/hostlist/hostlist.py 2009-02-23 23:53:55.000000000 +0100 @@ -34,7 +34,7 @@ compatibility with pdsh/dshbak/SLURM et al. """ -__version__ = "1.3" +__version__ = "1.5" import re import itertools @@ -50,7 +50,7 @@ def expand_hostlist(hostlist, allow_duplicates=False, sort=False): """Expand a hostlist expression string to a Python list. - Exemple: expand_hostlist("n[9-11],d[01-02]") ==> + Example: expand_hostlist("n[9-11],d[01-02]") ==> ['n9', 'n10', 'n11', 'd01', 'd02'] Unless allow_duplicates is true, duplicates will be purged diff -Nru slurm-llnl-1.3.13/contribs/python/hostlist/PKG-INFO slurm-llnl-1.3.15/contribs/python/hostlist/PKG-INFO --- slurm-llnl-1.3.13/contribs/python/hostlist/PKG-INFO 2008-09-30 18:28:41.000000000 +0200 +++ slurm-llnl-1.3.15/contribs/python/hostlist/PKG-INFO 2009-02-23 23:53:55.000000000 +0100 @@ -1,6 +1,6 @@ Metadata-Version: 1.0 Name: python-hostlist -Version: 1.3 +Version: 1.5 Summary: Python module for hostlist handling Home-page: http://www.nsc.liu.se/~kent/python-hostlist/ Author: Kent Engström @@ -14,3 +14,5 @@ Classifier: License :: OSI Approved :: GNU General Public License (GPL) Classifier: Topic :: System :: Clustering Classifier: Topic :: System :: Systems Administration +Classifier: Programming Language :: Python :: 2 +Classifier: Programming Language :: Python :: 3 diff -Nru slurm-llnl-1.3.13/contribs/python/hostlist/python-hostlist.spec slurm-llnl-1.3.15/contribs/python/hostlist/python-hostlist.spec --- slurm-llnl-1.3.13/contribs/python/hostlist/python-hostlist.spec 2008-09-30 18:28:41.000000000 +0200 +++ slurm-llnl-1.3.15/contribs/python/hostlist/python-hostlist.spec 2009-02-23 23:53:55.000000000 +0100 @@ -1,7 +1,7 @@ %{!?python_sitelib: %define python_sitelib %(%{__python} -c "from distutils.sysconfig import get_python_lib; print get_python_lib()")} Name: python-hostlist -Version: 1.3 +Version: 1.5 Release: 1 Summary: Python module for hostlist handling Vendor: NSC @@ -45,5 +45,7 @@ %doc CHANGES %{python_sitelib}/* /usr/bin/hostlist +/usr/bin/hostgrep /usr/share/man/man1/hostlist.1.gz +/usr/share/man/man1/hostgrep.1.gz %changelog diff -Nru slurm-llnl-1.3.13/contribs/python/hostlist/README slurm-llnl-1.3.15/contribs/python/hostlist/README --- slurm-llnl-1.3.13/contribs/python/hostlist/README 2008-09-30 18:28:41.000000000 +0200 +++ slurm-llnl-1.3.15/contribs/python/hostlist/README 2009-02-23 23:53:55.000000000 +0100 @@ -1,3 +1,6 @@ +INTRODUCTION +============ + The Python module hostlist.py knows how to expand and collect hostlist expressions. Example: @@ -32,22 +35,27 @@ n9 n10 -Install directly by running - python setup.py build (as yourself) - python setup.py install (as root) +BUILDING +======== -or just copy the hostlist.py, hostlist and hostlist.1 files to appropriate places. +Build RPM packages from the tar.gz archive by running: -Build RPM packages by running: + rpmbuild -ta python-hostlist-1.5.tar.gz - rpmbuild -ta python-hostlist-VERSION.tar.gz - -If you do not have the tar archive, build RPM packages by running: +If you do not have the tar archive, create it first: python setup.py sdist - cp dist/python-hostlist-VERSION.tar.gz ~/rpmbuild/SOURCES - rpmbuild -ba python-hostlist.spec + rpmbuild -ta dist/python-hostlist-1.5.tar.gz + +You may also install directly by running: + + python setup.py build (as yourself) + python setup.py install (as root) + + +RELEASES AND FEEDBACK +===================== You will find new releases at: diff -Nru slurm-llnl-1.3.13/contribs/python/hostlist/setup.py slurm-llnl-1.3.15/contribs/python/hostlist/setup.py --- slurm-llnl-1.3.13/contribs/python/hostlist/setup.py 2008-09-30 18:28:41.000000000 +0200 +++ slurm-llnl-1.3.15/contribs/python/hostlist/setup.py 2009-02-23 23:53:55.000000000 +0100 @@ -2,8 +2,27 @@ from distutils.core import setup +# Python 2/3 installation trick from .../Demo/distutils/test2to3 +try: + from distutils.command.build_py import build_py_2to3 as build_py +except ImportError: + from distutils.command.build_py import build_py + +try: + from distutils.command.build_scripts import build_scripts_2to3 as build_scripts +except ImportError: + from distutils.command.build_scripts import build_scripts + +# Version +VERSION = "1.5" +if "#" in VERSION: + import sys + sys.stderr.write("Bad version %s\n" % VERSION) + sys.exit(1) + + setup(name = "python-hostlist", - version = "1.3", # Change in hostlist{,.py,.1}, python-hostlist.spec too! + version = VERSION, description = "Python module for hostlist handling", long_description = "The hostlist.py module knows how to expand and collect hostlist expressions.", author = "Kent Engström", @@ -16,8 +35,14 @@ 'License :: OSI Approved :: GNU General Public License (GPL)', 'Topic :: System :: Clustering', 'Topic :: System :: Systems Administration', + 'Programming Language :: Python :: 2', + 'Programming Language :: Python :: 3', ], py_modules = ["hostlist"], - scripts = ["hostlist"], - data_files = [("share/man/man1", ["hostlist.1"])], + scripts = ["hostlist", "hostgrep"], + data_files = [("share/man/man1", ["hostlist.1", + "hostgrep.1"])], + cmdclass = {'build_py':build_py, + 'build_scripts':build_scripts, + } ) diff -Nru slurm-llnl-1.3.13/contribs/slurmdb-direct/config.slurmdb.pl slurm-llnl-1.3.15/contribs/slurmdb-direct/config.slurmdb.pl --- slurm-llnl-1.3.13/contribs/slurmdb-direct/config.slurmdb.pl 1970-01-01 01:00:00.000000000 +0100 +++ slurm-llnl-1.3.15/contribs/slurmdb-direct/config.slurmdb.pl 2009-03-12 21:15:20.000000000 +0100 @@ -0,0 +1,15 @@ +# Set database information +$db_name = "slurm_acct_db"; +$db_job_table = "job_table"; + +# These are the options you should change for your own site. +#$db_host = "fargo"; +#$db_port = "3306"; +#$db_user = "some_user"; +#$db_passwd = "some_password"; + +# Database connection line for the DBI +$db_conn_line = "DBI:mysql:database=${db_name};host=${db_host}"; + +# "1;" Required for file inclusion +1; diff -Nru slurm-llnl-1.3.13/contribs/slurmdb-direct/Makefile.am slurm-llnl-1.3.15/contribs/slurmdb-direct/Makefile.am --- slurm-llnl-1.3.13/contribs/slurmdb-direct/Makefile.am 1970-01-01 01:00:00.000000000 +0100 +++ slurm-llnl-1.3.15/contribs/slurmdb-direct/Makefile.am 2009-03-24 17:33:55.000000000 +0100 @@ -0,0 +1,39 @@ +# +# Makefile for slurmdb-direct scripts + +AUTOMAKE_OPTIONS = foreign + +sbin_SCRIPTS = moab_2_slurmdb +extra = config.slurmdb.pl + +moab_2_slurmdb: + +_perldir=$(exec_prefix)`perl -e 'use Config; $$T=$$Config{installsitearch}; $$P=$$Config{installprefix}; $$P1="$$P/local"; $$T =~ s/$$P1//; $$T =~ s/$$P//; print $$T;'` + +install-sbinSCRIPTS: $(sbin_SCRIPTS) + @$(NORMAL_INSTALL) + test -z "$(DESTDIR)$(sbindir)" || $(MKDIR_P) "$(DESTDIR)$(sbindir)" + @list='$(sbin_SCRIPTS)'; for p in $$list; do \ + echo "sed 's%use lib .*%use lib qw(${_perldir});%' $(top_srcdir)/contribs/slurmdb-direct/$$p.pl > $(DESTDIR)$(sbindir)/$$p"; \ + sed "s%use lib .*%use lib qw(${_perldir});%" $(top_srcdir)/contribs/slurmdb-direct/$$p.pl >$(DESTDIR)$(sbindir)/$$p; \ + chmod 755 $(DESTDIR)$(sbindir)/$$p;\ + done + + test -z "$(DESTDIR)${_perldir}" || $(MKDIR_P) "$(DESTDIR)${_perldir}" + @list='$(extra)'; for p in $$list; do \ + echo " cp $(top_srcdir)/contribs/slurmdb-direct/$$p $(DESTDIR)${_perldir}/$$p"; \ + cp $(top_srcdir)/contribs/slurmdb-direct/$$p $(DESTDIR)${_perldir}/$$p; \ + done + +uninstall-sbinSCRIPTS: + @$(NORMAL_UNINSTALL) + @list='$(sbin_SCRIPTS)'; for p in $$list; do \ + echo " rm -f '$(DESTDIR)$(sbindir)/$$p'"; \ + rm -f "$(DESTDIR)$(sbindir)/$$p"; \ + done + @list='$(extra)'; for p in $$list; do \ + echo " rm -f '$(DESTDIR)$(_perldir)/$$p'"; \ + rm -f "$(DESTDIR)$(_perldir)/$$p"; \ + done + +clean: diff -Nru slurm-llnl-1.3.13/contribs/slurmdb-direct/Makefile.in slurm-llnl-1.3.15/contribs/slurmdb-direct/Makefile.in --- slurm-llnl-1.3.13/contribs/slurmdb-direct/Makefile.in 1970-01-01 01:00:00.000000000 +0100 +++ slurm-llnl-1.3.15/contribs/slurmdb-direct/Makefile.in 2009-03-24 17:33:55.000000000 +0100 @@ -0,0 +1,450 @@ +# Makefile.in generated by automake 1.10.1 from Makefile.am. +# @configure_input@ + +# Copyright (C) 1994, 1995, 1996, 1997, 1998, 1999, 2000, 2001, 2002, +# 2003, 2004, 2005, 2006, 2007, 2008 Free Software Foundation, Inc. +# This Makefile.in is free software; the Free Software Foundation +# gives unlimited permission to copy and/or distribute it, +# with or without modifications, as long as this notice is preserved. + +# This program is distributed in the hope that it will be useful, +# but WITHOUT ANY WARRANTY, to the extent permitted by law; without +# even the implied warranty of MERCHANTABILITY or FITNESS FOR A +# PARTICULAR PURPOSE. + +@SET_MAKE@ + +# +# Makefile for slurmdb-direct scripts + +VPATH = @srcdir@ +pkgdatadir = $(datadir)/@PACKAGE@ +pkglibdir = $(libdir)/@PACKAGE@ +pkgincludedir = $(includedir)/@PACKAGE@ +am__cd = CDPATH="$${ZSH_VERSION+.}$(PATH_SEPARATOR)" && cd +install_sh_DATA = $(install_sh) -c -m 644 +install_sh_PROGRAM = $(install_sh) -c +install_sh_SCRIPT = $(install_sh) -c +INSTALL_HEADER = $(INSTALL_DATA) +transform = $(program_transform_name) +NORMAL_INSTALL = : +PRE_INSTALL = : +POST_INSTALL = : +NORMAL_UNINSTALL = : +PRE_UNINSTALL = : +POST_UNINSTALL = : +build_triplet = @build@ +host_triplet = @host@ +target_triplet = @target@ +subdir = contribs/slurmdb-direct +DIST_COMMON = $(srcdir)/Makefile.am $(srcdir)/Makefile.in +ACLOCAL_M4 = $(top_srcdir)/aclocal.m4 +am__aclocal_m4_deps = $(top_srcdir)/auxdir/acx_pthread.m4 \ + $(top_srcdir)/auxdir/slurm.m4 \ + $(top_srcdir)/auxdir/x_ac__system_configuration.m4 \ + $(top_srcdir)/auxdir/x_ac_affinity.m4 \ + $(top_srcdir)/auxdir/x_ac_aix.m4 \ + $(top_srcdir)/auxdir/x_ac_bluegene.m4 \ + $(top_srcdir)/auxdir/x_ac_cflags.m4 \ + $(top_srcdir)/auxdir/x_ac_databases.m4 \ + $(top_srcdir)/auxdir/x_ac_debug.m4 \ + $(top_srcdir)/auxdir/x_ac_elan.m4 \ + $(top_srcdir)/auxdir/x_ac_federation.m4 \ + $(top_srcdir)/auxdir/x_ac_gpl_licensed.m4 \ + $(top_srcdir)/auxdir/x_ac_gtk.m4 \ + $(top_srcdir)/auxdir/x_ac_munge.m4 \ + $(top_srcdir)/auxdir/x_ac_ncurses.m4 \ + $(top_srcdir)/auxdir/x_ac_pam.m4 \ + $(top_srcdir)/auxdir/x_ac_ptrace.m4 \ + $(top_srcdir)/auxdir/x_ac_readline.m4 \ + $(top_srcdir)/auxdir/x_ac_setpgrp.m4 \ + $(top_srcdir)/auxdir/x_ac_setproctitle.m4 \ + $(top_srcdir)/auxdir/x_ac_sgi_job.m4 \ + $(top_srcdir)/auxdir/x_ac_slurm_ssl.m4 \ + $(top_srcdir)/auxdir/x_ac_sun_const.m4 \ + $(top_srcdir)/auxdir/x_ac_xcpu.m4 $(top_srcdir)/configure.ac +am__configure_deps = $(am__aclocal_m4_deps) $(CONFIGURE_DEPENDENCIES) \ + $(ACLOCAL_M4) +mkinstalldirs = $(install_sh) -d +CONFIG_HEADER = $(top_builddir)/config.h $(top_builddir)/slurm/slurm.h +CONFIG_CLEAN_FILES = +am__installdirs = "$(DESTDIR)$(sbindir)" +sbinSCRIPT_INSTALL = $(INSTALL_SCRIPT) +SCRIPTS = $(sbin_SCRIPTS) +SOURCES = +DIST_SOURCES = +DISTFILES = $(DIST_COMMON) $(DIST_SOURCES) $(TEXINFOS) $(EXTRA_DIST) +ACLOCAL = @ACLOCAL@ +AMTAR = @AMTAR@ +AR = @AR@ +AUTHD_CFLAGS = @AUTHD_CFLAGS@ +AUTHD_LIBS = @AUTHD_LIBS@ +AUTOCONF = @AUTOCONF@ +AUTOHEADER = @AUTOHEADER@ +AUTOMAKE = @AUTOMAKE@ +AWK = @AWK@ +BG_INCLUDES = @BG_INCLUDES@ +BLUEGENE_LOADED = @BLUEGENE_LOADED@ +CC = @CC@ +CCDEPMODE = @CCDEPMODE@ +CFLAGS = @CFLAGS@ +CMD_LDFLAGS = @CMD_LDFLAGS@ +CPP = @CPP@ +CPPFLAGS = @CPPFLAGS@ +CXX = @CXX@ +CXXCPP = @CXXCPP@ +CXXDEPMODE = @CXXDEPMODE@ +CXXFLAGS = @CXXFLAGS@ +CYGPATH_W = @CYGPATH_W@ +DEFS = @DEFS@ +DEPDIR = @DEPDIR@ +DSYMUTIL = @DSYMUTIL@ +ECHO = @ECHO@ +ECHO_C = @ECHO_C@ +ECHO_N = @ECHO_N@ +ECHO_T = @ECHO_T@ +EGREP = @EGREP@ +ELAN_LIBS = @ELAN_LIBS@ +EXEEXT = @EXEEXT@ +F77 = @F77@ +FEDERATION_LDFLAGS = @FEDERATION_LDFLAGS@ +FFLAGS = @FFLAGS@ +GREP = @GREP@ +GTK2_CFLAGS = @GTK2_CFLAGS@ +GTK2_LIBS = @GTK2_LIBS@ +HAVEMYSQLCONFIG = @HAVEMYSQLCONFIG@ +HAVEPGCONFIG = @HAVEPGCONFIG@ +HAVEPKGCONFIG = @HAVEPKGCONFIG@ +HAVE_AIX = @HAVE_AIX@ +HAVE_ELAN = @HAVE_ELAN@ +HAVE_FEDERATION = @HAVE_FEDERATION@ +HAVE_OPENSSL = @HAVE_OPENSSL@ +HAVE_SOME_CURSES = @HAVE_SOME_CURSES@ +INSTALL = @INSTALL@ +INSTALL_DATA = @INSTALL_DATA@ +INSTALL_PROGRAM = @INSTALL_PROGRAM@ +INSTALL_SCRIPT = @INSTALL_SCRIPT@ +INSTALL_STRIP_PROGRAM = @INSTALL_STRIP_PROGRAM@ +LDFLAGS = @LDFLAGS@ +LIBOBJS = @LIBOBJS@ +LIBS = @LIBS@ +LIBTOOL = @LIBTOOL@ +LIB_LDFLAGS = @LIB_LDFLAGS@ +LN_S = @LN_S@ +LTLIBOBJS = @LTLIBOBJS@ +MAINT = @MAINT@ +MAKEINFO = @MAKEINFO@ +MKDIR_P = @MKDIR_P@ +MUNGE_CPPFLAGS = @MUNGE_CPPFLAGS@ +MUNGE_LDFLAGS = @MUNGE_LDFLAGS@ +MUNGE_LIBS = @MUNGE_LIBS@ +MYSQL_CFLAGS = @MYSQL_CFLAGS@ +MYSQL_LIBS = @MYSQL_LIBS@ +NCURSES = @NCURSES@ +NMEDIT = @NMEDIT@ +NUMA_LIBS = @NUMA_LIBS@ +OBJEXT = @OBJEXT@ +PACKAGE = @PACKAGE@ +PACKAGE_BUGREPORT = @PACKAGE_BUGREPORT@ +PACKAGE_NAME = @PACKAGE_NAME@ +PACKAGE_STRING = @PACKAGE_STRING@ +PACKAGE_TARNAME = @PACKAGE_TARNAME@ +PACKAGE_VERSION = @PACKAGE_VERSION@ +PAM_LIBS = @PAM_LIBS@ +PATH_SEPARATOR = @PATH_SEPARATOR@ +PGSQL_CFLAGS = @PGSQL_CFLAGS@ +PGSQL_LIBS = @PGSQL_LIBS@ +PLPA_LIBS = @PLPA_LIBS@ +PROCTRACKDIR = @PROCTRACKDIR@ +PROJECT = @PROJECT@ +PTHREAD_CC = @PTHREAD_CC@ +PTHREAD_CFLAGS = @PTHREAD_CFLAGS@ +PTHREAD_LIBS = @PTHREAD_LIBS@ +RANLIB = @RANLIB@ +READLINE_LIBS = @READLINE_LIBS@ +RELEASE = @RELEASE@ +SED = @SED@ +SEMAPHORE_LIBS = @SEMAPHORE_LIBS@ +SEMAPHORE_SOURCES = @SEMAPHORE_SOURCES@ +SET_MAKE = @SET_MAKE@ +SHELL = @SHELL@ +SLURMCTLD_PORT = @SLURMCTLD_PORT@ +SLURMDBD_PORT = @SLURMDBD_PORT@ +SLURMD_PORT = @SLURMD_PORT@ +SLURM_API_AGE = @SLURM_API_AGE@ +SLURM_API_CURRENT = @SLURM_API_CURRENT@ +SLURM_API_MAJOR = @SLURM_API_MAJOR@ +SLURM_API_REVISION = @SLURM_API_REVISION@ +SLURM_API_VERSION = @SLURM_API_VERSION@ +SLURM_MAJOR = @SLURM_MAJOR@ +SLURM_MICRO = @SLURM_MICRO@ +SLURM_MINOR = @SLURM_MINOR@ +SLURM_PREFIX = @SLURM_PREFIX@ +SLURM_VERSION = @SLURM_VERSION@ +SO_LDFLAGS = @SO_LDFLAGS@ +SSL_CPPFLAGS = @SSL_CPPFLAGS@ +SSL_LDFLAGS = @SSL_LDFLAGS@ +SSL_LIBS = @SSL_LIBS@ +STRIP = @STRIP@ +UTIL_LIBS = @UTIL_LIBS@ +VERSION = @VERSION@ +abs_builddir = @abs_builddir@ +abs_srcdir = @abs_srcdir@ +abs_top_builddir = @abs_top_builddir@ +abs_top_srcdir = @abs_top_srcdir@ +ac_ct_CC = @ac_ct_CC@ +ac_ct_CXX = @ac_ct_CXX@ +ac_ct_F77 = @ac_ct_F77@ +am__include = @am__include@ +am__leading_dot = @am__leading_dot@ +am__quote = @am__quote@ +am__tar = @am__tar@ +am__untar = @am__untar@ +bindir = @bindir@ +build = @build@ +build_alias = @build_alias@ +build_cpu = @build_cpu@ +build_os = @build_os@ +build_vendor = @build_vendor@ +builddir = @builddir@ +datadir = @datadir@ +datarootdir = @datarootdir@ +docdir = @docdir@ +dvidir = @dvidir@ +exec_prefix = @exec_prefix@ +host = @host@ +host_alias = @host_alias@ +host_cpu = @host_cpu@ +host_os = @host_os@ +host_vendor = @host_vendor@ +htmldir = @htmldir@ +includedir = @includedir@ +infodir = @infodir@ +install_sh = @install_sh@ +libdir = @libdir@ +libexecdir = @libexecdir@ +localedir = @localedir@ +localstatedir = @localstatedir@ +mandir = @mandir@ +mkdir_p = @mkdir_p@ +oldincludedir = @oldincludedir@ +pdfdir = @pdfdir@ +prefix = @prefix@ +program_transform_name = @program_transform_name@ +psdir = @psdir@ +sbindir = @sbindir@ +sharedstatedir = @sharedstatedir@ +srcdir = @srcdir@ +sysconfdir = @sysconfdir@ +target = @target@ +target_alias = @target_alias@ +target_cpu = @target_cpu@ +target_os = @target_os@ +target_vendor = @target_vendor@ +top_builddir = @top_builddir@ +top_srcdir = @top_srcdir@ +AUTOMAKE_OPTIONS = foreign +sbin_SCRIPTS = moab_2_slurmdb +extra = config.slurmdb.pl +_perldir = $(exec_prefix)`perl -e 'use Config; $$T=$$Config{installsitearch}; $$P=$$Config{installprefix}; $$P1="$$P/local"; $$T =~ s/$$P1//; $$T =~ s/$$P//; print $$T;'` +all: all-am + +.SUFFIXES: +$(srcdir)/Makefile.in: @MAINTAINER_MODE_TRUE@ $(srcdir)/Makefile.am $(am__configure_deps) + @for dep in $?; do \ + case '$(am__configure_deps)' in \ + *$$dep*) \ + cd $(top_builddir) && $(MAKE) $(AM_MAKEFLAGS) am--refresh \ + && exit 0; \ + exit 1;; \ + esac; \ + done; \ + echo ' cd $(top_srcdir) && $(AUTOMAKE) --foreign contribs/slurmdb-direct/Makefile'; \ + cd $(top_srcdir) && \ + $(AUTOMAKE) --foreign contribs/slurmdb-direct/Makefile +.PRECIOUS: Makefile +Makefile: $(srcdir)/Makefile.in $(top_builddir)/config.status + @case '$?' in \ + *config.status*) \ + cd $(top_builddir) && $(MAKE) $(AM_MAKEFLAGS) am--refresh;; \ + *) \ + echo ' cd $(top_builddir) && $(SHELL) ./config.status $(subdir)/$@ $(am__depfiles_maybe)'; \ + cd $(top_builddir) && $(SHELL) ./config.status $(subdir)/$@ $(am__depfiles_maybe);; \ + esac; + +$(top_builddir)/config.status: $(top_srcdir)/configure $(CONFIG_STATUS_DEPENDENCIES) + cd $(top_builddir) && $(MAKE) $(AM_MAKEFLAGS) am--refresh + +$(top_srcdir)/configure: @MAINTAINER_MODE_TRUE@ $(am__configure_deps) + cd $(top_builddir) && $(MAKE) $(AM_MAKEFLAGS) am--refresh +$(ACLOCAL_M4): @MAINTAINER_MODE_TRUE@ $(am__aclocal_m4_deps) + cd $(top_builddir) && $(MAKE) $(AM_MAKEFLAGS) am--refresh + +mostlyclean-libtool: + -rm -f *.lo + +clean-libtool: + -rm -rf .libs _libs +tags: TAGS +TAGS: + +ctags: CTAGS +CTAGS: + + +distdir: $(DISTFILES) + @srcdirstrip=`echo "$(srcdir)" | sed 's/[].[^$$\\*]/\\\\&/g'`; \ + topsrcdirstrip=`echo "$(top_srcdir)" | sed 's/[].[^$$\\*]/\\\\&/g'`; \ + list='$(DISTFILES)'; \ + dist_files=`for file in $$list; do echo $$file; done | \ + sed -e "s|^$$srcdirstrip/||;t" \ + -e "s|^$$topsrcdirstrip/|$(top_builddir)/|;t"`; \ + case $$dist_files in \ + */*) $(MKDIR_P) `echo "$$dist_files" | \ + sed '/\//!d;s|^|$(distdir)/|;s,/[^/]*$$,,' | \ + sort -u` ;; \ + esac; \ + for file in $$dist_files; do \ + if test -f $$file || test -d $$file; then d=.; else d=$(srcdir); fi; \ + if test -d $$d/$$file; then \ + dir=`echo "/$$file" | sed -e 's,/[^/]*$$,,'`; \ + if test -d $(srcdir)/$$file && test $$d != $(srcdir); then \ + cp -pR $(srcdir)/$$file $(distdir)$$dir || exit 1; \ + fi; \ + cp -pR $$d/$$file $(distdir)$$dir || exit 1; \ + else \ + test -f $(distdir)/$$file \ + || cp -p $$d/$$file $(distdir)/$$file \ + || exit 1; \ + fi; \ + done +check-am: all-am +check: check-am +all-am: Makefile $(SCRIPTS) +installdirs: + for dir in "$(DESTDIR)$(sbindir)"; do \ + test -z "$$dir" || $(MKDIR_P) "$$dir"; \ + done +install: install-am +install-exec: install-exec-am +install-data: install-data-am +uninstall: uninstall-am + +install-am: all-am + @$(MAKE) $(AM_MAKEFLAGS) install-exec-am install-data-am + +installcheck: installcheck-am +install-strip: + $(MAKE) $(AM_MAKEFLAGS) INSTALL_PROGRAM="$(INSTALL_STRIP_PROGRAM)" \ + install_sh_PROGRAM="$(INSTALL_STRIP_PROGRAM)" INSTALL_STRIP_FLAG=-s \ + `test -z '$(STRIP)' || \ + echo "INSTALL_PROGRAM_ENV=STRIPPROG='$(STRIP)'"` install +mostlyclean-generic: + +clean-generic: + +distclean-generic: + -test -z "$(CONFIG_CLEAN_FILES)" || rm -f $(CONFIG_CLEAN_FILES) + +maintainer-clean-generic: + @echo "This command is intended for maintainers to use" + @echo "it deletes files that may require special tools to rebuild." +clean-am: clean-generic clean-libtool mostlyclean-am + +distclean: distclean-am + -rm -f Makefile +distclean-am: clean-am distclean-generic + +dvi: dvi-am + +dvi-am: + +html: html-am + +info: info-am + +info-am: + +install-data-am: + +install-dvi: install-dvi-am + +install-exec-am: install-sbinSCRIPTS + +install-html: install-html-am + +install-info: install-info-am + +install-man: + +install-pdf: install-pdf-am + +install-ps: install-ps-am + +installcheck-am: + +maintainer-clean: maintainer-clean-am + -rm -f Makefile +maintainer-clean-am: distclean-am maintainer-clean-generic + +mostlyclean: mostlyclean-am + +mostlyclean-am: mostlyclean-generic mostlyclean-libtool + +pdf: pdf-am + +pdf-am: + +ps: ps-am + +ps-am: + +uninstall-am: uninstall-sbinSCRIPTS + +.MAKE: install-am install-strip + +.PHONY: all all-am check check-am clean clean-generic clean-libtool \ + distclean distclean-generic distclean-libtool distdir dvi \ + dvi-am html html-am info info-am install install-am \ + install-data install-data-am install-dvi install-dvi-am \ + install-exec install-exec-am install-html install-html-am \ + install-info install-info-am install-man install-pdf \ + install-pdf-am install-ps install-ps-am install-sbinSCRIPTS \ + install-strip installcheck installcheck-am installdirs \ + maintainer-clean maintainer-clean-generic mostlyclean \ + mostlyclean-generic mostlyclean-libtool pdf pdf-am ps ps-am \ + uninstall uninstall-am uninstall-sbinSCRIPTS + + +moab_2_slurmdb: + +install-sbinSCRIPTS: $(sbin_SCRIPTS) + @$(NORMAL_INSTALL) + test -z "$(DESTDIR)$(sbindir)" || $(MKDIR_P) "$(DESTDIR)$(sbindir)" + @list='$(sbin_SCRIPTS)'; for p in $$list; do \ + echo "sed 's%use lib .*%use lib qw(${_perldir});%' $(top_srcdir)/contribs/slurmdb-direct/$$p.pl > $(DESTDIR)$(sbindir)/$$p"; \ + sed "s%use lib .*%use lib qw(${_perldir});%" $(top_srcdir)/contribs/slurmdb-direct/$$p.pl >$(DESTDIR)$(sbindir)/$$p; \ + chmod 755 $(DESTDIR)$(sbindir)/$$p;\ + done + + test -z "$(DESTDIR)${_perldir}" || $(MKDIR_P) "$(DESTDIR)${_perldir}" + @list='$(extra)'; for p in $$list; do \ + echo " cp $(top_srcdir)/contribs/slurmdb-direct/$$p $(DESTDIR)${_perldir}/$$p"; \ + cp $(top_srcdir)/contribs/slurmdb-direct/$$p $(DESTDIR)${_perldir}/$$p; \ + done + +uninstall-sbinSCRIPTS: + @$(NORMAL_UNINSTALL) + @list='$(sbin_SCRIPTS)'; for p in $$list; do \ + echo " rm -f '$(DESTDIR)$(sbindir)/$$p'"; \ + rm -f "$(DESTDIR)$(sbindir)/$$p"; \ + done + @list='$(extra)'; for p in $$list; do \ + echo " rm -f '$(DESTDIR)$(_perldir)/$$p'"; \ + rm -f "$(DESTDIR)$(_perldir)/$$p"; \ + done + +clean: +# Tell versions [3.59,3.63) of GNU make to not export all variables. +# Otherwise a system limit (for SysV at least) may be exceeded. +.NOEXPORT: diff -Nru slurm-llnl-1.3.13/contribs/slurmdb-direct/moab_2_slurmdb.pl slurm-llnl-1.3.15/contribs/slurmdb-direct/moab_2_slurmdb.pl --- slurm-llnl-1.3.13/contribs/slurmdb-direct/moab_2_slurmdb.pl 1970-01-01 01:00:00.000000000 +0100 +++ slurm-llnl-1.3.15/contribs/slurmdb-direct/moab_2_slurmdb.pl 2009-03-13 22:04:16.000000000 +0100 @@ -0,0 +1,275 @@ +#! /usr/bin/perl -w +############################################################################### +# +# slurmdbd_direct - write directly into the slurmdbd. +# +# based off the index in +# http://www.clusterresources.com/products/mwm/docs/16.3.3workloadtrace.shtml +# +# +############################################################################### +# Copyright (C) 2009 Lawrence Livermore National Security. +# Produced at Lawrence Livermore National Laboratory (cf, DISCLAIMER). +# Written by Danny Auble +# CODE-OCEC-09-009. All rights reserved. +# +# This file is part of SLURM, a resource management program. +# For details, see . +# +# SLURM is free software; you can redistribute it and/or modify it under +# the terms of the GNU General Public License as published by the Free +# Software Foundation; either version 2 of the License, or (at your option) +# any later version. +# +# In addition, as a special exception, the copyright holders give permission +# to link the code of portions of this program with the OpenSSL library under +# certain conditions as described in each individual source file, and +# distribute linked combinations including the two. You must obey the GNU +# General Public License in all respects for all of the code used other than +# OpenSSL. If you modify file(s) with this exception, you may extend this +# exception to your version of the file(s), but you are not obligated to do +# so. If you do not wish to do so, delete this exception statement from your +# version. If you delete this exception statement from all source files in +# the program, then also delete it here. +# +# SLURM is distributed in the hope that it will be useful, but WITHOUT ANY +# WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS +# FOR A PARTICULAR PURPOSE. See the GNU General Public License for more +# details. +# +# You should have received a copy of the GNU General Public License along +# with SLURM; if not, write to the Free Software Foundation, Inc., +# 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA. +# +############################################################################### + +use strict; +use FindBin; +use Getopt::Long 2.24 qw(:config no_ignore_case require_order); +#use lib "${FindBin::Bin}/../lib/perl"; +use lib qw(/home/da/slurm/1.3/snowflake/lib/perl/5.8.8); +use autouse 'Pod::Usage' => qw(pod2usage); +use Slurm ':all'; +use Switch; +use DBI; +BEGIN { require "config.slurmdb.pl"; } +our ($logLevel, $db_conn_line, $db_job_table, $db_user, $db_passwd); + +my $set = 0; +my $submit_set = 0; +my $migrate_set = 0; +my $start_set = 0; +my $end_set = 0; + +my $submit_sql = "INSERT INTO $db_job_table " . + "(jobid, associd, wckeyid, track_steps, priority, uid, gid, cluster, " . + "account, partition, wckey, name, state, req_cpus, submit) VALUES "; + +my $migrate_sql = "INSERT INTO $db_job_table " . + "(jobid, associd, wckeyid, track_steps, priority, uid, gid, cluster, " . + "account, partition, wckey, name, state, req_cpus, " . + "submit, eligible) VALUES "; + +my $start_sql = "INSERT INTO $db_job_table " . + "(jobid, associd, wckeyid, track_steps, priority, uid, gid, cluster, " . + "account, partition, wckey, name, state, req_cpus, " . + "submit, eligible, start, nodelist, alloc_cpus) VALUES "; + +my $end_sql = "INSERT INTO $db_job_table " . + "(jobid, associd, wckeyid, track_steps, priority, uid, gid, cluster, " . + "account, partition, wckey, name, state, req_cpus, " . + "submit, eligible, start, nodelist, alloc_cpus, " . + "end, comp_code) VALUES "; + +foreach my $line () { + chomp $line; + # the below list is based off the index in +# http://www.clusterresources.com/products/mwm/docs/16.3.3workloadtrace.shtml + my ($hr_time, + $timestamp, + $type, + $id, + $event, + $req_nodes, + $req_tasks, + $user, + $group, + $wall_limit, + $state, + $partition, + $eligible_time, + $dispatch_time, + $start_time, + $end_time, + $network, + $arch, + $op, + $node_mem_comp, + $node_mem, + $node_disk_comp, + $node_disk, + $node_features, + $submit_time, + $alloc_tasks, + $tasks_per_node, + $qos, + $flags, + $account, + $executable, + $rm_ext, + $bypass_cnt, + $cpu_secs, + $cluster, + $procs_per_task, + $mem_per_task, + $disk_per_task, + $swap_per_task, + $other_time, + $timeout, + $alloc_hostlist, + $rm_name, + $req_hostlist, + $resv, + $app_sim_data, + $desc, + $message, + $cost, + $history, + $util, + $estimate, + $comp_code, + $ext_mem, + $ext_cpu, + @extra) = split /\s+/, $line; + next if !$type; + next if $type ne "job"; + + my $uid = getpwnam($user); + my $gid = getgrnam($group); + $uid = -2 if !$uid; + $gid = -2 if !$gid; + + # figure out the wckey + my $wckey = ""; + if ($rm_ext =~ /wckey:(\w*)/) { + $wckey = $1; + } + + if($partition =~ /\[(\w*)/) { + $partition = $1; + } + + #figure out the cluster + if($cluster eq "ALL") { + if ($node_features =~ /\[(\w*)\]/) { + $cluster = $1; + } elsif ($rm_ext =~ /partition:(\w*)/) { + $cluster = $1; + } elsif ($rm_ext =~ /feature:(\w*)/) { + $cluster = $1; + } else { + $cluster = ""; + } + } + + if($message =~ /job\\20exceeded\\20wallclock\\20limit/) { + $event = "JOBTIMEOUT"; + } + + my $alloc_hl = Slurm::Hostlist::create($alloc_hostlist); + if($alloc_hl) { + Slurm::Hostlist::uniq($alloc_hl); + $alloc_hl = Slurm::Hostlist::ranged_string($alloc_hl); + } + + if($event eq "JOBSUBMIT") { + $submit_sql .= ", " if $submit_set; + $submit_sql .= "($id, 0, 0, 0, 0, $uid, $gid, \"$cluster\", " . + "\"$account\", \"$partition\", \"$wckey\", " . + "\"$executable\", 0, $req_tasks, $submit_time)"; + $submit_set = 1; + $set = 1; + } elsif ($event eq "JOBMIGRATE") { + $migrate_sql .= ", " if $migrate_set; + # here for some reason the eligible time is really the + # elgible time, so we use the end time which appears + # to be the best guess. + $migrate_sql .= "($id, 0, 0, 0, 0, $uid, $gid, \"$cluster\", " . + "\"$account\", \"$partition\", \"$wckey\", " . + "\"$executable\", 0, $req_tasks, $submit_time, " . + "$end_time)"; + $migrate_set = 1; + $set = 1; + } elsif ($event eq "JOBSTART") { + $start_sql .= ", " if $start_set; + + # req_tasks is used for alloc_tasks on purpose. + # alloc_tasks isn't always correct. + $start_sql .= "($id, 0, 0, 0, 0, $uid, $gid, \"$cluster\", " . + "\"$account\", \"$partition\", \"$wckey\", " . + "\"$executable\", 1, $req_tasks, $submit_time, " . + "$eligible_time, $start_time, \"$alloc_hl\", " . + "$req_tasks)"; + $start_set = 1; + $set = 1; + } elsif (($event eq "JOBEND") || ($event eq "JOBCANCEL") + || ($event eq "JOBFAILURE") || ($event eq "JOBTIMEOUT")) { + if($event eq "JOBEND") { + $state = 3; + } elsif($event eq "JOBCANCEL") { + $state = 4; + } elsif($event eq "JOBFAILURE") { + $state = 5; + } else { + $state = 6; + } + + $end_sql .= ", " if $end_set; + $end_sql .= "($id, 0, 0, 0, 0, $uid, $gid, \"$cluster\", " . + "\"$account\", \"$partition\", \"$wckey\", " . + "\"$executable\", $state, $req_tasks, $submit_time, " . + "$eligible_time, $start_time, \"$alloc_hl\", " . + "$req_tasks, $end_time, $comp_code)"; + $end_set = 1; + $set = 1; + } else { + print "ERROR: unknown event of $event\n"; + next; + } +} + +exit 0 if !$set; + +$db_user = (getpwuid($<))[0] if !$db_user; +my $dbhandle = DBI->connect($db_conn_line, $db_user, $db_passwd, + {AutoCommit => 1, RaiseError => 1}); +if($submit_set) { + $submit_sql .= " on duplicate key update jobid=VALUES(jobid)"; + #print "submit\n$submit_sql\n\n"; + $dbhandle->do($submit_sql); +} + +if($migrate_set) { + $migrate_sql .= " on duplicate key update eligible=VALUES(eligible)"; + #print "migrate\n$migrate_sql\n\n"; + $dbhandle->do($migrate_sql); +} + +if($start_set) { + $start_sql .= " on duplicate key update nodelist=VALUES(nodelist), " . + "account=VALUES(account), partition=VALUES(partition), " . + "wckey=values(wckey), start=VALUES(start), " . + "name=VALUES(name), state=values(state), " . + "alloc_cpus=values(alloc_cpus)"; + #print "start\n$start_sql\n\n"; + $dbhandle->do($start_sql); +} + +if($end_set) { + $end_sql .= " on duplicate key update end=VALUES(end), " . + "state=VALUES(state), comp_code=VALUES(comp_code)"; + #print "end\n$end_sql\n\n"; + $dbhandle->do($end_sql); +} + +exit 0; diff -Nru slurm-llnl-1.3.13/contribs/torque/Makefile.am slurm-llnl-1.3.15/contribs/torque/Makefile.am --- slurm-llnl-1.3.13/contribs/torque/Makefile.am 2008-04-15 02:57:42.000000000 +0200 +++ slurm-llnl-1.3.15/contribs/torque/Makefile.am 2009-03-18 23:02:01.000000000 +0100 @@ -17,7 +17,7 @@ install-binSCRIPTS: $(bin_SCRIPTS) @$(NORMAL_INSTALL) - test -z "$(bindir)" || $(MKDIR_P) "$(DESTDIR)$(bindir)" + test -z "$(DESTDIR)$(bindir)" || $(MKDIR_P) "$(DESTDIR)$(bindir)" @list='$(bin_SCRIPTS)'; for p in $$list; do \ echo "sed 's%use lib .*%use lib qw(${_perldir});%' $(top_srcdir)/contribs/torque/$$p.pl > $(DESTDIR)$(bindir)/$$p"; \ sed "s%use lib .*%use lib qw(${_perldir});%" $(top_srcdir)/contribs/torque/$$p.pl >$(DESTDIR)$(bindir)/$$p; \ diff -Nru slurm-llnl-1.3.13/contribs/torque/Makefile.in slurm-llnl-1.3.15/contribs/torque/Makefile.in --- slurm-llnl-1.3.13/contribs/torque/Makefile.in 2008-12-29 22:56:17.000000000 +0100 +++ slurm-llnl-1.3.15/contribs/torque/Makefile.in 2009-03-18 23:02:01.000000000 +0100 @@ -425,7 +425,7 @@ install-binSCRIPTS: $(bin_SCRIPTS) @$(NORMAL_INSTALL) - test -z "$(bindir)" || $(MKDIR_P) "$(DESTDIR)$(bindir)" + test -z "$(DESTDIR)$(bindir)" || $(MKDIR_P) "$(DESTDIR)$(bindir)" @list='$(bin_SCRIPTS)'; for p in $$list; do \ echo "sed 's%use lib .*%use lib qw(${_perldir});%' $(top_srcdir)/contribs/torque/$$p.pl > $(DESTDIR)$(bindir)/$$p"; \ sed "s%use lib .*%use lib qw(${_perldir});%" $(top_srcdir)/contribs/torque/$$p.pl >$(DESTDIR)$(bindir)/$$p; \ diff -Nru slurm-llnl-1.3.13/debian/changelog slurm-llnl-1.3.15/debian/changelog --- slurm-llnl-1.3.13/debian/changelog 2009-04-22 11:23:58.000000000 +0200 +++ slurm-llnl-1.3.15/debian/changelog 2009-04-22 11:23:58.000000000 +0200 @@ -1,3 +1,13 @@ +slurm-llnl (1.3.15-1) unstable; urgency=high + + * New upstream release to fix supplementary group flaw + * New slurm-llnl-basic-plugins{,-dev} descriptions (Closes: #512059) + * New libpmi-dev description + * Compat version upgraded to 7 + * Standard version upgraded to 3.8.1 + + -- Gennaro Oliva Fri, 17 Apr 2009 02:26:25 +0200 + slurm-llnl (1.3.13-1) unstable; urgency=low * New upstream release (Closes: #511511) diff -Nru slurm-llnl-1.3.13/debian/compat slurm-llnl-1.3.15/debian/compat --- slurm-llnl-1.3.13/debian/compat 2009-04-22 11:23:58.000000000 +0200 +++ slurm-llnl-1.3.15/debian/compat 2009-04-22 11:23:58.000000000 +0200 @@ -1 +1 @@ -4 +7 diff -Nru slurm-llnl-1.3.13/debian/control slurm-llnl-1.3.15/debian/control --- slurm-llnl-1.3.13/debian/control 2009-04-22 11:23:58.000000000 +0200 +++ slurm-llnl-1.3.15/debian/control 2009-04-22 11:23:58.000000000 +0200 @@ -3,8 +3,8 @@ Priority: extra Maintainer: Gennaro Oliva Uploaders: Dirk Eddelbuettel -Build-Depends: debhelper (>= 4.0.0), autotools-dev, libmunge-dev, libncurses5-dev, libssl-dev, po-debconf, python, libglade2-dev, libgtk2.0-dev, libmysqlclient15-dev, postgresql-server-dev-8.3 -Standards-Version: 3.8.0 +Build-Depends: debhelper (>= 7.0.0), autotools-dev, libmunge-dev, libncurses5-dev, libssl-dev, po-debconf, python, libglade2-dev, libgtk2.0-dev, libmysqlclient15-dev, postgresql-server-dev-8.3 +Standards-Version: 3.8.1 Homepage: https://computing.llnl.gov/linux/slurm/ Package: slurm-llnl @@ -47,7 +47,7 @@ Section: libdevel Architecture: any Depends: ${shlibs:Depends}, ${misc:Depends}, libpmi0 (= ${binary:Version}), libslurm13 (= ${binary:Version}) -Description: SLURM PMI library implementation +Description: SLURM PMI library implementation development files SLURM, the Simple Linux Utility for Resource Management, is an open-source cluster resource management and job scheduling. This package contains the SLURM PMI implementation static library and header @@ -63,10 +63,9 @@ This package contains the SLURM html documentation. Package: slurm-llnl-basic-plugins -Section: admin Architecture: any Depends: ${shlibs:Depends}, ${misc:Depends} -Description: SLURM docmentation +Description: SLURM basic plugins SLURM, the Simple Linux Utility for Resource Management, is an open-source cluster resource management and job scheduling. This package contains the SLURM basic plugins @@ -75,7 +74,7 @@ Section: devel Architecture: any Depends: ${shlibs:Depends}, ${misc:Depends}, slurm-llnl-basic-plugins (= ${binary:Version}) -Description: SLURM docmentation +Description: SLURM basic plugins development files SLURM, the Simple Linux Utility for Resource Management, is an open-source cluster resource management and job scheduling. This package contains development files for the SLURM basic plugins diff -Nru slurm-llnl-1.3.13/debian/rules slurm-llnl-1.3.15/debian/rules --- slurm-llnl-1.3.13/debian/rules 2009-04-22 11:23:58.000000000 +0200 +++ slurm-llnl-1.3.15/debian/rules 2009-04-22 11:23:58.000000000 +0200 @@ -74,7 +74,7 @@ install-indep: dh_testdir dh_testroot - dh_clean -k -i + dh_prep -i dh_installdirs -i $(MAKE) -C doc/html install DESTDIR=$(CURDIR)/debian/slurm-llnl-doc dh_install -pslurm-llnl-doc debian/slurm-llnl-doc/usr/share/doc/slurm-[0-9]*/html usr/share/doc/slurm-llnl-doc @@ -86,7 +86,7 @@ install-arch: dh_testdir dh_testroot - dh_clean -k -s + dh_prep -s dh_installdirs -s # Add here commands to install the arch part of the package into diff -Nru slurm-llnl-1.3.13/debian/slurm-llnl.postinst slurm-llnl-1.3.15/debian/slurm-llnl.postinst --- slurm-llnl-1.3.13/debian/slurm-llnl.postinst 2009-04-22 11:23:58.000000000 +0200 +++ slurm-llnl-1.3.15/debian/slurm-llnl.postinst 2009-04-22 11:23:58.000000000 +0200 @@ -18,8 +18,6 @@ # case "$1" in configure) - chown slurm:slurm /var/run/slurm-llnl - chown slurm:slurm /var/run/slurm-llnl/slurmctld chown slurm:slurm /var/log/slurm-llnl ;; diff -Nru slurm-llnl-1.3.13/debian/slurm-llnl-slurmdbd.postinst slurm-llnl-1.3.15/debian/slurm-llnl-slurmdbd.postinst --- slurm-llnl-1.3.13/debian/slurm-llnl-slurmdbd.postinst 2009-04-22 11:23:58.000000000 +0200 +++ slurm-llnl-1.3.15/debian/slurm-llnl-slurmdbd.postinst 2009-04-22 11:23:58.000000000 +0200 @@ -18,7 +18,6 @@ # case "$1" in configure) - chown slurm:slurm /var/run/slurm-llnl chown slurm:slurm /var/log/slurm-llnl ;; diff -Nru slurm-llnl-1.3.13/doc/html/accounting.shtml slurm-llnl-1.3.15/doc/html/accounting.shtml --- slurm-llnl-1.3.13/doc/html/accounting.shtml 2008-10-07 23:15:47.000000000 +0200 +++ slurm-llnl-1.3.15/doc/html/accounting.shtml 2009-03-02 23:26:35.000000000 +0100 @@ -8,13 +8,14 @@

SLURM can be configured to collect accounting information for every job and job step executed. -Accounting records can be written to a simple file or a database. +Accounting records can be written to a simple text file or a database. Information is available about both currently executing jobs and -jobs which have already terminated and can be viewed using the -sacct command. -sacct can also report resource usage for individual tasks, -which can be useful to detect load imbalance between the tasks. -The sstat tool can be used to status a currently running job. +jobs which have already terminated. +The sacct command can report resource usage for running or terminated +jobs including individual tasks, which can be useful to detect load imbalance +between the tasks. +The sstat command can be used to status only currently running jobs. +It also can give you valuable information about imbalance between tasks. The sreport can be used to generate reports based upon all jobs executed in a particular time interval.

@@ -22,6 +23,15 @@ The SLURM configuration parameters (in slurm.conf) associated with these plugins include:

    +
  • AccountingStorageType controls how detailed job and job +step information is recorded. You can store this information in a +text file, MySQL or +PostgreSQL +database, optionally using SlurmDBD for added security.
    • +
  • JobAcctGatherType is operating system dependent and +controls what mechanism is used to collect accounting information. +Supported values are jobacct_gather/aix, jobacct_gather/linux +and jobacct_gather/none (no information collected).
  • JobCompType controls how job completion information is recorded. This can be used to record basic job information such as job name, user name, allocated nodes, start time, completion @@ -31,25 +41,18 @@ text file, MySQL or PostgreSQL database
    • -
  • JobAcctGatherType is operating system dependent and -controls what mechanisms are used to collect accounting information. -Supported values are jobacct_gather/aix, jobacct_gather/linux -and jobacct_gather/none (no information collected).
    • -
  • AccountingStorageType controls how detailed job and job -step information is recorded. You can store this information in a -text file, MySQL or -PostgreSQL -database optionally using either -Gold -or SlurmDBD for added security.
-

The use of sacct or sstat to view information about completed jobs -is dependent upon both JobAcctGatherType and AccountingStorageType +

The use of sacct to view information about jobs +is dependent upon AccountingStorageType being configured to collect and store that information. The use of sreport is dependent upon some database being used to store that information.

+

The use of sacct or sstat to view information about resource usage + within jobs is dependent upon both JobAcctGatherType and AccountingStorageType + being configured to collect and store that information.

+

Storing the accounting information into text files is very simple. Just configure the appropriate plugin (e.g. AccountingStorageType=accounting_storage/filetxt and/or @@ -71,14 +74,11 @@ Making possibly sensitive information available to all users makes database security more difficult to provide, sending the data through an intermediate daemon can provide better security and performance -(through caching data). -Gold and SlurmDBD are two such services. -Our initial implementation relied upon Gold, but we found its -performance to be inadequate for our needs and developed SlurmDBD. +(through caching data) and SlurmDBD provides such services. SlurmDBD (SLURM Database Daemon) is written in C, multi-threaded, -secure, and considerably faster than Gold. +secure and fast. The configuration required to use SlurmDBD will be described below. -Direct database or Gold use would be similar.

+Storing information directly into database would be similar.

Note that SlurmDBD relies upon existing SLURM plugins for authentication and database use, but the other SLURM @@ -128,9 +128,9 @@ and SlurmDBD configuration files (slurm.conf and slurmdbd.conf respectively, more details are provided below).

-Whether you use any authentication module or not you will need to have +Slurm JobComp Configuration

Presently job completion is not supported with the SlurmDBD, but can be -written directly to a database, script or flat file.If you are +written directly to a database, script or flat file. If you are running with the accounting storage, you may not need to run this -since it contains much of the same information.If you would like +since it contains much of the same information. If you would like to configure this, some of the more important parameters include:

    @@ -155,7 +155,8 @@
  • JobCompPass: Only needed if using a database. Password for the user connecting to -the database.
    • +the database. Since the password can not be security maintained, +storing the information directly in a database is not recommended.
  • JobCompPort: Only needed if using a database. The network port that the database @@ -177,23 +178,21 @@ "associations" between a user and account. A database allows such a configuration. -

    -MySQL is the preferred database, PostgreSQL is +

    MySQL is the preferred database, PostgreSQL is supported for job and step accounting only. The infrastructure for PostgresSQL for use with associations is not yet supported, meaning sacctmgr will not work correcting. If interested in adding this -capabilty for PostgresSQL please email slurm-dev@lists.llnl.gov. +capability for PostgresSQL, please contact us at slurm-dev@lists.llnl.gov. -

    -To enable this database support +

    To enable this database support one only needs to have the development package for the database they wish to use on the system. The slurm configure script uses mysql_config and pg-config to find out the information it needs about installed libraries and headers. You can specify where your mysql_config script is with the --with-mysql_conf=/path/to/mysql_config option when configuring your -slurm build. A similar option is available for PostgreSQL also. On -a successful configure, output is something like this:

    +slurm build. A similar option is also available for PostgreSQL. +On a successful configure, output is something like this: 

     checking for mysql_config... /usr/bin/mysql_config
 MySQL test program built properly.
@@ -203,7 +202,7 @@
 
 
    For simplicity sake we are going to reference everything as if you
 are running with the SlurmDBD. You can communicate with a storage plugin
-directly, but that offers minimal authentication. 
+directly, but that offers minimal security. 
    
 
 
    Several SLURM configuration parameters must be set to support
 archiving information in SlurmDBD. SlurmDBD has a separate configuration
@@ -212,21 +211,36 @@
 while job completion records are written to a text file or 
 not maintained at all. 
 If you don't set the configuration parameters that begin 
-with "JobComp" then job completion records will not be recorded.
    
+with "AccountingStorage" then accounting information will not be
+referenced or recorded.
    
 
 
      
 
    • AccountingStorageEnforce:
-If you want to prevent users from running jobs if their association
-is not in the database, then set this to "1". 
+This option contains a comma separated list of options you may want to
+ enforce.  The valid options are 
+
        
+
      • associations - This will prevent users from running jobs if
+their association is not in the database. This option will
+prevent users from accessing invalid accounts.  
+
        • 
+
      • limits - This will enforce limits set to associations.  By setting
+  this option, the 'associations' option is also set.
+
        • 
+
      • wckeys - This will prevent users from running jobs under a wckey
+  that they don't have access to.  By using this option, the
+  'associations' option is also set.  The 'TrackWCKey' option is also
+  set to true.
+
        • 
+
      
 (NOTE: The association is a combination of cluster, account, 
 user names and optional partition name.)
+
      
 Without AccountingStorageEnforce being set (the default behavior) 
-jobs will be executed based upon policies configured in SLURM on each cluster. 
-This option will prevent users from accessing invalid accounts.  
-Setting this to "2" will also cause association limits to be
-enforced.  When set to "1" association limits will not be
-enforced.  It is a good idea to run in this mode when running a
-scheduler on top of slurm, like Moab, that does not update in real
+jobs will be executed based upon policies configured in SLURM on each
+cluster.
+
       
+It is advisable to run without the option 'limits' set when running a
+scheduler on top of SLURM, like Moab, that does not update in real
 time their limits per association.
      • 
 
 
    • AccountingStorageHost: The name or address of the host where 
@@ -245,6 +259,13 @@
 
    • ClusterName:
 Set to a unique name for each Slurm-managed cluster so that 
 accounting records from each can be identified.
      • 
+
    • TrackWCKey:
+Boolean.  If you want to track wckeys (Workload Characterization Key)
+  of users.  A Wckey is an orthogonal way to do accounting against
+  maybe a group of unrelated accounts. WCKeys can be defined using
+  sacctmgr add wckey 'name'.  When a job is run use srun --wckey and 
+  time will be summed up for this wckey.
+
      • 
 
    
 
 
    SlurmDBD Configuration
    
@@ -322,12 +343,7 @@
 
  • StorageType:
 Define the accounting storage mechanism type.
 Acceptable values at present include 
-"accounting_storage/gold", "accounting_storage/mysql", and
-"accounting_storage/pgsql".
-The value "accounting_storage/gold" indicates that account records
-will be written to Gold, which maintains its own database.
-Use of Gold is not recommended due to reduced performance without 
-providing any additional security.
+"accounting_storage/mysql" and "accounting_storage/pgsql".
 The value "accounting_storage/mysql" indicates that accounting records
 should be written to a MySQL database specified by the 
 StorageLoc parameter.
@@ -391,7 +407,7 @@
 
    See the man pages for each command for more information.
    
 
 
    Web interfaces with graphical output is currently under
-development and should be available in the Fall of 2008.
+development and should be available in the Fall of 2009.
 A tool to report node state information is also under development.
    
 
 
    Database Configuration
    
@@ -446,8 +462,8 @@
 sacctmgr add user da default=test
    • -

    If AccountingStorageEnforce=1 is configured in the slurm.conf of -the cluster snowflake then user da would be +

    If AccountingStorageEnforce=associations is configured in +the slurm.conf of the cluster snowflake then user da would be allowed to run in account test and any other accounts added in the future. Any attempt to use other accounts will result in the job being @@ -459,7 +475,6 @@ Partition='partitionname' option to specify an association specific to a slurm partition.

    -

    Account Options

    When either adding or modifying an account, the following sacctmgr options are available:

      -
    • Description= Description of the account. (Required on creation)
      • - -
    • Organization=Organization of the account. (Required on creation)
      • - -
    • Name= Name of account
      • -
    • Cluster= Only add this account to these clusters. The account is added to all defined clusters by default.
      • -
    • Parent= Make this account a child of this other account.
      • - -

    User Options

    @@ -517,10 +512,7 @@ options are available:
      -
    • Name= User name
      • - -
    • DefaultAccount= Default account for the user, used when no account -is specified when a job is submitted. (Required on creation)
      • +
    • Account= Account(s) to add user to
    • AdminLevel= This field is used to allow a user to add accounting privileges to this user. Valid options are @@ -531,37 +523,117 @@ and remove accounts and clusters
    -
  • Account= Account(s) to add user to
    • -
  • Cluster= Only add to accounts on these clusters (default is all clusters)
    • -
  • Partition= Name of Slurm partition this association applies to
    • - -
- +

All of the above entities can include limits as described below... + +

    + +
  • Fairshare= Used for determining priority. Essentially + this is the amount of claim this association and it's children have + to the above system.
    • + + + + + + +
  • GrpJobs= The total number of jobs able to run at any given + time from this association and its children. If + this limit is reached new jobs will be queued but only allowed to + run after previous jobs complete from this group. +
    • + +
  • GrpNodes= The total count of nodes able to be used at any given + time from jobs running from this association and its children. If + this limit is reached new jobs will be queued but only allowed to + run after resources have been relinquished from this group. +
    • + +
  • GrpSubmitJobs= The total number of jobs able to be submitted + to the system at any given time from this association and its children. If + this limit is reached new submission requests will be denied until + previous jobs complete from this group. +
    • + +
  • GrpWall= The maximum wall clock time any job submitted to + this group can run for. If this limit is reached submission requests + will be denied. +
    • + + + + + +
  • MaxJobs= The total number of jobs able to run at any given + time from this association. If this limit is reached new jobs will + be queued but only allowed to run after previous jobs complete from + this association. +
    • + +
  • MaxNodesPerJob= The maximum size in nodes any given job can + have from this association. If this limit is reached the job will + be denied at submission. +
    • + +
  • MaxSubmitJobs= The maximum number of jobs able to be submitted + to the system at any given time from this association. If + this limit is reached new submission requests will be denied until + previous jobs complete from this association. +
    • + +
  • MaxWallDurationPerJob= The maximum wall clock time any job + submitted to this association can run for. If this limit is reached + the job will be denied at submission. +
    • + +
  • QOS= comma separated list of QOS's this association is + able to run. + +
    • +

Modifying Entities

@@ -595,7 +667,7 @@ If an entity has existed for less than 1 day, the entity will be removed completely. This is meant to clean up after typographic errors.

-

Last modified 27 June 2008

+

Last modified 2 March 2009

diff -Nru slurm-llnl-1.3.13/doc/html/accounting_storageplugins.shtml slurm-llnl-1.3.15/doc/html/accounting_storageplugins.shtml --- slurm-llnl-1.3.13/doc/html/accounting_storageplugins.shtml 1970-01-01 01:00:00.000000000 +0100 +++ slurm-llnl-1.3.15/doc/html/accounting_storageplugins.shtml 2009-02-11 23:52:55.000000000 +0100 @@ -0,0 +1,885 @@ + + +

SLURM Accounting Storage Plugin API

+ +

Overview

+

This document describes SLURM Accounting Storage plugins and the API that +defines them. It is intended as a resource to programmers wishing to write +their own SLURM Job Accounting Storage plugins. This is version 1 of the API. + +

SLURM Accounting Storage plugins must conform to the +SLURM Plugin API with the following specifications: + +

const char +plugin_name[]="full text name" +

+A free-formatted ASCII text string that identifies the plugin. + +

const char +plugin_type[]="major/minor"
+

+The major type must be "accounting_storage." +The minor type can be any suitable name +for the type of accounting package. We currently use +

    +
  • filetxt—Information written to a text file. +
  • mysql— Store information in a mysql database. +
  • pgsql— Store information in a postgresql database. +
  • none— Information is not stored anywhere. +
+

The programmer is urged to study +src/plugins/accounting_storage/mysql +for a sample implementation of a SLURM Accounting Storage plugin. +

The Accounting Storage plugin was written to be a interface +to storage data collected by the Job Accounting Gather plugin. When +adding a new database you may want to add common functions in a common +file in the src/database dir. Refer to src/database/mysql_common.c|.h for an +example so other plugins can also use that database type to write out +information. +

top + + +

API Functions

+ +The Job Accounting Storage API uses hooks in the slurmctld. + +

All of the following functions are required. Functions which are not +implemented must be stubbed. + +

Functions called by the accounting_storage plugin

+ +

void *acct_storage_p_get_connection(bool + make_agent, int conn_num, bool rollback, char *location) +

Description:
+acct_storage_p_get_connection() is called to get a connection to the + storage medium. acct_storage_p_close_connection() should be used to + free the pointer returned by this function. +

Arguments:
+make_agent (input) to make an agent +thread of not. This is primarily used in the slurmdbd plugin.
+conn_num (input) connection number to +the plugin. In many cases you should plan on multiple simultanious +connections to the plugin. This number is useful since the debug +messages can print this out to determine which connection the message +is from.
+rollback (input) Allow rollback to +happen or not (in use with databases that support rollback).
+

Returns:
+void * which is an opaque structure +used inside the plugin to connection to the storage type on success, or
+NULL on failure. + +

int acct_storage_p_close_connection(void **db_conn) +

Description:
+acct_storage_p_close_connection() is called at the end of the program that has +called acct_storage_p_get_connection this function closes the connection to +the storage type. +

Arguments:
+db_conn (input/output) connection to +the storage type, all memory will be freed inside this function and +NULLed out. +

Returns:
+SLURM_SUCCESS on success, or
+SLURM_ERROR on failure. + +

int acct_storage_p_commit(void *db_conn, bool commit) +

Description:
+acct_storage_p_commit() is called at a point where you would either + want changes to storage be committed or rolled back. This function + should also send appropriate update messages to the various slurmctlds. +

Arguments:
+db_conn (input) connection to +the storage type.
+commit (input) true for commit, false +to rollback if connection was set up to rollback.
+

Returns:
+SLURM_SUCCESS on success, or
+SLURM_ERROR on failure. + +

+int acct_storage_p_add_users(void *db_conn, uint32_t uid, List user_list) +

Description:
+Called to add users to the storage type. +

Arguments:
+db_conn (input) connection to +the storage type.
+uid (input) uid of user calling the function.
+user_list (input) list of +acct_user_rec_t *'s containing information about the users being added.
+

Returns:
+SLURM_SUCCESS on success, or
+SLURM_ERROR on failure. + +

+int acct_storage_p_add_coord(void *db_conn, uint32_t uid, List acct_list, acct_user_cond_t *user_cond) +

Description:
+Called to link specified users to the specified accounts as coordinators. +

Arguments:
+db_conn (input) connection to +the storage type.
+uid (input) uid of user calling the function.
+acct_list (input) list of +acct_account_rec_t *'s containing information about the accounts to +add the coordinators to.
+user_cond (input) contain a list of +users to add to be coordinators of the acct_list.
+

Returns:
+SLURM_SUCCESS on success, or
+SLURM_ERROR on failure. + +

+int acct_storage_p_add_accts(void *db_conn, uint32_t uid, List acct_list) +

Description:
+Called to add accounts to the storage type. +

Arguments:
+db_conn (input) connection to +the storage type.
+uid (input) uid of user calling the function.
+acct_list (input) list of +acct_account_rec_t *'s containing information about the accounts to add.
+

Returns:
+SLURM_SUCCESS on success, or
+SLURM_ERROR on failure. + +

+int acct_storage_p_add_clusters(void *db_conn, uint32_t uid, List cluster_list) +

Description:
+Called to add clusters to the storage type. +

Arguments:
+db_conn (input) connection to +the storage type.
+uid (input) uid of user calling the function.
+cluster_list (input) list of +acct_cluster_rec_t *'s containing information about the clusters to add.
+

Returns:
+SLURM_SUCCESS on success, or
+SLURM_ERROR on failure. + +

+int acct_storage_p_add_associations(void *db_conn, uint32_t uid, List association_list) +

Description:
+Called to add associations to the storage type. +

Arguments:
+db_conn (input) connection to +the storage type.
+uid (input) uid of user calling the function.
+association_list (input) list of +acct_association_rec_t *'s containing information about the +associations to add.
+

Returns:
+SLURM_SUCCESS on success, or
+SLURM_ERROR on failure. + +

+int acct_storage_p_add_qos(void *db_conn, uint32_t uid, List qos_list) +

Description:
+Called to add QOS' to the storage type. +

Arguments:
+db_conn (input) connection to +the storage type.
+uid (input) uid of user calling the function.
+qos_list (input) list of +acct_qos_rec_t *'s containing information about the qos to add.
+

Returns:
+SLURM_SUCCESS on success, or
+SLURM_ERROR on failure. + +

+int acct_storage_p_add_wckeys(void *db_conn, uint32_t uid, List wckey_list) +

Description:
+Called to add wckeys to the storage type. +

Arguments:
+db_conn (input) connection to +the storage type.
+uid (input) uid of user calling the function.
+wckey_list (input) list of +acct_wckey_rec_t *'s containing information about the wckeys to add.
+

Returns:
+SLURM_SUCCESS on success, or
+SLURM_ERROR on failure. + +

+List acct_storage_p_modify_users(void *db_conn, uint32_t uid, +acct_user_cond_t *user_cond, acct_user_rec_t *user) +

Description:
+Used to modify existing users in the storage type. The condition + could include very vaque information about the user, so this + function should be robust in the ability to give everything the user + is asking for. This is the reason a list of modified users is + returned so the caller knows what has been changed, sometimes by mistake. +

Arguments:
+db_conn (input) connection to +the storage type.
+uid (input) uid of user calling the function.
+user_cond (input) conditional about +which users need to change. User names or ids should not need to be stated.
+user (input) what the changes +should be on the users identified by the conditional.
+

Returns:
+List containing names of users +modified on success, or
+NULL on failure. + +

+List acct_storage_p_modify_accounts(void *db_conn, uint32_t uid, +acct_account_cond_t *acct_cond, acct_account_rec_t *acct) +

Description:
+Used to modify existing accounts in the storage type. The condition + could include very vaque information about the account, so this + function should be robust in the ability to give everything the account + is asking for. This is the reason a list of modified accounts is + returned so the caller knows what has been changed, sometimes by mistake. +

Arguments:
+db_conn (input) connection to +the storage type.
+uid (input) uid of user calling the function.
+acct_cond (input) conditional about +which accounts need to change. Account names should not need to be stated.
+acct (input) what the changes +should be on the accounts identified by the conditional.
+

Returns:
+List containing names of users +modified on success, or
+NULL on failure. + +

+List acct_storage_p_modify_clusters(void *db_conn, uint32_t uid, +acct_cluster_cond_t *cluster_cond, acct_cluster_rec_t *cluster) +

Description:
+Used to modify existing clusters in the storage type. The condition + could include very vaque information about the cluster, so this + function should be robust in the ability to give everything the cluster + is asking for. This is the reason a list of modified clusters is + returned so the caller knows what has been changed, sometimes by mistake. +

Arguments:
+db_conn (input) connection to +the storage type.
+uid (input) uid of user calling the function.
+cluster_cond (input) conditional about +which clusters need to change. Cluster names should not need to be stated.
+cluster (input) what the changes +should be on the clusters identified by the conditional.
+

Returns:
+List containing names of clusters +modified on success, or
+NULL on failure. + +

+List acct_storage_p_modify_associations(void *db_conn, uint32_t uid, +acct_association_cond_t *assoc_cond, acct_association_rec_t *assoc) +

Description:
+Used to modify existing associations in the storage type. The condition + could include very vaque information about the association, so this + function should be robust in the ability to give everything the association + is asking for. This is the reason a list of modified associations is + returned so the caller knows what has been changed, sometimes by mistake. +

Arguments:
+db_conn (input) connection to +the storage type.
+uid (input) uid of user calling the function.
+assoc_cond (input) conditional about +which associations need to change. Association ids should not need to be stated.
+assoc (input) what the changes +should be on the associations identified by the conditional.
+

Returns:
+List containing names of associations +modified on success, or
+NULL on failure. + +

+List acct_storage_p_modify_qos(void *db_conn, uint32_t uid, +acct_qos_cond_t *qos_cond, acct_qos_rec_t *qos) +

Description:
+Used to modify existing qos in the storage type. The condition + could include very vaque information about the qos, so this + function should be robust in the ability to give everything the qos + is asking for. This is the reason a list of modified qos is + returned so the caller knows what has been changed, sometimes by mistake. +

Arguments:
+db_conn (input) connection to +the storage type.
+uid (input) uid of user calling the function.
+qos_cond (input) conditional about +which qos need to change. Qos names should not need to be stated.
+qos (input) what the changes +should be on the qos identified by the conditional.
+

Returns:
+List containing names of qos +modified on success, or
+NULL on failure. + +

+List acct_storage_p_modify_wckeys(void *db_conn, uint32_t uid, +acct_wckey_cond_t *wckey_cond, acct_wckey_rec_t *wckey) +

Description:
+Used to modify existing wckeys in the storage type. The condition + could include very vaque information about the wckeys, so this + function should be robust in the ability to give everything the wckey + is asking for. This is the reason a list of modified wckey is + returned so the caller knows what has been changed, sometimes by mistake. +

Arguments:
+db_conn (input) connection to +the storage type.
+uid (input) uid of user calling the function.
+wckey_cond (input) conditional about +which wckeys need to change. Wckey names should not need to be stated.
+wckey (input) what the changes +should be on the wckey identified by the conditional.
+

Returns:
+List containing names of wckeys +modified on success, or
+NULL on failure. + +

+List acct_storage_p_remove_users(void *db_conn, uint32_t uid, +acct_user_cond_t *user_cond) +

Description:
+Used to remove users from the storage type. This will remove all + associations. Must check to make sure all running jobs are finished + before this is allowed to execute. +

Arguments:
+db_conn (input) connection to +the storage type.
+uid (input) uid of user calling the function.
+user_cond (input) conditional about +which users to be removed. User names or ids should not need to be stated.
+

Returns:
+List containing names of users +removed on success, or
+NULL on failure. + +

+List acct_storage_p_remove_coord(void *db_conn, uint32_t uid, +List acct_list, acct_user_cond_t *user_cond) +

Description:
+Used to remove coordinators from the storage type. +

Arguments:
+db_conn (input) connection to +the storage type.
+uid (input) uid of user calling the function.
+acct_list (input) list of accounts +associated with the users.
+user_cond (input) conditional about +which users to be removed as coordinators. User names or ids should be stated.
+

Returns:
+List containing names of users +removed as coordinators on success, or
+NULL on failure. + +

+List acct_storage_p_remove_accounts(void *db_conn, uint32_t uid, +acct_account_cond_t *acct_cond) +

Description:
+Used to remove accounts from the storage type. This will remove all + associations from these accounts. You need to make sure no jobs are + running with any association that is to be removed. If any of these + accounts are default accounts for users that must also change before + an account can be removed. +

Arguments:
+db_conn (input) connection to +the storage type.
+uid (input) uid of user calling the function.
+acct_cond (input) conditional about +which accounts to be removed. Account names should not need to be stated.
+

Returns:
+List containing names of accounts +removed on success, or
+NULL on failure. + +

+List acct_storage_p_remove_clusters(void *db_conn, uint32_t uid, +acct_cluster_cond_t *cluster_cond) +

Description:
+Used to remove clusters from the storage type. This will remove all + associations from these clusters. You need to make sure no jobs are + running with any association that is to be removed. +

Arguments:
+db_conn (input) connection to +the storage type.
+uid (input) uid of user calling the function.
+cluster_cond (input) conditional about +which clusters to be removed. Cluster names should not need to be stated.
+

Returns:
+List containing names of clusters +removed on success, or
+NULL on failure. + +

+List acct_storage_p_remove_associations(void *db_conn, uint32_t uid, +acct_association_cond_t *assoc_cond) +

Description:
+Used to remove associations from the storage type. You need to make + sure no jobs are running with any association that is to be removed. +

Arguments:
+db_conn (input) connection to +the storage type.
+uid (input) uid of user calling the function.
+assoc_cond (input) conditional about +which associations to be removed. Association ids should not need to be stated.
+

Returns:
+List containing names of associations +removed on success, or
+NULL on failure. + +

+List acct_storage_p_remove_qos(void *db_conn, uint32_t uid, +acct_qos_cond_t *qos_cond) +

Description:
+Used to remove qos from the storage type. +

Arguments:
+db_conn (input) connection to +the storage type.
+uid (input) uid of user calling the function.
+qos_cond (input) conditional about +which qos to be removed. Qos names should not need to be stated.
+

Returns:
+List containing names of qos +removed on success, or
+NULL on failure. + +

+List acct_storage_p_remove_wckeys(void *db_conn, uint32_t uid, +acct_wckey_cond_t *wckey_cond) +

Description:
+Used to remove wckeys from the storage type. +

Arguments:
+db_conn (input) connection to +the storage type.
+uid (input) uid of user calling the function.
+wckey_cond (input) conditional about +which wckeys to be removed. Wckey names should not need to be stated.
+

Returns:
+List containing names of wckeys +removed on success, or
+NULL on failure. + +

+List acct_storage_p_get_users(void *db_conn, uint32_t uid, +acct_user_cond_t *user_cond) +

Description:
+Get a list of acct_user_rec_t *'s based on the conditional sent. +

Arguments:
+db_conn (input) connection to +the storage type.
+uid (input) uid of user calling the +function.
+user_cond (input) conditional about +which users are to be returned. User names or ids should not need to +be stated.
+

Returns:
+List containing acct_user_rec_t *'s +on success, or
+NULL on failure. + +

+List acct_storage_p_get_accts(void *db_conn, uint32_t uid, +acct_account_cond_t *acct_cond) +

Description:
+Get a list of acct_account_rec_t *'s based on the conditional sent. +

Arguments:
+db_conn (input) connection to +the storage type.
+uid (input) uid of user calling the +function.
+acct_cond (input) conditional about +which accounts are to be returned. Account names should not need to +be stated.
+

Returns:
+List containing acct_account_rec_t *'s +on success, or
+NULL on failure. + +

+List acct_storage_p_get_clusters(void *db_conn, uint32_t uid, +acct_cluster_cond_t *cluster_cond) +

Description:
+Get a list of acct_cluster_rec_t *'s based on the conditional sent. +

Arguments:
+db_conn (input) connection to +the storage type.
+uid (input) uid of user calling the +function.
+cluster_cond (input) conditional about +which clusters are to be returned. Cluster names should not need to +be stated.
+

Returns:
+List containing acct_cluster_rec_t *'s +on success, or
+NULL on failure. + +

+List acct_storage_p_get_associations(void *db_conn, uint32_t uid, +acct_association_cond_t *assoc_cond) +

Description:
+Get a list of acct_association_rec_t *'s based on the conditional sent. +

Arguments:
+db_conn (input) connection to +the storage type.
+uid (input) uid of user calling the +function.
+assoc_cond (input) conditional about +which associations are to be returned. Association names should not need to +be stated.
+

Returns:
+List containing acct_association_rec_t *'s +on success, or
+NULL on failure. + +

+List acct_storage_p_get_qos(void *db_conn, uint32_t uid, +acct_qos_cond_t *qos_cond) +

Description:
+Get a list of acct_qos_rec_t *'s based on the conditional sent. +

Arguments:
+db_conn (input) connection to +the storage type.
+uid (input) uid of user calling the +function.
+qos_cond (input) conditional about +which qos are to be returned. Qos names should not need to +be stated.
+

Returns:
+List containing acct_qos_rec_t *'s +on success, or
+NULL on failure. + +

+List acct_storage_p_get_wckeys(void *db_conn, uint32_t uid, +acct_wckey_cond_t *wckey_cond) +

Description:
+Get a list of acct_wckey_rec_t *'s based on the conditional sent. +

Arguments:
+db_conn (input) connection to +the storage type.
+uid (input) uid of user calling the +function.
+wckey_cond (input) conditional about +which wckeys are to be returned. Wckey names should not need to +be stated.
+

Returns:
+List containing acct_wckey_rec_t *'s +on success, or
+NULL on failure. + +

+List acct_storage_p_get_txn(void *db_conn, uint32_t uid, +acct_txn_cond_t *txn_cond) +

Description:
+Get a list of acct_txn_rec_t *'s (transactions) based on the conditional sent. +

Arguments:
+db_conn (input) connection to +the storage type.
+uid (input) uid of user calling the +function.
+txn_cond (input) conditional about +which transactions are to be returned. Transaction ids should not need to +be stated.
+

Returns:
+List containing acct_txn_rec_t *'s +on success, or
+NULL on failure. + +

+int acct_storage_p_get_usage(void *db_conn, uint32_t uid, void *in, int type, +time_t start, time_t end) +

Description:
+Get usage for a specific association or wckey. +

Arguments:
+db_conn (input) connection to +the storage type.
+uid (input) uid of user calling the +function.
+in (input/out) can be anything that +gathers usage like acct_associaiton_rec_t * or acct_wckey_rec_t *.
+type (input) really +slurmdbd_msg_type_t should let the plugin know what the structure is +that was sent in some how.
+start (input) start time of the usage.
+end (input) end time of the usage.
+

Returns:
+SLURM_SUCCESS on success, or
+SLURM_ERROR on failure. + +

+int acct_storage_p_roll_usage(void *db_conn, time_t sent_start) +

Description:
+roll up association, cluster, and wckey usage in the storage. +

Arguments:
+db_conn (input) connection to +the storage type.
+start (input) start time of the rollup.
+

Returns:
+SLURM_SUCCESS on success, or
+SLURM_ERROR on failure. + +

+int clusteracct_storage_p_node_down(void *db_conn, char *cluster, +struct node_record *node_ptr, time_t event_time, char *reason) +

Description:
+Mark nodes down in the storage type. +

Arguments:
+db_conn (input) connection to +the storage type.
+cluster (input) name of cluster node +is on.
+node_ptr (input) pointer to the node +structure marked down.
+event_time (input) time event happened.
+reason (input) if different from what +is set in the node_ptr, the reason the node is down.
+

Returns:
+SLURM_SUCCESS on success, or
+SLURM_ERROR on failure. + +

+int clusteracct_storage_p_node_up(void *db_conn, char *cluster, +struct node_record *node_ptr, time_t event_time) +

Description:
+Mark nodes up in the storage type. +

Arguments:
+db_conn (input) connection to +the storage type.
+cluster (input) name of cluster node +is on.
+node_ptr (input) pointer to the node +structure marked up.
+event_time (input) time event happened.
+

Returns:
+SLURM_SUCCESS on success, or
+SLURM_ERROR on failure. + +

+int clusteracct_storage_p_cluster_procs(void *db_conn, char *cluster, +uint32_t procs, time_t event_time) +

Description:
+Update storage type with the current number of processors on a given cluster. +

Arguments:
+db_conn (input) connection to +the storage type.
+cluster (input) name of cluster.
+procs (input) number of processors on +system.
+event_time (input) time event happened.
+

Returns:
+SLURM_SUCCESS on success, or
+SLURM_ERROR on failure. + +

+int clusteracct_storage_p_get_usage(void *db_conn, uint32_t uid, void +*cluster_rec, int type, time_t start, time_t end) +

Description:
+Get usage for a specific cluster. +

Arguments:
+db_conn (input) connection to +the storage type.
+uid (input) uid of user calling the +function.
+cluster_rec (input/out) +acct_cluster_rec_t * already set with the cluster name. Usage will be +filled in.
+type (input) really +slurmdbd_msg_type_t should let the plugin know what the structure is +that was sent in some how for this it is just DBD_GET_CLUSTER_USAGE.
+start (input) start time of the usage.
+end (input) end time of the usage.
+

Arguments:
+db_conn (input) connection to +the storage type. +

Returns:
+SLURM_SUCCESS on success, or
+SLURM_ERROR on failure. + +

+int clusteracct_storage_p_register_ctld(void *db_conn, char *cluster, +uint16_t port) +

Description:
+Used when a controller is turned on to tell the storage type where the + slurmctld for a given cluster is located at. +

Arguments:
+db_conn (input) connection to +the storage type.
+cluster (input) name of cluster.
+port (input) port on host cluster is +running on the host is grabbed from the connection.
+

Returns:
+SLURM_SUCCESS on success, or
+SLURM_ERROR on failure. + +

+int jobacct_storage_p_job_start(void *db_conn, struct job_record *job_ptr) +

Description:
+jobacct_storage_p_job_start() is called in the jobacct plugin when a +job starts, inserting information into the database about the new job. +

Arguments:
+db_conn (input) connection to +the storage type.
+job_ptr (input) information about the job in +slurmctld. +

Returns:
+SLURM_SUCCESS on success, or
+SLURM_ERROR on failure. + +

+int jobacct_storage_p_job_complete(void *db_conn, struct job_record *job_ptr) +

Description:
+jobacct_storage_p_job_complete() is called in the jobacct plugin when +a job completes, this updates info about end of a job. +

Arguments:
+db_conn (input) connection to +the storage type.
+job_ptr (input) information about the job in +slurmctld. +

Returns:
+SLURM_SUCCESS on success, or
+SLURM_ERROR on failure. + +

+int jobacct_storage_p_step_start(void *db_conn, struct step_record *step_ptr) +

Description:
+jobacct_storage_p_step_start() is called in the jobacct plugin at the +allocation of a new step in the slurmctld, this inserts info about the +beginning of a step. +

Arguments:
+db_conn (input) connection to +the storage type.
+step_ptr (input) information about the step in +slurmctld. +

Returns:
+SLURM_SUCCESS on success, or
+SLURM_ERROR on failure. + +

+int jobacct_storage_p_step_complete(void *db_conn, struct step_record *step_ptr) +

Description:
+jobacct_storage_p_step_complete() is called in the jobacct plugin at +the end of a step in the slurmctld, this updates the ending +information about a step. +

Arguments:
+db_conn (input) connection to +the storage type.
+step_ptr (input) information about the step in +slurmctld. +

Returns:
+SLURM_SUCCESS on success, or
+SLURM_ERROR on failure. + +

+int jobacct_storage_p_job_suspend(void *db_conn, struct job_record *job_ptr) +

Description:
+jobacct_storage_p_suspend() is called in the jobacct plugin when a +job is suspended or resumed in the slurmctld, this updates the +database about the suspended time of the job. +

Arguments:
+db_conn (input) connection to +the storage type.
+job_ptr (input) information about the job in +slurmctld. +

Returns:
+none + +

+List jobacct_storage_p_get_jobs_cond(void *db_conn, uint32_t uid, +acct_job_cond_t *job_cond) +

Description:
+jobacct_storage_p_get_jobs_cond() is called to get a list of jobs from the +database given the conditional. +

Arguments:
+db_conn (input) connection to +the storage type.
+uid (input) uid of user calling the function.
+job_cond (input) conditional about +which jobs to get. Job ids should not need to be stated.
+

Returns:
+List of job_rec_t's on success, or
+NULL on failure. + +

+int jobacct_storage_p_archive(void *db_conn, acct_archive_cond_t *arch_cond) +

Description:
+used to archive old data. +

Arguments:
+db_conn (input) connection to +the storage type.
+arch_cond (input) conditional about +what to archive.
+

Returns:
+SLURM_SUCCESS on success, or
+SLURM_ERROR on failure. + +

+int jobacct_storage_p_archive_load(void *db_conn, acct_archive_rect *arch_rec) +

Description:
+used to load old archive data. +

Arguments:
+db_conn (input) connection to +the storage type.
+arch_rec (input) information about +what to load.
+

Returns:
+SLURM_SUCCESS on success, or
+SLURM_ERROR on failure. + +

+int acct_storage_p_update_shares_used(void *db_conn, List acct_list) +

Description:
+Used to update shares used in the storage type. +

Arguments:
+db_conn (input) connection to +the storage type.
+acct_list (input) List of shares_used_object_t.
+

Returns:
+SLURM_SUCCESS on success, or
+SLURM_ERROR on failure. + +

+int acct_storage_p_flush_jobs_on_cluster(void *db_conn, char *cluster, time_t event_time) +

Description:
+used to mark all jobs in the storage type as finished. +

Arguments:
+db_conn (input) connection to +the storage type.
+cluster (input) name of cluster to +apply end to.
+event_time (input) when the flush happened.
+

Returns:
+SLURM_SUCCESS on success, or
+SLURM_ERROR on failure. + +

top + + +

Parameters

+

These parameters can be used in the slurm.conf to set up +connections to the database all have defaults based on the plugin type +used. +

+
AccountingStorageType +
Specifies which plugin should be used. +
AccountingStorageLoc +
Let the plugin the name of the logfile/database name to use. +
AccountingStorageHost +
Let the plugin know the host where the database is. +
AccountingStoragePort +
Let the plugin know the port to connect to. +
AccountingStorageUser +
Let the plugin know the name of the user to connect to the +database with. +
AccountingStoragePass +
Let the plugin know the password of the user connecting to the database. +
AccountingStorageEnforce +
Specifies if we should enforce certain things be in existance + before allowing job submissions and such valid options are + "associations, limits, and wckeys". You can use any combination of + those listed. +
+ +

Versioning

+

This document describes version 1 of the SLURM Accounting Storage API. Future +releases of SLURM may revise this API. An Accounting Storage plugin conveys its +ability to implement a particular API version using the mechanism outlined +for SLURM plugins. +

top + +

Last modified 10 February 2009

+ + diff -Nru slurm-llnl-1.3.13/doc/html/bluegene.shtml slurm-llnl-1.3.15/doc/html/bluegene.shtml --- slurm-llnl-1.3.13/doc/html/bluegene.shtml 2008-09-09 19:26:00.000000000 +0200 +++ slurm-llnl-1.3.15/doc/html/bluegene.shtml 2009-03-17 22:38:37.000000000 +0100 @@ -44,10 +44,12 @@ --geometry (specify job size in each dimension), --no-rotate (disable rotation of geometry), --conn-type (specify interconnect type between base partitions, mesh or torus). ---blrts-image (specify alternative blrts image for bluegene block. Default if not set). ---linux-image (specify alternative linux image for bluegene block. Default if not set). +--blrts-image (specify alternative blrts image for bluegene --block. Default if not set, BGL only.) +--cnload-image (specify alternative c-node image for bluegene block. Default if not set, BGP only.) +--ioload-image (specify alternative io image for bluegene block. Default if not set, BGP only.) +--linux-image (specify alternative linux image for bluegene block. Default if not set, BGL only.) --mloader-image (specify alternative mloader image for bluegene block. Default if not set). ---ramdisk-image (specify alternative ramdisk image for bluegene block. Default if not set). +--ramdisk-image (specify alternative ramdisk image for bluegene block. Default if not set, BGL only.) The --nodes option with a minimum and (optionally) maximum node count continues to be available. @@ -193,16 +195,13 @@

System Administration

-

As of IBM's REV 2 driver SLURM must be built in 64-bit mode. -This can be done by specifying CFLAGS=-m64. -CFLAGS must be set for SLURM to compile correctly.

-

Building a BlueGene compatible system is dependent upon the configure program locating some expected files. -In particular, the configure script searches for libdb2.so in the -directories /home/bgdb2cli/sqllib and /u/bgdb2cli/sqllib. -If your DB2 library file is in a different location, use the configure -option --with-db2-dir=PATH to specify the parent directory. +In particular for a BlueGene/L system, the configure script searches +for libdb2.so in the directories /home/bgdb2cli/sqllib +and /u/bgdb2cli/sqllib. If your DB2 library file is in a +different location, use the configure +option --with-db2-dir=PATH to specify the parent directory. If you have the same version of the operating system on both the Service Node (SN) and the Front End Nodes (FEN) then you can configure and build one set of files on the SN and install them on both the SN and FEN. @@ -238,7 +237,6 @@ The value of SelectType must be set to "select/bluegene" in order to have node selection performed using a system aware of the system's topography and interfaces. -The value of SchedulerType should be set to "sched/builtin". The value of Prolog should be set to the full pathname of a program that will delay execution until the bgblock identified by the MPIRUN_PARTITION environment variable is ready for use. It is recommended that you construct a script @@ -420,7 +418,7 @@ the user. If you change the bgblock layout, then slurmctld and slurmd should both be cold-started (e.g. /etc/init.d/slurm startclean). -If you which to modify the Numpsets values +If you wish to modify the Numpsets values for existing bgblocks, either modify them manually or destroy the bgblocks and let SLURM recreate them. Note that in addition to the bgblocks defined in bluegene.conf, an @@ -435,7 +433,7 @@ # # BlrtsImage: BlrtsImage used for creation of all bgblocks. # LinuxImage: LinuxImage used for creation of all bgblocks. -# tMloaderImage: MloaderImage used for creation of all bgblocks. +# MloaderImage: MloaderImage used for creation of all bgblocks. # RamDiskImage: RamDiskImage used for creation of all bgblocks. # # You may add extra images which a user can specify from the srun @@ -527,17 +525,18 @@ # BPs=[000x001] Type=TORUS # 1x1x2 = 2 midplanes ############################################################################### # volume = 1x1x1 = 1 -BPs=[000x000] Type=TORUS # 1x1x1 = 1 midplane -BPs=[001x001] Type=SMALL NodeCards=4 Quarters=3 # 1x1x1 = 4-Nodecard sized - # cnode blocks 3-Base - # Partition Quarter sized - # c-node blocks +BPs=[000x000] Type=TORUS # 1x1x1 = 1 midplane +BPs=[001x001] Type=SMALL 32CNBlocks=4 128CNBlocks=3 # 1x1x1 = 4-Nodecard sized + # cnode blocks 3-Base + # Partition Quarter sized + # c-node blocks

The above bluegene.conf file defines multiple bgblocks to be created in a single midplane (see the "SMALL" option). -Using this mechanism, up to 32 independent jobs can be executed +Using this mechanism, up to 32 independent jobs each consisting of 1 + 32 cnodes can be executed simultaneously on a one-rack BlueGene system. If defining bgblocks of Type=SMALL, the SLURM partition containing them as defined in slurm.conf must have the @@ -549,6 +548,10 @@ As in all SLURM configuration files, parameters and values are case insensitive.

+

With a BlueGene/P system the image names are different. The + correct image names are CnloadImage, MloaderImage, and IoloadImage. + You can also use alternate images just the same as discribed above. +

One more thing is required to support SLURM interactions with the DB2 database (at least as of the time this was written). DB2 database access is required by the slurmctld daemon only. @@ -624,10 +627,15 @@ non-BlueGene systems, the slurmd actually does run on the compute nodes, so the message is more meaningful there).

-

Note that you can emulate a BlueGene system on stand-alone Linux +

Note that you can emulate a BlueGene/L system on stand-alone Linux system. -Run configure with the --enable-bluegene-emulation option. -This will define "HAVE_BG" and "HAVE_FRONT_END" in the config.h file. +Run configure with the --enable-bgl-emulation option. +This will define "HAVE_BG", "HAVE_BGL", and "HAVE_FRONT_END" in the +config.h file. +You can also emulate a BlueGene/P system with + the --enable-bgp-emulation option. +This will define "HAVE_BG", "HAVE_BGP", and "HAVE_FRONT_END" in the +config.h file. Then execute make normally. These variables will build the code as if it were running on an actual BlueGene computer, but avoid making calls to the @@ -637,6 +645,6 @@

top

-

Last modified 9 September 2008

+

Last modified 17 March 2009

diff -Nru slurm-llnl-1.3.13/doc/html/documentation.shtml slurm-llnl-1.3.15/doc/html/documentation.shtml --- slurm-llnl-1.3.13/doc/html/documentation.shtml 2008-06-03 18:57:07.000000000 +0200 +++ slurm-llnl-1.3.15/doc/html/documentation.shtml 2009-02-11 23:52:55.000000000 +0100 @@ -40,7 +40,7 @@
  • Authentication Plugin Programmer Guide
  • Cryptographic Plugin Programmer Guild
  • Job Accounting Gather Plugin Programmer Guide
    • -
  • Job Accounting Storage Plugin Programmer Guide
    • +
  • Accounting Storage Plugin Programmer Guide
  • Job Checkpoint Plugin Programmer Guide
  • Job Completion Logging Plugin Programmer Guide
  • MPI Plugin Programmer Guide
    • diff -Nru slurm-llnl-1.3.13/doc/html/jobacct_storageplugins.shtml slurm-llnl-1.3.15/doc/html/jobacct_storageplugins.shtml --- slurm-llnl-1.3.13/doc/html/jobacct_storageplugins.shtml 2007-09-11 23:23:44.000000000 +0200 +++ slurm-llnl-1.3.15/doc/html/jobacct_storageplugins.shtml 1970-01-01 01:00:00.000000000 +0100 @@ -1,204 +0,0 @@ - - -

    SLURM Job Accounting Storage Plugin API

    - -

    Overview

    -

    This document describes SLURM Job Accounting Storage plugins and the API that -defines them. It is intended as a resource to programmers wishing to write -their own SLURM Job Accounting Storage plugins. This is version 1 of the API. - -

    SLURM Job Accounting Storage plugins must conform to the -SLURM Plugin API with the following specifications: - -

    const char -plugin_name[]="full text name" -

    -A free-formatted ASCII text string that identifies the plugin. - -

    const char -plugin_type[]="major/minor"
    -

    -The major type must be "jobacct_storage." -The minor type can be any suitable name -for the type of accounting package. We currently use -

      -
    • filetxt—Information written to a text file. -
    • mysql— Store information in a mysql database. -
    • pgsql— Store information in a postgresql database. -
    • none— Information is not stored anywhere. -
    -

    The programmer is urged to study -src/plugins/jobacct_storage/mysql -for a sample implementation of a SLURM Job Accounting Storage plugin. -

    The Job Accounting Storage plugin was written to be a interface -to storage data collected by the Job Accounting Gather plugin. When -adding a new database you may want to add common functions in a common -file in the src/common dir. Refer to src/common/mysql_common.c/.h for an -example so other plugins can also use that database type to write out -information. -

    top - - -

    API Functions

    - -The Job Accounting Storage API uses hooks in the slurmctld. - -

    All of the following functions are required. Functions which are not -implemented must be stubbed. - -

    Functions called by the jobacct_storage plugin

    - -

    int jobacct_storage_p_init(char *location) -

    Description:
    -jobacct_storage_p_init() is called to initiate a connection to the -database server and check the state of the database table to make sure -they are in sync with the table definitions in the plugin. -Put global initialization here. Or open file or anything to initialize -the plugin. -

    Arguments:
    -location (input) database name or log -file location. -

    Returns:
    -SLURM_SUCCESS on success, or
    -SLURM_ERROR on failure. - -

    int jobacct_storage_p_fini() -

    Description:
    -jobacct_storage_p_fini() is called at the end of the program that has -called jobacct_storage_p_init this function closes the connection to -the database or logfile. -

    Arguments:
    -none -

    Returns:
    -SLURM_SUCCESS on success, or
    -SLURM_ERROR on failure. - -

    -int jobacct_storage_p_job_start(struct job_record *job_ptr) -

    Description:
    -jobacct_storage_p_job_start() is called in the jobacct plugin when a -job starts, inserting information into the database about the new job. -

    Arguments:
    -job_ptr (input) information about the job in -slurmctld. -

    Returns:
    -SLURM_SUCCESS on success, or
    -SLURM_ERROR on failure. - -

    -int jobacct_storage_p_job_complete(struct job_record *job_ptr) -

    Description:
    -jobacct_storage_p_job_complete() is called in the jobacct plugin when -a job completes, this updates info about end of a job. -

    Arguments:
    -job_ptr (input) information about the job in -slurmctld. -

    Returns:
    -SLURM_SUCCESS on success, or
    -SLURM_ERROR on failure. - -

    -int jobacct_storage_p_step_start(struct step_record *step_ptr) -

    Description:
    -jobacct_storage_p_step_start() is called in the jobacct plugin at the -allocation of a new step in the slurmctld, this inserts info about the -beginning of a step. -

    Arguments:
    -step_ptr (input) information about the step in -slurmctld. -

    Returns:
    -SLURM_SUCCESS on success, or
    -SLURM_ERROR on failure. - -

    -int jobacct_storage_p_step_complete(struct step_record *step_ptr) -

    Description:
    -jobacct_storage_p_step_complete() is called in the jobacct plugin at -the end of a step in the slurmctld, this updates the ending -information about a step. -

    Arguments:
    -step_ptr (input) information about the step in -slurmctld. -

    Returns:
    -SLURM_SUCCESS on success, or
    -SLURM_ERROR on failure. - -

    -int jobacct_storage_p_suspend(struct job_record *job_ptr) -

    Description:
    -jobacct_storage_p_suspend() is called in the jobacct plugin when a -job is suspended or resumed in the slurmctld, this updates the -database about the suspended time of the job. -

    Arguments:
    -job_ptr (input) information about the job in -slurmctld. -

    Returns:
    -none - -

    -void jobacct_storage_p_get_jobs(List job_list, List selected_steps, -List selected_parts, void *params) -

    Description:
    -jobacct_storage_p_get_jobs() is called to get a list of jobs from the -database given the specific inputs. -

    Arguments:
    -List job_list (input/output) list to -be filled with jobacct_job_rec_t.
    -List selected_steps -(input) list containing type jobacct_select_step_t to query against.
    -List selected_parts -(input) list containing char *'s of names of partitions to query against.
    -void *params -(input) to be cast as sacct_parameters_t in the plugin. - -

    jobacct_job_rec_t, jobacct_select_step_t, -and sacct_parameters_t are -all defined in common/slurm_jobacct.h -

    Returns:
    -none - -

    -void jobacct_storage_p_archive(List selected_parts, void *params) -

    Description:
    -database_p_jobcomp_archive() used to archive old data. -

    Arguments:
    -List selected_parts -(input) list containing char *'s of names of partitions to query against.
    -void *params -(input) to be cast as sacct_parameters_t in the plugin. -

    Returns:
    -none - -

    top - - -

    Parameters

    -

    These parameters can be used in the slurm.conf to set up -connections to the database all have defaults based on the plugin type -used. -

    -
    JobAcctStorageType -
    Specifies which plugin should be used. -
    JobAcctStorageLoc -
    Let the plugin the name of the logfile/database name to use. -
    JobAcctStorageHost -
    Let the plugin know the host where the database is. -
    JobAcctStoragePort -
    Let the plugin know the port to connect to. -
    JobAcctStorageUser -
    Let the plugin know the name of the user to connect to the -database with. -
    JobAcctStoragePass -
    Let the plugin know the password of the user connecting to the database. -
    - -

    Versioning

    -

    This document describes version 1 of the SLURM Job Accounting Storage API. Future -releases of SLURM may revise this API. A Job Accounting Storage plugin conveys its -ability to implement a particular API version using the mechanism outlined -for SLURM plugins. -

    top - -

    Last modified 23 May 2007

    - - diff -Nru slurm-llnl-1.3.13/doc/html/Makefile.am slurm-llnl-1.3.15/doc/html/Makefile.am --- slurm-llnl-1.3.13/doc/html/Makefile.am 2008-04-12 00:09:57.000000000 +0200 +++ slurm-llnl-1.3.15/doc/html/Makefile.am 2009-02-11 23:52:55.000000000 +0100 @@ -19,7 +19,7 @@ help.html \ ibm.html \ jobacct_gatherplugins.html \ - jobacct_storageplugins.html \ + accounting_storageplugins.html \ jobcompplugins.html \ mail.html \ maui.html \ diff -Nru slurm-llnl-1.3.13/doc/html/Makefile.in slurm-llnl-1.3.15/doc/html/Makefile.in --- slurm-llnl-1.3.13/doc/html/Makefile.in 2008-12-29 22:56:17.000000000 +0100 +++ slurm-llnl-1.3.15/doc/html/Makefile.in 2009-02-11 23:52:55.000000000 +0100 @@ -265,7 +265,7 @@ help.html \ ibm.html \ jobacct_gatherplugins.html \ - jobacct_storageplugins.html \ + accounting_storageplugins.html \ jobcompplugins.html \ mail.html \ maui.html \ diff -Nru slurm-llnl-1.3.13/doc/html/overview.shtml slurm-llnl-1.3.15/doc/html/overview.shtml --- slurm-llnl-1.3.13/doc/html/overview.shtml 2008-03-11 21:27:01.000000000 +0100 +++ slurm-llnl-1.3.15/doc/html/overview.shtml 2009-02-11 23:52:55.000000000 +0100 @@ -54,9 +54,8 @@
  • Cryptography: Munge or OpenSSL
  • Job Accounting Gather: AIX, Linux, or none(default)
    • -
  • Job Accounting Storage: +
  • Accounting Storage: text file (default if jobacct_gather != none), -Gold MySQL, PGSQL, SlurmDBD (Slurm Database Daemon) or none
  • Job completion logging: @@ -67,7 +66,7 @@ other versions of MPI including MPICH2 and MVAPICH2).
  • Node selection: -Blue Gene (a 3-D torus interconnect), +Bluegene (a 3-D torus interconnect BGL or BGP), consumable resources (to allocate individual processors and memory) or linear (to dedicate entire nodes).
    • diff -Nru slurm-llnl-1.3.13/doc/html/programmer_guide.shtml slurm-llnl-1.3.15/doc/html/programmer_guide.shtml --- slurm-llnl-1.3.13/doc/html/programmer_guide.shtml 2008-10-07 22:14:38.000000000 +0200 +++ slurm-llnl-1.3.15/doc/html/programmer_guide.shtml 2009-02-11 23:52:55.000000000 +0100 @@ -95,11 +95,11 @@ plugins—Plugin functions for various infrastructure. A separate subdirectory is used for each plugin class:
      +
    • accounting_storage for specifing the type of storage for accounting,
    • auth for user authentication,
    • checkpoint for system-initiated checkpoint and restart of user jobs,
    • crypto for cryptographic functions,
    • jobacct_gather for job accounting,
      -
    • jobacct_storage for specifing the type of storage for job accounting,
    • jobcomp for job completion logging,
    • mpi for MPI support,
    • proctrack for process tracking,
      @@ -109,6 +109,8 @@
    • task for task affinity to processors.

    +sacct—User command to view accounting information about jobs.
    +sacctmgr—User and administrator tool to manage accounting.
    salloc—User command to allocate resources for a job.
    sattach—User command to attach standard input, output and error files to a running job or job step.
    @@ -122,12 +124,15 @@ slurmd—SLURM daemon code to manage the compute server nodes including the execution of user applications.
    smap—User command to view layout of nodes, partitions, and jobs. -This is particularly valuable on systems like Blue Gene, which has a three +This is particularly valuable on systems like Bluegene, which has a three dimension torus topography.
    squeue—User command to get information on SLURM jobs and job steps.
    +sreport—User command to view various reports about past +usage across the enterprise.
    srun—User command to submit a job, get an allocation, and/or initiation a parallel job step.
    -strigger—User and administrator to manage event triggers.
    +sstat—User tool to status running jobs.
    +strigger—User and administrator tool to manage event triggers.
    sview—User command to view and update node, partition, and job job state information.
    diff -Nru slurm-llnl-1.3.13/doc/html/review_release.html slurm-llnl-1.3.15/doc/html/review_release.html --- slurm-llnl-1.3.13/doc/html/review_release.html 2008-03-25 23:31:59.000000000 +0100 +++ slurm-llnl-1.3.15/doc/html/review_release.html 2009-02-11 23:52:55.000000000 +0100 @@ -9,6 +9,7 @@ NOTE: Do not follow links.

    -

    Last modified 1 November 2008

    +

    Last modified 20 January 2009

    diff -Nru slurm-llnl-1.3.13/doc/man/man1/sacct.1 slurm-llnl-1.3.15/doc/man/man1/sacct.1 --- slurm-llnl-1.3.13/doc/man/man1/sacct.1 2008-12-09 23:22:06.000000000 +0100 +++ slurm-llnl-1.3.15/doc/man/man1/sacct.1 2009-04-22 11:23:58.000000000 +0200 @@ -24,7 +24,7 @@ command displays information on jobs, job steps, status, and exitcodes by default. You can tailor the output with the use of the -\f3\-\-fields=\fP +\f3\-\-format=\fP option to specify the fields to be shown. .PP For the root user, the @@ -51,6 +51,17 @@ \f3sacct\fP reports values of 0 for this missing data. See your systems \f2getrusage(3)\fP man page for information about which data are actually available on your system. +.IP +If \-\-dump is specified, the field selection options (\-\-brief, +\-\-format, ...) have no effect. +.IP +Elapsed time fields are presented as 2 fields, integral seconds and integral microseconds +.IP +If \-\-dump is not specified, elapsed time fields are presented as +[[days-]hours:]minutes:seconds.hundredths. +.IP +The default input file is the file named in the jobacct_logfile +parameter in slurm.conf. .SS "Options" .TP "10" @@ -60,7 +71,13 @@ This is the default behavior when the .BR "sacct " command is executed by the root user. - +.TP +\f3\-A \fP\f2account_list\fP\f3,\fP \f3\-\-accounts\fP\f3=\fP\f2account_list\fP +Displays the statistics only for the jobs started on the accounts specified by +the \f2account_list\fP operand, which is a comma\-separated list of +account names. +Space characters are not allowed in the \f2account_list\fP. Default is +all accounts\&. .TP \f3\-b \fP\f3,\fP \f3\-\-brief\fP Displays a brief listing, which includes the following data: @@ -81,7 +98,7 @@ option is also specified. .TP -\f3\-C \fP\f2cluster_list\fP\f3,\fP \f3\-\-cluster\fP\f3=\fP\f2cluster_list\fP +\f3\-C \fP\f2cluster_list\fP\f3,\fP \f3\-\-clusters\fP\f3=\fP\f2cluster_list\fP Displays the statistics only for the jobs started on the clusters specified by the \f2cluster_list\fP operand, which is a comma\-separated list of clusters. Space characters are not allowed in the \f2cluster_list\fP. \-1 for @@ -95,12 +112,31 @@ This option overrides the \f3\-\-brief\fP and -\f3\-\-fields=\fP +\f3\-\-format=\fP options. .IP The section titled "INTERPRETING THE \-\-dump OPTION OUTPUT" describes the data output when this option is used. +.TP +\f3\-\-duplicates\fP +If SLURM job ids are reset, but the job accounting log file isn't +reset at the same time (with \-e, for example), some job numbers will +probably appear more than once in the accounting log file to refer to +different jobs; such jobs can be distinguished by the "submit" time +stamp in the data records. +.IP +When data for specific jobs are requested with the \-\-jobs option, we +assume that the user wants to see only the most recent job with that +number. This behavior can be overridden by specifying \-\-duplicates, in +which case all records that match the selection criteria will be returned. +.IP +When \-\-jobs is not specified, we report data for all jobs that match +the selection criteria, even if some of the job numbers are +reused. Specify that you only want the most recent job for each +selected job number with the \-\-noduplicates option. + + .TP \f3\-e \fP\f2time_spec\fP \f3,\fP \f3\-\-expire=\fP\f2time_spec\fP .IP @@ -121,30 +157,38 @@ the accounting log file is /var/log/slurmacct.log, the expired log file will be /var/log/slurmacct.log.expired. +.TP +\f3\-\-endtime\fP\f3=\fP\f2endtime\fP +Select jobs eligible before this time. +Valid Formats are. + HH:MM[:SS] [AM|PM] + MMDD[YY] or MM/DD[/YY] or MM.DD[.YY] + MM/DD[/YY]-HH:MM[:SS] + .TP -\f3\-F \fP\f2field_list\fP \f3,\fP \f3\-\-fields\fP\f3=\fP\f2field_list\fP +\f3\-F \fP\f2format_list\fP \f3,\fP \f3\-\-format\fP\f3=\fP\f2format_list\fP Displays the job accounting data specified by the -\f2field_list\fP +\f2format_list\fP operand, which is a comma\-separated list of fields. Space characters are not allowed in the -\f2field_list\fP\c +\f2format_list\fP\c \&. .IP See the -\f3\-\-help\-fields\fP +\f3\-\-helpformat\fP option for a list of the available fields. See the section titled "Job Accounting Fields" for a description of each field. .IP The job accounting data is displayed in the order specified by the -\f2field_list\fP +\f2format_list\fP operand. Thus, the following two commands display the same data but in different order: .RS .PP .nf .ft 3 -# sacct \-\-fields=jobid,status +# sacct \-\-format=jobid,status Jobid Status \-\-\-\-\-\-\-\-\-\- \-\-\-\-\-\-\-\-\-\- 3 COMPLETED @@ -157,7 +201,7 @@ .PP .nf .ft 3 -# sacct \-\-fields=status,jobid +# sacct \-\-format=status,jobid Status Jobid \-\-\-\-\-\-\-\-\-\- \-\-\-\-\-\-\-\-\-\- COMPLETED 3 @@ -204,9 +248,9 @@ \f3\-h \fP\f3,\fP \f3\-\-help\fP Displays a general help message. .TP -\f3\-\-help\-fields\fP +\f3\-\-helpformat\fP Displays a list of fields that can be specified with the -\f3\-\-fields\fP +\f3\-\-format\fP option. .RS .PP @@ -214,16 +258,16 @@ .ft 3 Fields available: account associd cluster cpu -cputime elapsed end exitcode -finished gid group job -jobid jobname ncpus nodes -nnodes nprocs ntasks pages -partition rss start state -status submit timelimit submitted -systemcpu uid user usercpu -vsize blockid connection geo -max_procs reboot rotate bg_start_point -wckey +cputime elapsed eligible end +exitcode finished gid group +job jobid jobname ncpus +nodes nnodes nprocs ntasks +pages partition rss start +state status submit timelimit +submitted systemcpu uid user +usercpu vsize blockid connection +geo max_procs reboot rotate +bg_start_point wckey .ft 1 .fi @@ -285,6 +329,10 @@ .RE .TP +\f3\-\-noduplicates\fP +See the discussion under \-\-duplicates. + +.TP \f3\-\-noheader\fP Prevents the display of the heading over the output. The default action is to display a header. @@ -299,6 +347,10 @@ .IP This option is provided for debugging. +.TP +\f3\-P \fP\f3,\fP \f3\-\-purge\fP +Used in conjunction with \-\-expire to remove invalid data from the job accounting log. + .TP \f3\-p \fP\f2partition_list\fP \f3,\fP \f3\-\-partition\fP\f3=\fP\f2partition_list\fP Displays information about jobs and job steps specified by the @@ -380,6 +432,14 @@ \f2state_list\fP\c \&. +.TP +\f3\-\-starttime\fP\f3=\fP\f2starttime\fP +Select jobs eligible after this time. +Valid Formats are. + HH:MM[:SS] [AM|PM] + MMDD[YY] or MM/DD[/YY] or MM.DD[.YY] + MM/DD[/YY]-HH:MM[:SS] + .TP \f3\-t \fP\f3,\fP \f3\-\-total\fP Displays only the cumulative statistics for each job. @@ -410,6 +470,14 @@ Reports the state of certain variables during processing. This option is primarily used for debugging. +.TP +\f3\-W \fP\f2wckey_list\fP\f3,\fP \f3\-\-wckeys\fP\f3=\fP\f2wckey_list\fP +Displays the statistics only for the jobs started on the wckeys specified by +the \f2wckey_list\fP operand, which is a comma\-separated list of +wckey names. +Space characters are not allowed in the \f2wckey_list\fP. Default is +all wckeys\&. + .SS "Job Accounting Fields" The following describes each job accounting field: .RS diff -Nru slurm-llnl-1.3.13/doc/man/man1/sacctmgr.1 slurm-llnl-1.3.15/doc/man/man1/sacctmgr.1 --- slurm-llnl-1.3.13/doc/man/man1/sacctmgr.1 2009-01-07 18:07:16.000000000 +0100 +++ slurm-llnl-1.3.15/doc/man/man1/sacctmgr.1 2009-02-02 23:36:37.000000000 +0100 @@ -228,7 +228,7 @@ before it is enforced.) .TP -\fIGrpJobs\fP= +\fIGrpJobs\fP= Maximum number of running jobs in aggregate for this association and all association which are children of this association. To clear a previously set value use the modify command with a new value of \-1. diff -Nru slurm-llnl-1.3.13/doc/man/man1/salloc.1 slurm-llnl-1.3.15/doc/man/man1/salloc.1 --- slurm-llnl-1.3.13/doc/man/man1/salloc.1 2008-12-09 23:22:06.000000000 +0100 +++ slurm-llnl-1.3.15/doc/man/man1/salloc.1 2009-03-17 19:50:45.000000000 +0100 @@ -593,12 +593,25 @@ applicable to other systems as well. .TP +\fB\-\-blrts\-image\fR[=]<\fIpath\fR> +Path to blrts image for bluegene block. BGL only. +Default from \fIblugene.conf\fR if not set. + +.TP +\fB\-\-cnload\-image\fR=\fIpath\fR +Path to compute node image for bluegene block. BGP only. +Default from \fIblugene.conf\fR if not set. + +.TP \fB\-\-conn\-type\fR[=]<\fItype\fR> Require the partition connection type to be of a certain type. On Blue Gene the acceptable of \fItype\fR are MESH, TORUS and NAV. If NAV, or if not set, then SLURM will try to fit a TORUS else MESH. You should not normally set this option. SLURM will normally allocate a TORUS if possible for a given geometry. +If running on a BGP system and wanting to run in HTC mode (only for 1 +midplane and below). You can use HTC_S for SMP, HTC_D for Dual, HTC_V +for virtual node mode, and HTC_L for Linux mode. .TP \fB\-g\fR, \fB\-\-geometry\fR[=]<\fIXxYxZ\fR> @@ -609,12 +622,32 @@ Blue Gene). .TP +\fB\-\-ioload\-image\fR=\fIpath\fR +Path to io image for bluegene block. BGP only. +Default from \fIblugene.conf\fR if not set. + +.TP +\fB\-\-linux\-image\fR[=]<\fIpath\fR> +Path to linux image for bluegene block. BGL only. +Default from \fIblugene.conf\fR if not set. + +.TP +\fB\-\-mloader\-image\fR[=]<\fIpath\fR> +Path to mloader image for bluegene block. +Default from \fIblugene.conf\fR if not set. + +.TP \fB\-R\fR, \fB\-\-no\-rotate\fR Disables rotation of the job's requested geometry in order to fit an appropriate partition. By default the specified geometry can rotate in three dimensions. .TP +\fB\-\-ramdisk\-image\fR[=]<\fIpath\fR> +Path to ramdisk image for bluegene block. BGL only. +Default from \fIblugene.conf\fR if not set. + +.TP \fB\-\-reboot\fR Force the allocated nodes to reboot before starting the job. diff -Nru slurm-llnl-1.3.13/doc/man/man1/sbatch.1 slurm-llnl-1.3.15/doc/man/man1/sbatch.1 --- slurm-llnl-1.3.13/doc/man/man1/sbatch.1 2008-12-09 23:22:06.000000000 +0100 +++ slurm-llnl-1.3.15/doc/man/man1/sbatch.1 2009-03-17 19:50:45.000000000 +0100 @@ -657,6 +657,13 @@ slurm.conf this value does not get looked at. .TP +\fB\-\-wrap\fR[=]<\fIcommand string\fR> +Sbatch will wrap the specified command string in a simple "sh" shell script, +and submit that script to the slurm controller. When \-\-wrap is used, +a script name and arguments may not be specified on the command line; instead +the sbatch-generated wrapper script is used. + +.TP \fB\-x\fR, \fB\-\-exclude\fR[=]<\fInode name list\fR> Explicitly exclude certain nodes from the resources granted to the job. @@ -666,7 +673,12 @@ .TP \fB\-\-blrts\-image\fR[=]<\fIpath\fR> -Path to blrts image for bluegene block. +Path to blrts image for bluegene block. BGL only. +Default from \fIblugene.conf\fR if not set. + +.TP +\fB\-\-cnload\-image\fR=\fIpath\fR +Path to compute node image for bluegene block. BGP only. Default from \fIblugene.conf\fR if not set. .TP @@ -676,6 +688,9 @@ If NAV, or if not set, then SLURM will try to fit a TORUS else MESH. You should not normally set this option. SLURM will normally allocate a TORUS if possible for a given geometry. +If running on a BGP system and wanting to run in HTC mode (only for 1 +midplane and below). You can use HTC_S for SMP, HTC_D for Dual, HTC_V +for virtual node mode, and HTC_L for Linux mode. .TP \fB\-g\fR, \fB\-\-geometry\fR[=]<\fIXxYxZ\fR> @@ -686,8 +701,13 @@ Blue Gene). .TP +\fB\-\-ioload\-image\fR=\fIpath\fR +Path to io image for bluegene block. BGP only. +Default from \fIblugene.conf\fR if not set. + +.TP \fB\-\-linux\-image\fR[=]<\fIpath\fR> -Path to linux image for bluegene block. +Path to linux image for bluegene block. BGL only. Default from \fIblugene.conf\fR if not set. .TP @@ -703,20 +723,13 @@ .TP \fB\-\-ramdisk\-image\fR[=]<\fIpath\fR> -Path to ramdisk image for bluegene block. +Path to ramdisk image for bluegene block. BGL only. Default from \fIblugene.conf\fR if not set. .TP \fB\-\-reboot\fR Force the allocated nodes to reboot before starting the job. -.TP -\fB\-\-wrap\fR[=]<\fIcommand string\fR> -Sbatch will wrap the specified command string in a simple "sh" shell script, -and submit that script to the slurm controller. When \-\-wrap is used, -a script name and arguments may not be specified on the command line; instead -the sbatch-generated wrapper script is used. - .SH "INPUT ENVIRONMENT VARIABLES" .PP Upon startup, sbatch will read and handle the options set in the following diff -Nru slurm-llnl-1.3.13/doc/man/man1/scontrol.1 slurm-llnl-1.3.15/doc/man/man1/scontrol.1 --- slurm-llnl-1.3.13/doc/man/man1/scontrol.1 2008-12-04 02:16:52.000000000 +0100 +++ slurm-llnl-1.3.15/doc/man/man1/scontrol.1 2009-02-13 01:46:19.000000000 +0100 @@ -447,7 +447,12 @@ .TP \fIDefault\fP= Specify if this partition is to be used by jobs which do not explicitly -identify a partition to use. Possible values are"YES" and "NO". +identify a partition to use. +Possible output values are "YES" and "NO". +In order to change the default partition of a running system, +use the scontrol update command and set Default=yes for the partition +that you want to become the new default. + .TP \fIHidden\fP= Specify if the partition and its jobs should be hidden from view. diff -Nru slurm-llnl-1.3.13/doc/man/man1/squeue.1 slurm-llnl-1.3.15/doc/man/man1/squeue.1 --- slurm-llnl-1.3.13/doc/man/man1/squeue.1 2008-10-15 00:38:17.000000000 +0200 +++ slurm-llnl-1.3.15/doc/man/man1/squeue.1 2009-01-29 19:16:08.000000000 +0100 @@ -265,6 +265,11 @@ .RE .TP +\fB\-U \fR, \fB\-\-account=\fR +Specify the accounts of the jobs to view. Accepts a comma separated +list of account names. This have no effect when listing job steps. + +.TP \fB\-p \fR, \fB\-\-partition=\fR Specify the partitions of the jobs or steps to view. Accepts a comma separated list of partition names. diff -Nru slurm-llnl-1.3.13/doc/man/man1/sreport.1 slurm-llnl-1.3.15/doc/man/man1/sreport.1 --- slurm-llnl-1.3.13/doc/man/man1/sreport.1 2008-12-30 21:48:54.000000000 +0100 +++ slurm-llnl-1.3.15/doc/man/man1/sreport.1 2009-04-02 01:06:12.000000000 +0200 @@ -4,10 +4,10 @@ sreport \- Used to generate reports from the slurm accounting data. .SH "SYNOPSIS" -\fBssreport\fR [\fIOPTIONS\fR...] [\fICOMMAND\fR...] +\fBsreport\fR [\fIOPTIONS\fR...] [\fICOMMAND\fR...] .SH "DESCRIPTION" -\fBssreport\fR is used to generate certain reports. More can be added +\fBsreport\fR is used to generate certain reports. More can be added at any time. It provides a view into accounting data gathered from slurm via the account information maintained within a database with the interface being provided by the \fBslurmdbd\fR (Slurm Database daemon). @@ -20,7 +20,7 @@ .TP \fB\-h\fR, \fB\-\-help\fR -Print a help message describing the usage of \fBssreport\fR. +Print a help message describing the usage of \fBsreport\fR. This is equivalent to the \fBhelp\fR command. .TP @@ -130,7 +130,7 @@ Various reports are as follows... cluster \- AccountUtilizationByUser, UserUtilizationByAccount, UserUtilizationByWckey, Utilization, WCKeyUtilizationByUser - job \- Sizes + job \- SizesByAccount, SizesByWckey user \- TopUsage .TP @@ -168,12 +168,17 @@ each user on each cluster. .TP -.B job Sizes +.B job SizesByAccount This report will dispay the amount of time used for job ranges specified by the 'grouping=' option. Only a single level in the tree is displayed defaulting to the root dir. If you specify other -accounts with the 'account=' option you will receive those accounts -sub accounts. +accounts with the 'account=' option sreport will use those accounts as +the root account and you will receive the sub accounts for the +accounts listed. +.TP +.B job SizesByWckey +This report will dispay the amount of time for each wckey for job ranges +specified by the 'grouping=' option. .TP .B user TopUsage @@ -191,26 +196,47 @@ .B All_Clusters Use all monitored clusters default is local cluster. .TP +.B Clusters= +List of clusters to include in report. Default is local cluster. +.TP .B End= Period ending for report. Default is 23:59:59 of previous day. +Valid time formats are... +HH:MM[:SS] [AM|PM] +MMDD[YY] or MM/DD[/YY] or MM.DD[.YY] +MM/DD[/YY]-HH:MM[:SS] .TP .B Format= Comma separated list of fields to display in report. .TP .B Start= Period start for report. Default is 00:00:00 of previous day. +Valid time formats are... +HH:MM[:SS] [AM|PM] +MMDD[YY] or MM/DD[/YY] or MM.DD[.YY] +MM/DD[/YY]-HH:MM[:SS] .RE .TP CLUSTER .RS .TP -.B Names= -List of clusters to include in report. Default is local cluster. +.B Accounts= +When used with the UserUtilizationByAccount, or +AccountUtilizationByUser, List of accounts to include in report. +Default is all. .TP .B Tree When used with the AccountUtilizationByUser report will span the accounts as they in the hierarchy. +.TP +.B Users= +When used with any report other than Utilization, List of users to +include in report. Default is all. +.TP +.B Wckeys= +When used with the UserUtilizationByWckey or WCKeyUtilizationByUser, +List of wckeys to include in report. Default is all. .RE .TP @@ -218,10 +244,17 @@ .RS .TP .B Accounts= -List of accounts to use for the report Default is all. -.TP -.B Clusters= -List of clusters to include in report. Default is local cluster. +List of accounts to use for the report Default is all. The SizesByAccount +report only displays 1 hierarchical level. If accounts are specified +the next layer of accounts under those specified will be displayed, +not the accounts specified. In the SizesByAccount reports the default +for accounts is root. This explanation does not apply when ran with +the FlatView option. +.TP +.B FlatView +When used with the SizesbyAccount will not group accounts in a +hierarchical level, but print each account where jobs ran on a +separate line without any hierarchy. .TP .B GID= List of group ids to include in report. Default is all. @@ -239,38 +272,83 @@ When used with the Sizes report will print number of jobs ran instead of time used. .TP -.B TopCount= -Used in the TopUsage report. Change the number of users displayed. -Default is 10. -.TP .B Users= List of users jobs to include in report. Default is all. +.TP +.B Wckeys= +List of wckeys to use for the report. Default is all. The +SizesbyWckey report all users summed together. If you want only +certain users specify them them with the Users= option. .RE .TP USER .RS .TP -.B Clusters= -List of clusters to include in report. Default is local cluster. +.B Accounts= +List of accounts to use for the report. Default is all. .TP .B Group -Group all accounts together for each user. Default is a separate entry for each user and account reference. +Group all accounts together for each user. Default is a separate +entry for each user and account reference. +.TP +.B TopCount= +Used in the TopUsage report. Change the number of users displayed. +Default is 10. .TP .B Users= List of users jobs to include in report. Default is all. .RE +.TP + +.SH "Format Options for Each Report" +\fBCluster\fP + AccountUtilizationByUser + UserUtilizationByAccount + \- Accounts, Cluster, CPUCount, Login, Proper, Used + UserUtilizationByWckey + WCKeyUtilizationByUser + \- Cluster, CPUCount, Login, Proper, Used, Wckey + Utilization + \- Allocated, Cluster, CPUCount, Down, Idle, Overcommited, + PlannedDown, Reported, Reserved + +\fBJob\fP + SizesByAccount + \- Account, Cluster + SizesByWckey + \- Wckey, Cluster + +\fBUser\fP + TopUsage + \- Account, Cluster, Login, Proper, Used + +.TP .SH "EXAMPLES" -.eo -.br -> sreport job sizes -.br -> sreport cluster utilization -.br -> sreport user top -.ec +.TP +\fBsreport job sizesbyaccount\fP +.TP +\fBreport cluster utilization\fP +.TP +\fBsreport user top\fP +.TP +\fBsreport job sizesbyaccount All_Clusters users=gore1 account=environ PrintJobCount\fP +Report number of jobs by user gore1 within the environ account +.TP +\fBsreport cluster AccountUtilizationByUser cluster=zeus user=gore1 start=2/23/08 end=2/24/09 format=Accounts,Cluster,CPU_Count,Login,Proper,Used\fP +Report cluster account utilization with the specified fields during +the specified 24 hour day of February 23, 2009, by user gore1 +.TP +\fBsreport cluster AccountUtilizationByUser cluster=zeus accounts=lc start=2/23/08 end=2/24/09\fP +Report cluster account utilization by user in the LC account on +cluster zeus +.TP +\fBsreport user topusage start=2/16/09 end=2/23/09 \-t percent account=lc\fP +Report top usage in percent of the lc account during the specified week +.TP + .SH "COPYING" Copyright (C) 2008 Lawrence Livermore National Security. diff -Nru slurm-llnl-1.3.13/doc/man/man1/srun.1 slurm-llnl-1.3.15/doc/man/man1/srun.1 --- slurm-llnl-1.3.13/doc/man/man1/srun.1 2008-12-31 21:17:53.000000000 +0100 +++ slurm-llnl-1.3.15/doc/man/man1/srun.1 2009-03-17 19:50:45.000000000 +0100 @@ -967,7 +967,12 @@ .TP \fB\-\-blrts\-image\fR=\fIpath\fR -Path to blrts image for bluegene block. +Path to blrts image for bluegene block. BGL only. +Default from \fIblugene.conf\fR if not set. + +.TP +\fB\-\-cnload\-image\fR=\fIpath\fR +Path to compute node image for bluegene block. BGP only. Default from \fIblugene.conf\fR if not set. .TP @@ -977,6 +982,9 @@ If NAV, or if not set, then SLURM will try to fit a TORUS else MESH. You should not normally set this option. SLURM will normally allocate a TORUS if possible for a given geometry. +If running on a BGP system and wanting to run in HTC mode (only for 1 +midplane and below). You can use HTC_S for SMP, HTC_D for Dual, HTC_V +for virtual node mode, and HTC_L for Linux mode. .TP \fB\-g\fR, \fB\-\-geometry\fR=\fIXxYxZ\fR @@ -987,8 +995,13 @@ Blue Gene). .TP +\fB\-\-ioload\-image\fR=\fIpath\fR +Path to io image for bluegene block. BGP only. +Default from \fIblugene.conf\fR if not set. + +.TP \fB\-\-linux\-image\fR=\fIpath\fR -Path to linux image for bluegene block. +Path to linux image for bluegene block. BGL only. Default from \fIblugene.conf\fR if not set. .TP @@ -1004,7 +1017,7 @@ .TP \fB\-\-ramdisk\-image\fR=\fIpath\fR -Path to ramdisk image for bluegene block. +Path to ramdisk image for bluegene block. BGL only. Default from \fIblugene.conf\fR if not set. .TP diff -Nru slurm-llnl-1.3.13/doc/man/man5/bluegene.conf.5 slurm-llnl-1.3.15/doc/man/man5/bluegene.conf.5 --- slurm-llnl-1.3.13/doc/man/man5/bluegene.conf.5 2008-11-10 20:17:20.000000000 +0100 +++ slurm-llnl-1.3.15/doc/man/man5/bluegene.conf.5 2009-02-11 18:57:20.000000000 +0100 @@ -17,8 +17,10 @@ SLURM daemons, daemon receipt of the SIGHUP signal, or execution of the command "scontrol reconfigure" unless otherwise noted. .LP -The overall configuration parameters available include: +There are some differences between Bluegene/L and Bluegene/P in respects to the contents of the bluegene.conf file. + +.SH "The Bluegene/L specific options are:" .TP \fBAltBlrtsImage\fR Alternative BlrtsImage. This is an optional field only used for @@ -36,16 +38,59 @@ groups. You can but as many alternative images as you want in the conf file. .TP -\fBAltMloaderImage\fR -Alternative MloaderImage. This is an optional field only used for +\fBAltRamDiskImage\fR +Alternative RamDiskImage. This is an optional field only used for mulitple images on a system and should be followed by a Groups= with the user groups allowed to use this image (i.e. Groups=da,jette) if Groups= is not stated then this image will be able to be used by all groups. You can but as many alternative images as you want in the conf file. .TP -\fBAltRamDiskImage\fR -Alternative RamDiskImage. This is an optional field only used for +\fBBlrtsImage\fR +BlrtsImage used for creation of all bgblocks. +There is no default value and this must be specified. + +.TP +\fBLinuxImage\fR +LinuxImage used for creation of all bgblocks. +There is no default value and this must be specified. + +.TP +\fBRamDiskImage\fR +RamDiskImage used for creation of all bgblocks. +There is no default value and this must be specified. + +.SH "The Bluegene/P specific options are:" +.TP +\fBAltCnloadImage\fR +Alternative CnloadImage. This is an optional field only used for +mulitple images on a system and should be followed by a Groups= with +the user groups allowed to use this image (i.e. Groups=da,jette) if +Groups= is not stated then this image will be able to be used by all +groups. You can but as many alternative images as you want in the conf file. + +.TP +\fBAltIoloadImage\fR +Alternative IoloadImage. This is an optional field only used for +mulitple images on a system and should be followed by a Groups= with +the user groups allowed to use this image (i.e. Groups=da,jette) if +Groups= is not stated then this image will be able to be used by all +groups. You can but as many alternative images as you want in the conf file. + +.TP +\fBCnloadImage\fR +CnloadImage used for creation of all bgblocks. +There is no default value and this must be specified. + +.TP +\fBIoloadImage\fR +IoloadImage used for creation of all bgblocks. +There is no default value and this must be specified. + +.SH "All options below are common on all Bluegene systems:" +.TP +\fBAltMloaderImage\fR +Alternative MloaderImage. This is an optional field only used for mulitple images on a system and should be followed by a Groups= with the user groups allowed to use this image (i.e. Groups=da,jette) if Groups= is not stated then this image will be able to be used by all @@ -58,11 +103,6 @@ is usually 512) .TP -\fBBlrtsImage\fR -BlrtsImage used for creation of all bgblocks. -There is no default value and this must be specified. - -.TP \fBBridgeAPILogFile\fR Fully qualified pathname of a into which the Bridge API logs are to be written. @@ -86,6 +126,11 @@ .RE .TP +\fBDenyPassthrough\fR +Specify which dimsions you do not want to allow pass throughs. Valid options are X, Y, Z or all. +example: If you don't want to allow passthroughs in the X and Y diminsions you would specify DenyPassthrough=X,Y + +.TP \fBLayoutMode\fR Describes how SLURM should create bgblocks. .RS @@ -105,11 +150,6 @@ .RE .TP -\fBLinuxImage\fR -LinuxImage used for creation of all bgblocks. -There is no default value and this must be specified. - -.TP \fBMloaderImage\fR MloaderImage used for creation of all bgblocks. There is no default value and this must be specified. @@ -132,11 +172,6 @@ For bgp IO rich systems 32 is the value that should be used to create small blocks since you can only have 2 ionodes per nodecard instead of 4 like on bgl. -.TP -\fBRamDiskImage\fR -RamDiskImage used for creation of all bgblocks. -There is no default value and this must be specified. - .LP Each bgblock is defined by the base partitions used to construct it. Ordering is very important for laying out switch wires. Please create diff -Nru slurm-llnl-1.3.13/doc/man/man5/slurm.conf.5 slurm-llnl-1.3.15/doc/man/man5/slurm.conf.5 --- slurm-llnl-1.3.13/doc/man/man5/slurm.conf.5 2008-12-30 18:28:15.000000000 +0100 +++ slurm-llnl-1.3.15/doc/man/man5/slurm.conf.5 2009-03-17 16:46:32.000000000 +0100 @@ -217,7 +217,7 @@ The slurmctld daemon must be restarted for a change in \fBCryptoType\fR to take effect. Acceptable values at present include "crypto/munge" and "crypto/openssl". -The default value is "crypto/munge". +The default value is "crypto/openssl". .TP \fBDefMemPerCPU\fR @@ -379,7 +379,11 @@ \fBHealthCheckProgram\fR Fully qualified pathname of a script to execute as user root periodically on all compute nodes that are not in the DOWN state. This may be used to -verify the node is fully operational and DRAIN the it otherwise. +verify the node is fully operational and DRAIN the node or send email +if a problem is detected. +Any action to be taken must be explicitly performed by the program +(e.g. execute "scontrol update NodeName=foo State=drain Reason=tmp_file_system_full" +to drain a node). The interval is controlled using the \fBHealthCheckInterval\fR parameter. Note that the \fBHealthCheckProgram\fR will be executed at the same time on all nodes to minimize its impact upon parallel programs. @@ -1222,6 +1226,9 @@ Standard output from this program of the form "export NAME=value" will be used to set environment variables for the task being spawned. +Standard output from this program of the form +"print ..." will cause that line (without the leading "print ") +to be printed to the job's standard output. The order of task prolog/epilog execution is as follows: .RS .TP @@ -1641,7 +1648,7 @@ .TP \fBMaxNodes\fR -Maximum count of nodes (or base partitions for BlueGene systems) which +Maximum count of nodes (c\-nodes for BlueGene systems) which may be allocated to any single job. The default value is "UNLIMITED", which is represented internally as \-1. This limit does not apply to jobs executed by SlurmUser or user root. diff -Nru slurm-llnl-1.3.13/doc/man/man5/slurmdbd.conf.5 slurm-llnl-1.3.15/doc/man/man5/slurmdbd.conf.5 --- slurm-llnl-1.3.13/doc/man/man5/slurmdbd.conf.5 2008-12-09 23:22:06.000000000 +0100 +++ slurm-llnl-1.3.15/doc/man/man5/slurmdbd.conf.5 2009-02-10 18:32:01.000000000 +0100 @@ -1,4 +1,4 @@ -.TH "slurmdbd.conf" "5" "June 2008" "slurmdbd.conf 1.3" "Slurm configuration file" +.TH "slurmdbd.conf" "5" "Feb 2009" "slurmdbd.conf 1.3" "Slurm configuration file" .SH "NAME" slurmdbd.conf \- Slurm Database Daemon (SlurmDBD) configuration file @@ -22,18 +22,38 @@ The overall configuration parameters available include: .TP -\fBArchiveAge\fR -Move data over this age out of the database to an archive. -The time is a numeric value and is a number of days. -If zero, then never archive the data. -The default value is zero. +\fBArchiveDir\fR +If ArchiveScript is not set the slurmdbd will generate a text file that can be +read in anytime with sacctmgr load filename. This directory is where the +file will be placed archive has ran. Default is /tmp. + +.TP +\fBArchiveJobs\fR +Boolean, yes to archive job data, no other wise. Default is no. .TP \fBArchiveScript\fR This script is executed periodically in order to transfer accounting records out of the database into an archive. The script is executed -with a single argument, the value of \fBArchiveTime\fR as described -below. +with a no arguments, The following environment variables are set. +.RS +.TP +\fBSLURM_ARCHIVE_STEPS\fR +1 for archive steps 0 otherwise. +.TP +\fBSLURM_ARCHIVE_LAST_STEP\fR +Time of last step start to archive. +.TP +\fBSLURM_ARCHIVE_JOBS\fR +1 for achive jobs 0 otherwise. +.TP +\fBSLURM_ARCHIVE_LAST_JOB\fR +Time of last job submit to archive. +.RE + +.TP +\fBArchiveSteps\fR +Boolean, yes to archive step data, no other wise. Default is no. .TP \fBAuthInfo\fR @@ -99,12 +119,16 @@ The default value is 3. .TP +\fBDefaultQOS\fR +When adding a new cluster this will be used as the qos for the cluster +unless something is explicitly set by the admin with the create. + +.TP \fBJobPurge\fR Individual job records over this age are purged from the database. Aggregated information will be preserved indefinitely. -The time is a numeric value and is a number of days. -If zero, then job records are never purged. -The default value is 360 days. +The time is a numeric value and is a number of months. +If zero (default), then job records are never purged. .TP \fBLogFile\fR @@ -173,9 +197,8 @@ \fBStepPurge\fR Individual job step records over this age are purged from the database. Aggregated information will be preserved indefinitely. -The time is a numeric value and is a number of days. -If zero, then job step records are never purged. -The default value is 30 days. +The time is a numeric value and is a number of months. +If zero (default), then job step records are never purged. .TP \fBStorageHost\fR @@ -237,9 +260,11 @@ .br # .br -ArchiveAge=365 # keep 1 year of data online +ArchiveJobs=yes +.br +ArchiveSteps=no .br -ArchiveScript=/usr/sbin/slurm.dbd.archive +#ArchiveScript=/usr/sbin/slurm.dbd.archive .br AuthInfo=/var/run/munge/munge.socket.2 .br @@ -249,9 +274,9 @@ .br DebugLevel=4 .br -JobPurge=90 +JobPurge=12 .br -StepPurge=30 +StepPurge=1 .br LogFile=/var/log/slurmdbd.log .br diff -Nru slurm-llnl-1.3.13/doc/man/man8/spank.8 slurm-llnl-1.3.15/doc/man/man8/spank.8 --- slurm-llnl-1.3.13/doc/man/man8/spank.8 2008-11-14 21:34:15.000000000 +0100 +++ slurm-llnl-1.3.15/doc/man/man8/spank.8 2009-02-24 17:30:23.000000000 +0100 @@ -1,4 +1,4 @@ -.TH "SPANK" "8" "Jul 2008" "SPANK" "SLURM plug\-in architecture for Node and job (K)control" +.TH "SPANK" "8" "February 2009" "SPANK" "SLURM plug\-in architecture for Node and job (K)control" .SH "NAME" \fBSPANK\fR \- SLURM Plug\-in Architecture for Node and job (K)control @@ -20,8 +20,8 @@ .SH "SPANK PLUGINS" \fBSPANK\fR plugins are loaded in two separate contexts during a -\fBSLURM\fR job. In "local" context, the plugin is loaded by \fBsrun\fR, -\fBsbatch\fR or other \fBSLURM\fR user interface. +\fBSLURM\fR job. In "local" context, the plugin is loaded by \fBsrun\fR +(NOTE: the \fBsalloc\fR and \fBsbatch\fR commands do not support \fBSPANK\fR). In local context, options provided by plugins are read by \fBSPANK\fR, and these options are presented to the user. In "remote" context, the plugin is loaded on a compute node of the job, @@ -47,12 +47,11 @@ action in \fBslurm_spank_init_post_opt\fR if necessary. .TP \fBslurm_spank_local_user_init\fR -Called in local (\fBsrun\fR or \fBsbatch\fR) context only after all +Called in local (\fBsrun\fR) context only after all options have been processed. This is called after the job ID and step IDs are available. This happens in \fBsrun\fR after the allocation is made, but before tasks are launched. -This happens in \fBsbatch\fR after the job is submitted. .TP \fBslurm_spank_user_init\fR Called after privileges are temporarily dropped. (remote context only) @@ -72,7 +71,7 @@ .TP \fBslurm_spank_exit\fR Called once just before \fBslurmstepd\fR exits in remote context. -In local context, called before \fBsrun\fR or \fBsbatch\fR exits. +In local context, called before \fBsrun\fR exits. .LP All of these functions have the same prototype, for example: .nf @@ -119,9 +118,7 @@ User id for running job. (uid_t *) is third arg of \fBspank_get_item\fR .TP \fBS_JOB_STEPID\fR -Job step id for running job. (uint32_t *) is third arg of \fBspank_get_item\fR -For batch jobs (initiated by \fBsbatch\fR), the step id will be -\fBSLURM_BATCH_SCRIPT\fR as defined in the \fBslurm.h\fR file. +Job step id for running job. (uint32_t *) is third arg of \fBspank_get_item\fR. .TP \fBS_TASK_EXIT_STATUS\fR Exit status for exited task. Only valid from \fBslurm_spank_task_exit\fR. @@ -163,8 +160,7 @@ .LP SPANK plugins also have an interface through which they may define and implement extra job options. These options are made available to -the user through SLURM commands such as \fBsrun\fR(1) or -\fBsbatch\fR(1), and if the +the user through SLURM commands such as \fBsrun\fR(1), and if the option is specified, its value is forwarded and registered with the plugin on the remote side. In this way, \fBSPANK\fR plugins may dynamically provide new options and functionality to SLURM. @@ -217,7 +213,7 @@ Where \fIval\fR is the value of the \fIval\fR field in the \fBspank_option\fR struct, \fIoptarg\fR is the supplied argument if applicable, and \fIremote\fR is 0 if the function is being called from the "local" host -(e.g. \fBsrun\fR or \fBsbatch\fR) or 1 from the "remote" host (\fBslurmd\fR). +(e.g. \fBsrun\fR) or 1 from the "remote" host (\fBslurmd\fR). .LP There are two methods by which the plugin can register these options with SLURM. The simplest method is for the plugin to define an array @@ -505,4 +501,4 @@ \fB/usr/include/slurm/spank.h\fR \- SPANK header file. .SH "SEE ALSO" .LP -\fBsbatch\fR(1), \fBsrun\fR(1), \fBslurm.conf\fR(5) +\fBsrun\fR(1), \fBslurm.conf\fR(5) diff -Nru slurm-llnl-1.3.13/etc/bluegene.conf.example slurm-llnl-1.3.15/etc/bluegene.conf.example --- slurm-llnl-1.3.13/etc/bluegene.conf.example 2008-04-18 20:12:30.000000000 +0200 +++ slurm-llnl-1.3.15/etc/bluegene.conf.example 2009-02-11 23:42:39.000000000 +0100 @@ -73,6 +73,8 @@ BridgeAPILogFile=/var/log/slurm/bridgeapi.log BridgeAPIVerbose=0 +#DenyPassthrough=X,Y,Z + ############################################################################### # Define the static/overlap partitions (bgblocks) # diff -Nru slurm-llnl-1.3.13/etc/slurmdbd.conf.example slurm-llnl-1.3.15/etc/slurmdbd.conf.example --- slurm-llnl-1.3.13/etc/slurmdbd.conf.example 1970-01-01 01:00:00.000000000 +0100 +++ slurm-llnl-1.3.15/etc/slurmdbd.conf.example 2009-02-10 18:32:01.000000000 +0100 @@ -0,0 +1,39 @@ +# +# Example slurmdbd.conf file. +# +# See the slurmdbd.conf man page for more information. +# +# Archive info +#ArchiveJobs=yes +#ArchiveDir="/tmp" +#ArchiveSteps=yes +#ArchiveScript= +#JobPurge=12 +#StepPurge=1 +# +# Authentication info +AuthType=auth/munge +#AuthInfo=/var/run/munge/munge.socket.2 +# +# slurmDBD info +DbdAddr=localhost +DbdHost=localhost +#DbdPort=7031 +SlurmUser=slurm +#MessageTimeout=300 +DebugLevel=4 +#DefaultQOS=normal,standby +LogFile=/var/log/slurm/slurmdbd.log +PidFile=/var/run/slurmdbd.pid +#PluginDir=/usr/lib/slurm +#PrivateData=accounts,users,usage,jobs +#TrackWCKey=yes +# +# Database info +StorageType=accounting_storage/mysql +#StorageHost=localhost +#StoragePort=1234 +StoragePassword=password +StorageUser=slurm +#StorageLoc=slurm_acct_db + diff -Nru slurm-llnl-1.3.13/META slurm-llnl-1.3.15/META --- slurm-llnl-1.3.13/META 2009-01-13 18:36:19.000000000 +0100 +++ slurm-llnl-1.3.15/META 2009-04-14 19:26:36.000000000 +0200 @@ -3,9 +3,9 @@ Api_revision: 0 Major: 1 Meta: 1 - Micro: 13 + Micro: 15 Minor: 3 Name: slurm Release: 1 Release_tags: dist - Version: 1.3.13 + Version: 1.3.15 diff -Nru slurm-llnl-1.3.13/NEWS slurm-llnl-1.3.15/NEWS --- slurm-llnl-1.3.13/NEWS 2009-01-13 18:34:26.000000000 +0100 +++ slurm-llnl-1.3.15/NEWS 2009-04-10 21:25:52.000000000 +0200 @@ -1,6 +1,81 @@ This file describes changes in recent versions of SLURM. It primarily documents those changes that are of interest to users and admins. +* Changes in SLURM 1.3.16 +========================= + +* Changes in SLURM 1.3.15 +========================= + -- Fix bug in squeue command with sort on job name ("-S j" option) for jobs + that lack a name. Previously generated an invalid memory reference. + -- Permit the TaskProlog to write to the job's standard output by writing + a line containing the prefix "print " to it's standard output. + -- Fix for making the slurmdbd agent thread start up correctly when + stopped and then started again. + -- Add squeue option to report jobs by account (-U or --account). Patch from + Par Andersson, National Supercomputer Centre, Sweden. + -- Add -DNUMA_VERSION1_COMPATIBILITY to Makefile CFLAGS for proper behavior + when building with NUMA version 2 APIs. + -- BLUEGENE - slurm works on a BGP system. + -- BLUEGENE - slurm handles HTC blocks + -- BLUEGENE - Added option DenyPassthrough in the bluegene.conf. Can be set + to any combination of X,Y,Z to not allow passthroughs when running in + dynamic layout mode. + -- Fix bug in logic to remove a job's dependency, could result in abort. + -- Add new error message to sched/wiki and sched/wiki2 (Maui and Moab) for + STARTJOB request: "TASKLIST includes non-responsive nodes". + -- Fix bug in task layout for heterogeneous nodes and srun --exclusive + option. + -- Fix bug in select/linear when used with sched/gang that can result in a + job's required or excluded node specification being ignored. + -- Add logic to handle message connect timeouts (timed-out.patch from + Chuck Clouston, Bull). + -- BLUEGENE - CFLAGS=-m64 is no longer required in configure + -- Update python-hostlist code from Kent Engström (NSC) to v1.5 + - Add hostgrep utility to search for lines matching a hostlist. + - Make each "-" on the command line count as one hostlist argument. + If multiple hostslists are given on stdin they are combined to a + union hostlist before being used in the way requested by the + options. + -- When using -j option in sacct no user restriction will applied unless + specified with the -u option. + -- For sched/wiki and sched/wiki2, change logging of wiki message traffic + from debug() to debug2(). Only seen if SlurmctldDebug is configured to + 6 or higher. + -- Significant speed up for association based reports in sreport + -- BLUEGENE - fix for checking if job can run with downed nodes. Previously + sbatch etc would tell you node configuration not available now jobs are + accepted but held until nodes are back up. + -- Fix in accounting so if any nodes are removed from the system when they + were previously down will be recorded correctly. + -- For sched/wiki2 (Moab), add flag to note if job is restartable and + prevent deadlock of job requeue fails. + -- Modify squeue to return non-zero exit code on failure. Patch from + Par Andersson (NSC). + -- Correct logic in select/cons_res to allocate a job the maximum node + count from a range rather than minimum (e.g. "sbatch -N1-4 my.sh"). + -- In accounting_storage/filetxt and accounting_storage/pgsql fix + possible invalid memory reference when a job lacks a name. + -- Give srun command an exit code of 1 if the prolog fails. + -- BLUEGENE - allows for checking nodecard states in the system instead + of midplane state so as to not down an entire midplane if you don't + have to. + -- BLUEGENE - fix creation of MESH blocks + -- BLUEGENE - on job cancellation we call jm_cancel_job and then wait until + the system cleans up the job. Before we would send a SIGKILL right + at the beginning. + -- BLUEGENE - if a user specifies a node count that can not be met the job + will be refused instead of before the plugin would search for the next + larger size that could be created. This prevents users asking for + things that can't be created, and then getting something back they might + not be expecting. + +* Changes in SLURM 1.3.14 +========================= + -- SECURITY BUG: Fix in sbcast logic that permits users to write files based + upon supplimental groups of the slurmd daemon. Similar logic for event + triggers if slurmctld is run as user root (not typical). + * Changes in SLURM 1.3.13 ========================= -- Added ability for slurmdbd to archive and purge step and/or job records. @@ -17,7 +92,7 @@ associated with that partition rather than blocking all partitions with any overlapping nodes). -- Correct logic to log in a job's stderr that it was "CANCELLED DUE TO - NODE FAILURE" rather than just "CANCLLED". + NODE FAILURE" rather than just "CANCELLED". -- Fix to crypto/openssl plugin that could result in job launch requests being spoofed through the use of an improperly formed credential. This bug could permit a user to launch tasks on compute nodes not allocated for @@ -160,7 +235,7 @@ given user with a given name will execute with this dependency type. From Matthieu Hautreux, CEA. -- Updated contribs/python/hostlist to version 1.3: See "CHANGES" file in - that directory for details. From Kent Engstrom, NSC. + that directory for details. From Kent Engström, NSC. -- Add SLURM_JOB_NAME environment variable for jobs submitted using sbatch. In order to prevent the job steps from all having the same name as the batch job that spawned them, the SLURM_JOB_NAME environment variable is @@ -272,6 +347,7 @@ slurmctld daemon and leave the slurmd daemons running. -- Do not require JobCredentialPrivateKey or JobCredentialPublicCertificate in slurm.conf if using CryptoType=crypto/munge. + -- Remove SPANK support from sbatch. * Changes in SLURM 1.3.6 ======================== @@ -3678,4 +3754,4 @@ -- Change directory to /tmp in slurmd if daemonizing. -- Logfiles are reopened on reconfigure. -$Id: NEWS 16197 2009-01-13 17:34:26Z jette $ +$Id: NEWS 17225 2009-04-10 19:25:52Z da $ diff -Nru slurm-llnl-1.3.13/slurm/slurm_errno.h slurm-llnl-1.3.15/slurm/slurm_errno.h --- slurm-llnl-1.3.13/slurm/slurm_errno.h 2008-06-23 19:36:46.000000000 +0200 +++ slurm-llnl-1.3.15/slurm/slurm_errno.h 2009-02-03 00:02:44.000000000 +0100 @@ -159,6 +159,7 @@ ESLURM_NEED_RESTART, ESLURM_ACCOUNTING_POLICY, ESLURM_INVALID_TIME_LIMIT, + ESLURM_INVALID_WCKEY, /* switch specific error codes, specific values defined in plugin module */ ESLURM_SWITCH_MIN = 3000, diff -Nru slurm-llnl-1.3.13/slurm/slurm.h.in slurm-llnl-1.3.15/slurm/slurm.h.in --- slurm-llnl-1.3.13/slurm/slurm.h.in 2008-12-04 22:39:19.000000000 +0100 +++ slurm-llnl-1.3.15/slurm/slurm.h.in 2009-04-07 18:56:32.000000000 +0200 @@ -252,7 +252,11 @@ SELECT_MESH, /* nodes wired in mesh */ SELECT_TORUS, /* nodes wired in torus */ SELECT_NAV, /* nodes wired in torus else mesh */ - SELECT_SMALL /* nodes in a small partition */ + SELECT_SMALL, /* nodes in a small partition */ + SELECT_HTC_S, /* nodes in a htc running SMP mode */ + SELECT_HTC_D, /* nodes in a htc running Dual mode */ + SELECT_HTC_V, /* nodes in a htc running VN mode */ + SELECT_HTC_L /* nodes in a htc running in Linux mode */ }; enum node_use_type { diff -Nru slurm-llnl-1.3.13/slurm.spec slurm-llnl-1.3.15/slurm.spec --- slurm-llnl-1.3.13/slurm.spec 2008-12-30 00:24:59.000000000 +0100 +++ slurm-llnl-1.3.15/slurm.spec 2009-03-24 17:33:55.000000000 +0100 @@ -1,4 +1,4 @@ -# $Id: slurm.spec 16103 2008-12-29 23:24:59Z jette $ +# $Id: slurm.spec 16983 2009-03-24 16:33:55Z da $ # # Note that this package is not relocatable @@ -73,14 +73,14 @@ %endif Name: slurm -Version: 1.3.13 +Version: 1.3.15 Release: 1%{?dist} Summary: Simple Linux Utility for Resource Management License: GPL Group: System Environment/Base -Source: slurm-1.3.13.tar.bz2 +Source: slurm-1.3.15.tar.bz2 BuildRoot: %{_tmppath}/%{name}-%{version}-%{release} URL: https://computing.llnl.gov/linux/slurm/ @@ -230,10 +230,17 @@ %package torque Summary: Torque/PBS wrappers for transitition from Torque/PBS to SLURM. Group: Development/System -Requires: slurm +Requires: slurm-perlapi %description torque Torque wrapper scripts used for helping migrate from Torque/PBS to SLURM. +%package slurmdb-direct +Summary: Wrappers to write directly to the slurmdb. +Group: Development/System +Requires: slurm-perlapi +%description slurmdb-direct +Wrappers to write directly to the slurmdb. + %if %{slurm_with aix} %package aix-federation Summary: SLURM interfaces to IBM AIX and Federation switch. @@ -258,7 +265,7 @@ ############################################################################# %prep -%setup -n slurm-1.3.13 +%setup -n slurm-1.3.15 %build %configure --program-prefix=%{?_program_prefix:%{_program_prefix}} \ @@ -287,6 +294,7 @@ install -D -m755 etc/init.d.slurmdbd $RPM_BUILD_ROOT/etc/init.d/slurmdbd fi install -D -m644 etc/slurm.conf.example ${RPM_BUILD_ROOT}%{_sysconfdir}/slurm.conf.example +install -D -m644 etc/slurmdbd.conf.example ${RPM_BUILD_ROOT}%{_sysconfdir}/slurmdbd.conf.example install -D -m755 etc/slurm.epilog.clean ${RPM_BUILD_ROOT}%{_sysconfdir}/slurm.epilog.clean # Delete unpackaged files: @@ -426,6 +434,7 @@ %{_sbindir}/slurmdbd %{_mandir}/man5/slurmdbd.* %{_mandir}/man8/slurmdbd.* +%config %{_sysconfdir}/slurmdbd.conf.example ############################################################################# %files -f plugins.files plugins @@ -480,6 +489,12 @@ %{_bindir}/mpiexec ############################################################################# +%files slurmdb-direct +%defattr(-,root,root) +%config (noreplace) %{_perldir}/config.slurmdb.pl +%{_sbindir}/moab_2_slurmdb +############################################################################# + %if %{slurm_with aix} %files aix-federation %defattr(-,root,root) diff -Nru slurm-llnl-1.3.13/src/api/job_info.c slurm-llnl-1.3.15/src/api/job_info.c --- slurm-llnl-1.3.13/src/api/job_info.c 2008-12-04 22:47:09.000000000 +0100 +++ slurm-llnl-1.3.15/src/api/job_info.c 2009-03-17 22:38:37.000000000 +0100 @@ -538,8 +538,13 @@ xstrcat(out, " "); else xstrcat(out, "\n "); +#ifdef HAVE_BGL snprintf(tmp_line, sizeof(tmp_line), "LinuxImage=%s", select_buf); +#else + snprintf(tmp_line, sizeof(tmp_line), + "CnloadImage=%s", select_buf); +#endif xstrcat(out, tmp_line); } /****** Line 22 (optional) ******/ @@ -564,8 +569,13 @@ xstrcat(out, " "); else xstrcat(out, "\n "); +#ifdef HAVE_BGL snprintf(tmp_line, sizeof(tmp_line), "RamDiskImage=%s", select_buf); +#else + snprintf(tmp_line, sizeof(tmp_line), + "IoloadImage=%s", select_buf); +#endif xstrcat(out, tmp_line); } xstrcat(out, "\n\n"); diff -Nru slurm-llnl-1.3.13/src/common/assoc_mgr.c slurm-llnl-1.3.15/src/common/assoc_mgr.c --- slurm-llnl-1.3.13/src/common/assoc_mgr.c 2008-12-12 22:35:56.000000000 +0100 +++ slurm-llnl-1.3.15/src/common/assoc_mgr.c 2009-03-24 21:32:18.000000000 +0100 @@ -339,7 +339,7 @@ if(local_cluster_name) { assoc_q.cluster_list = list_create(NULL); list_append(assoc_q.cluster_list, local_cluster_name); - } else if(enforce && !slurmdbd_conf) { + } else if((enforce & ACCOUNTING_ENFORCE_ASSOCS) && !slurmdbd_conf) { error("_get_local_association_list: " "no cluster name here going to get " "all associations."); @@ -356,9 +356,10 @@ if(!local_association_list) { /* create list so we don't keep calling this if there isn't anything there */ - local_association_list = list_create(NULL); + local_association_list = + list_create(destroy_acct_association_rec); slurm_mutex_unlock(&assoc_mgr_association_lock); - if(enforce) { + if(enforce & ACCOUNTING_ENFORCE_ASSOCS) { error("_get_local_association_list: " "no list was made."); return SLURM_ERROR; @@ -387,7 +388,7 @@ if(!local_qos_list) { slurm_mutex_unlock(&local_qos_lock); - if(enforce) { + if(enforce & ACCOUNTING_ENFORCE_ASSOCS) { error("_get_local_qos_list: no list was made."); return SLURM_ERROR; } else { @@ -414,7 +415,7 @@ if(!local_user_list) { slurm_mutex_unlock(&local_user_lock); - if(enforce) { + if(enforce & ACCOUNTING_ENFORCE_ASSOCS) { error("_get_local_user_list: " "no list was made."); return SLURM_ERROR; @@ -443,7 +444,7 @@ if(local_cluster_name) { wckey_q.cluster_list = list_create(NULL); list_append(wckey_q.cluster_list, local_cluster_name); - } else if(enforce && !slurmdbd_conf) { + } else if((enforce & ACCOUNTING_ENFORCE_WCKEYS) && !slurmdbd_conf) { error("_get_local_wckey_list: " "no cluster name here going to get " "all wckeys."); @@ -460,9 +461,9 @@ if(!assoc_mgr_wckey_list) { /* create list so we don't keep calling this if there isn't anything there */ - assoc_mgr_wckey_list = list_create(NULL); + assoc_mgr_wckey_list = list_create(destroy_acct_wckey_rec); slurm_mutex_unlock(&assoc_mgr_wckey_lock); - if(enforce) { + if(enforce & ACCOUNTING_ENFORCE_WCKEYS) { error("_get_local_wckey_list: " "no list was made."); return SLURM_ERROR; @@ -494,7 +495,7 @@ if(local_cluster_name) { assoc_q.cluster_list = list_create(NULL); list_append(assoc_q.cluster_list, local_cluster_name); - } else if(enforce && !slurmdbd_conf) { + } else if((enforce & ACCOUNTING_ENFORCE_ASSOCS) && !slurmdbd_conf) { error("_refresh_local_association_list: " "no cluster name here going to get " "all associations."); @@ -634,7 +635,7 @@ if(local_cluster_name) { wckey_q.cluster_list = list_create(NULL); list_append(wckey_q.cluster_list, local_cluster_name); - } else if(enforce && !slurmdbd_conf) { + } else if((enforce & ACCOUNTING_ENFORCE_WCKEYS) && !slurmdbd_conf) { error("_refresh_local_wckey_list: " "no cluster name here going to get " "all wckeys."); @@ -745,7 +746,7 @@ return SLURM_ERROR; } if((!local_association_list || !list_count(local_association_list)) - && !enforce) + && !(enforce & ACCOUNTING_ENFORCE_ASSOCS)) return SLURM_SUCCESS; if(!assoc->id) { @@ -753,7 +754,7 @@ acct_user_rec_t user; if(assoc->uid == (uint32_t)NO_VAL) { - if(enforce) { + if(enforce & ACCOUNTING_ENFORCE_ASSOCS) { error("get_assoc_id: " "Not enough info to " "get an association"); @@ -766,7 +767,7 @@ user.uid = assoc->uid; if(assoc_mgr_fill_in_user(db_conn, &user, enforce) == SLURM_ERROR) { - if(enforce) + if(enforce & ACCOUNTING_ENFORCE_ASSOCS) return SLURM_ERROR; else { return SLURM_SUCCESS; @@ -835,7 +836,7 @@ if(!ret_assoc) { slurm_mutex_unlock(&assoc_mgr_association_lock); - if(enforce) + if(enforce & ACCOUNTING_ENFORCE_ASSOCS) return SLURM_ERROR; else return SLURM_SUCCESS; @@ -895,7 +896,8 @@ if(_get_local_user_list(db_conn, enforce) == SLURM_ERROR) return SLURM_ERROR; - if((!local_user_list || !list_count(local_user_list)) && !enforce) + if((!local_user_list || !list_count(local_user_list)) + && !(enforce & ACCOUNTING_ENFORCE_ASSOCS)) return SLURM_SUCCESS; slurm_mutex_lock(&local_user_lock); @@ -940,7 +942,7 @@ return SLURM_ERROR; } if((!assoc_mgr_wckey_list || !list_count(assoc_mgr_wckey_list)) - && !enforce) + && !(enforce & ACCOUNTING_ENFORCE_WCKEYS)) return SLURM_SUCCESS; if(!wckey->id) { @@ -948,7 +950,7 @@ acct_user_rec_t user; if(wckey->uid == (uint32_t)NO_VAL && !wckey->user) { - if(enforce) { + if(enforce & ACCOUNTING_ENFORCE_WCKEYS) { error("get_wckey_id: " "Not enough info to " "get an wckey"); @@ -962,7 +964,7 @@ user.name = wckey->user; if(assoc_mgr_fill_in_user(db_conn, &user, enforce) == SLURM_ERROR) { - if(enforce) + if(enforce & ACCOUNTING_ENFORCE_WCKEYS) return SLURM_ERROR; else { return SLURM_SUCCESS; @@ -972,7 +974,7 @@ wckey->user = user.name; wckey->name = user.default_wckey; } else if(wckey->uid == (uint32_t)NO_VAL && !wckey->user) { - if(enforce) { + if(enforce & ACCOUNTING_ENFORCE_WCKEYS) { error("get_wckey_id: " "Not enough info 2 to " "get an wckey"); @@ -1017,13 +1019,22 @@ wckey->name, found_wckey->name); continue; } - + /* only check for on the slurmdbd */ - if(!local_cluster_name && found_wckey->cluster - && strcasecmp(wckey->cluster, - found_wckey->cluster)) { - debug4("not the right cluster"); - continue; + if(!local_cluster_name) { + if(!wckey->cluster) { + error("No cluster name was given " + "to check against, " + "we need one to get a wckey."); + continue; + } + + if(found_wckey->cluster + && strcasecmp(wckey->cluster, + found_wckey->cluster)) { + debug4("not the right cluster"); + continue; + } } } ret_wckey = found_wckey; @@ -1033,7 +1044,7 @@ if(!ret_wckey) { slurm_mutex_unlock(&assoc_mgr_wckey_lock); - if(enforce) + if(enforce & ACCOUNTING_ENFORCE_WCKEYS) return SLURM_ERROR; else return SLURM_SUCCESS; @@ -1137,6 +1148,7 @@ ListIterator itr = NULL; int rc = SLURM_SUCCESS; int parents_changed = 0; + List remove_list = NULL; if(!local_association_list) return SLURM_SUCCESS; @@ -1276,9 +1288,19 @@ //rc = SLURM_ERROR; break; } - if (remove_assoc_notify) - remove_assoc_notify(rec); - list_delete_item(itr); + if(remove_assoc_notify) { + /* since there are some deadlock + issues while inside our lock here + we have to process a notify later + */ + if(!remove_list) + remove_list = list_create( + destroy_acct_association_rec); + list_remove(itr); + list_append(remove_list, rec); + } else + list_delete_item(itr); + break; default: break; @@ -1330,6 +1352,18 @@ list_iterator_destroy(itr); slurm_mutex_unlock(&assoc_mgr_association_lock); + + /* This needs to happen outside of the + assoc_mgr_association_lock */ + if(remove_list) { + itr = list_iterator_create(remove_list); + + while((rec = list_next(itr))) + remove_assoc_notify(rec); + + list_iterator_destroy(itr); + list_destroy(remove_list); + } return rc; } @@ -1620,7 +1654,7 @@ return SLURM_ERROR; if((!local_association_list || !list_count(local_association_list)) - && !enforce) + && !(enforce & ACCOUNTING_ENFORCE_ASSOCS)) return SLURM_SUCCESS; slurm_mutex_lock(&assoc_mgr_association_lock); @@ -1632,7 +1666,7 @@ list_iterator_destroy(itr); slurm_mutex_unlock(&assoc_mgr_association_lock); - if(found_assoc || !enforce) + if(found_assoc || !(enforce & ACCOUNTING_ENFORCE_ASSOCS)) return SLURM_SUCCESS; return SLURM_ERROR; @@ -1836,6 +1870,8 @@ break; } slurm_mutex_lock(&assoc_mgr_association_lock); + if(local_association_list) + list_destroy(local_association_list); local_association_list = msg->my_list; _post_association_list(local_association_list); debug("Recovered %u associations", @@ -1854,6 +1890,8 @@ break; } slurm_mutex_lock(&local_user_lock); + if(local_user_list) + list_destroy(local_user_list); local_user_list = msg->my_list; _post_user_list(local_user_list); debug("Recovered %u users", @@ -1872,6 +1910,8 @@ break; } slurm_mutex_lock(&local_qos_lock); + if(local_qos_list) + list_destroy(local_qos_list); local_qos_list = msg->my_list; debug("Recovered %u qos", list_count(local_qos_list)); @@ -1889,6 +1929,8 @@ break; } slurm_mutex_lock(&assoc_mgr_wckey_lock); + if(assoc_mgr_wckey_list) + list_destroy(assoc_mgr_wckey_list); assoc_mgr_wckey_list = msg->my_list; debug("Recovered %u wckeys", list_count(assoc_mgr_wckey_list)); diff -Nru slurm-llnl-1.3.13/src/common/checkpoint.c slurm-llnl-1.3.15/src/common/checkpoint.c --- slurm-llnl-1.3.13/src/common/checkpoint.c 2008-06-06 21:15:24.000000000 +0200 +++ slurm-llnl-1.3.15/src/common/checkpoint.c 2009-03-24 22:57:43.000000000 +0100 @@ -1,6 +1,6 @@ /*****************************************************************************\ * checkpoint.c - implementation-independent checkpoint functions - * $Id: checkpoint.c 14208 2008-06-06 19:15:24Z da $ + * $Id: checkpoint.c 17005 2009-03-24 21:57:43Z da $ ***************************************************************************** * Copyright (C) 2004 The Regents of the University of California. * Produced at Lawrence Livermore National Laboratory (cf, DISCLAIMER). @@ -119,13 +119,14 @@ static int _slurm_checkpoint_context_destroy( slurm_checkpoint_context_t c ) { + int rc = SLURM_SUCCESS; /* * Must check return code here because plugins might still * be loaded and active. */ if ( c->plugin_list ) { if ( plugrack_destroy( c->plugin_list ) != SLURM_SUCCESS ) { - return SLURM_ERROR; + rc = SLURM_ERROR; } } else { plugin_unload(c->cur_plugin); @@ -134,7 +135,7 @@ xfree( c->checkpoint_type ); xfree( c ); - return SLURM_SUCCESS; + return rc; } /* diff -Nru slurm-llnl-1.3.13/src/common/env.c slurm-llnl-1.3.15/src/common/env.c --- slurm-llnl-1.3.13/src/common/env.c 2008-11-12 00:06:36.000000000 +0100 +++ slurm-llnl-1.3.15/src/common/env.c 2009-02-05 18:48:11.000000000 +0100 @@ -267,7 +267,6 @@ int rc = SLURM_SUCCESS; char *dist = NULL; char *lllp_dist = NULL; - char *bgl_part_id = NULL; char addrbuf[INET_ADDRSTRLEN]; if (env == NULL) @@ -529,10 +528,22 @@ rc = SLURM_FAILURE; } +#ifdef HAVE_BG if(env->select_jobinfo) { + char *bgl_part_id = NULL; select_g_get_jobinfo(env->select_jobinfo, SELECT_DATA_BLOCK_ID, &bgl_part_id); if (bgl_part_id) { +#ifndef HAVE_BGL + uint16_t conn_type = (uint16_t)NO_VAL; + select_g_get_jobinfo(env->select_jobinfo, + SELECT_DATA_CONN_TYPE, &conn_type); + if(conn_type > SELECT_SMALL) { + if(setenvf(&env->env, + "SUBMIT_POOL", "%s", bgl_part_id)) + rc = SLURM_FAILURE; + } +#endif if(setenvf(&env->env, "MPIRUN_PARTITION", "%s", bgl_part_id)) rc = SLURM_FAILURE; @@ -541,15 +552,17 @@ rc = SLURM_FAILURE; if(setenvf(&env->env, "MPIRUN_NOALLOCATE", "%d", 1)) rc = SLURM_FAILURE; + xfree(bgl_part_id); } else rc = SLURM_FAILURE; if(rc == SLURM_FAILURE) error("Can't set MPIRUN_PARTITION " "environment variable"); - xfree(bgl_part_id); + } - +#endif + if (env->jobid >= 0 && setenvf(&env->env, "SLURM_JOBID", "%d", env->jobid)) { error("Unable to set SLURM_JOBID environment"); @@ -786,7 +799,7 @@ env_array_for_job(char ***dest, const resource_allocation_response_msg_t *alloc, const job_desc_msg_t *desc) { - char *bgl_part_id = NULL, *tmp; + char *tmp = NULL; slurm_step_layout_t *step_layout = NULL; uint32_t num_tasks = desc->num_tasks; @@ -807,16 +820,28 @@ env_array_overwrite(dest, "LOADLBATCH", "yes"); #endif +#ifdef HAVE_BG /* BlueGene only */ select_g_get_jobinfo(alloc->select_jobinfo, SELECT_DATA_BLOCK_ID, - &bgl_part_id); - if (bgl_part_id) { + &tmp); + if (tmp) { +#ifndef HAVE_BGL + uint16_t conn_type = (uint16_t)NO_VAL; + select_g_get_jobinfo(alloc->select_jobinfo, + SELECT_DATA_CONN_TYPE, &conn_type); + if(conn_type > SELECT_SMALL) { + env_array_overwrite_fmt(dest, "SUBMIT_POOL", "%s", + tmp); + } +#endif env_array_overwrite_fmt(dest, "MPIRUN_PARTITION", "%s", - bgl_part_id); + tmp); env_array_overwrite_fmt(dest, "MPIRUN_NOFREE", "%d", 1); env_array_overwrite_fmt(dest, "MPIRUN_NOALLOCATE", "%d", 1); - } + xfree(tmp); + } +#endif /* OBSOLETE, but needed by MPI, do not remove */ env_array_overwrite_fmt(dest, "SLURM_JOBID", "%u", alloc->job_id); env_array_overwrite_fmt(dest, "SLURM_NNODES", "%u", alloc->node_cnt); diff -Nru slurm-llnl-1.3.13/src/common/jobacct_common.c slurm-llnl-1.3.15/src/common/jobacct_common.c --- slurm-llnl-1.3.13/src/common/jobacct_common.c 2008-12-04 21:29:38.000000000 +0100 +++ slurm-llnl-1.3.15/src/common/jobacct_common.c 2009-02-20 17:05:46.000000000 +0100 @@ -176,8 +176,10 @@ xfree(job->jobname); xfree(job->partition); xfree(job->nodes); - if(job->steps) + if(job->steps) { list_destroy(job->steps); + job->steps = NULL; + } xfree(job->user); xfree(job->wckey); xfree(job); @@ -558,7 +560,6 @@ { struct jobacctinfo *jobacct = (struct jobacctinfo *)object; xfree(jobacct); - jobacct = NULL; } extern int jobacct_common_setinfo(struct jobacctinfo *jobacct, diff -Nru slurm-llnl-1.3.13/src/common/mpi.c slurm-llnl-1.3.15/src/common/mpi.c --- slurm-llnl-1.3.13/src/common/mpi.c 2008-06-06 21:15:24.000000000 +0200 +++ slurm-llnl-1.3.15/src/common/mpi.c 2009-03-24 22:57:43.000000000 +0100 @@ -110,13 +110,15 @@ static int _slurm_mpi_context_destroy( slurm_mpi_context_t c ) { + int rc = SLURM_SUCCESS; + /* * Must check return code here because plugins might still * be loaded and active. */ if ( c->plugin_list ) { if ( plugrack_destroy( c->plugin_list ) != SLURM_SUCCESS ) { - return SLURM_ERROR; + rc = SLURM_ERROR; } } else { plugin_unload(c->cur_plugin); @@ -125,7 +127,7 @@ xfree(c->mpi_type); xfree(c); - return SLURM_SUCCESS; + return rc; } /* diff -Nru slurm-llnl-1.3.13/src/common/node_select.c slurm-llnl-1.3.15/src/common/node_select.c --- slurm-llnl-1.3.13/src/common/node_select.c 2008-11-18 00:20:37.000000000 +0100 +++ slurm-llnl-1.3.15/src/common/node_select.c 2009-03-24 22:57:43.000000000 +0100 @@ -9,7 +9,7 @@ * the plugin. This is because functions required by the plugin can not be * resolved on the front-end nodes, so we can't load the plugins there. * - * $Id: node_select.c 15717 2008-11-17 23:20:37Z da $ + * $Id: node_select.c 17005 2009-03-24 21:57:43Z da $ ***************************************************************************** * Copyright (C) 2002-2006 The Regents of the University of California. * Produced at Lawrence Livermore National Laboratory (cf, DISCLAIMER). @@ -268,13 +268,14 @@ */ static int _select_context_destroy( slurm_select_context_t *c ) { + int rc = SLURM_SUCCESS; /* * Must check return code here because plugins might still * be loaded and active. */ if ( c->plugin_list ) { if ( plugrack_destroy( c->plugin_list ) != SLURM_SUCCESS ) { - return SLURM_ERROR; + rc = SLURM_ERROR; } } else { plugin_unload(c->cur_plugin); @@ -283,7 +284,7 @@ xfree( c->select_type ); xfree( c ); - return SLURM_SUCCESS; + return rc; } /* @@ -772,14 +773,33 @@ static char *_job_conn_type_string(uint16_t inx) { - if (inx == SELECT_TORUS) + switch(inx) { + case SELECT_TORUS: return "torus"; - else if (inx == SELECT_MESH) + break; + case SELECT_MESH: return "mesh"; - else if (inx == SELECT_SMALL) + break; + case SELECT_SMALL: return "small"; - else + break; +#ifndef HAVE_BGL + case SELECT_HTC_S: + return "htc_s"; + break; + case SELECT_HTC_D: + return "htc_d"; + break; + case SELECT_HTC_V: + return "htc_v"; + break; + case SELECT_HTC_L: + return "htc_l"; + break; +#endif + default: return "n/a"; + } } static char *_yes_no_string(uint16_t inx) diff -Nru slurm-llnl-1.3.13/src/common/parse_time.c slurm-llnl-1.3.15/src/common/parse_time.c --- slurm-llnl-1.3.13/src/common/parse_time.c 2009-01-05 22:57:49.000000000 +0100 +++ slurm-llnl-1.3.15/src/common/parse_time.c 2009-01-30 18:07:34.000000000 +0100 @@ -592,3 +592,48 @@ hours, minutes, seconds); } } + +extern char *mon_abbr(int mon) +{ + switch(mon) { + case 0: + return "Ja"; + break; + case 1: + return "Fe"; + break; + case 2: + return "Ma"; + break; + case 3: + return "Ap"; + break; + case 4: + return "Ma"; + break; + case 5: + return "Ju"; + break; + case 6: + return "Jl"; + break; + case 7: + return "Au"; + break; + case 8: + return "Se"; + break; + case 9: + return "Oc"; + break; + case 10: + return "No"; + break; + case 11: + return "De"; + break; + default: + return "Un"; + break; + } +} diff -Nru slurm-llnl-1.3.13/src/common/parse_time.h slurm-llnl-1.3.15/src/common/parse_time.h --- slurm-llnl-1.3.13/src/common/parse_time.h 2008-10-07 02:16:53.000000000 +0200 +++ slurm-llnl-1.3.15/src/common/parse_time.h 2009-01-30 18:07:34.000000000 +0100 @@ -96,4 +96,7 @@ extern void secs2time_str(time_t time, char *string, int size); extern void mins2time_str(uint32_t time, char *string, int size); +/* used to get a 2 char abbriviated month name from int 0-11 */ +extern char *mon_abbr(int mon); + #endif diff -Nru slurm-llnl-1.3.13/src/common/proc_args.c slurm-llnl-1.3.15/src/common/proc_args.c --- slurm-llnl-1.3.13/src/common/proc_args.c 2008-06-20 01:58:28.000000000 +0200 +++ slurm-llnl-1.3.15/src/common/proc_args.c 2009-02-05 01:01:19.000000000 +0100 @@ -125,6 +125,7 @@ */ int verify_conn_type(const char *arg) { +#ifdef HAVE_BG int len = strlen(arg); if (!strncasecmp(arg, "MESH", len)) @@ -133,7 +134,18 @@ return SELECT_TORUS; else if (!strncasecmp(arg, "NAV", len)) return SELECT_NAV; - +#ifndef HAVE_BGL + else if (!strncasecmp(arg, "HTC", len) + || !strncasecmp(arg, "HTC_S", len)) + return SELECT_HTC_S; + else if (!strncasecmp(arg, "HTC_D", len)) + return SELECT_HTC_D; + else if (!strncasecmp(arg, "HTC_V", len)) + return SELECT_HTC_V; + else if (!strncasecmp(arg, "HTC_L", len)) + return SELECT_HTC_L; +#endif +#endif error("invalid --conn-type argument %s ignored.", arg); return NO_VAL; } diff -Nru slurm-llnl-1.3.13/src/common/read_config.c slurm-llnl-1.3.15/src/common/read_config.c --- slurm-llnl-1.3.13/src/common/read_config.c 2008-12-04 22:39:19.000000000 +0100 +++ slurm-llnl-1.3.15/src/common/read_config.c 2009-02-03 00:02:44.000000000 +0100 @@ -125,7 +125,7 @@ s_p_hashtbl_t *hashtbl); s_p_options_t slurm_conf_options[] = { - {"AccountingStorageEnforce", S_P_UINT16}, + {"AccountingStorageEnforce", S_P_STRING}, {"AccountingStorageHost", S_P_STRING}, {"AccountingStorageLoc", S_P_STRING}, {"AccountingStoragePass", S_P_STRING}, @@ -1764,6 +1764,10 @@ if (!s_p_get_string(&conf->mpi_default, "MpiDefault", hashtbl)) conf->mpi_default = xstrdup(DEFAULT_MPI_DEFAULT); + if(!s_p_get_boolean((bool *)&conf->track_wckey, + "TrackWCKey", hashtbl)) + conf->track_wckey = false; + if (!s_p_get_string(&conf->accounting_storage_type, "AccountingStorageType", hashtbl)) { if(default_storage_type) @@ -1775,9 +1779,30 @@ xstrdup(DEFAULT_ACCOUNTING_STORAGE_TYPE); } - if (!s_p_get_uint16(&conf->accounting_storage_enforce, - "AccountingStorageEnforce", hashtbl)) - conf->accounting_storage_enforce = DEFAULT_ACCOUNTING_ENFORCE; + if (s_p_get_string(&temp_str, "AccountingStorageEnforce", hashtbl)) { + if (strstr(temp_str, "1") || strstr(temp_str, "associations")) + conf->accounting_storage_enforce + |= ACCOUNTING_ENFORCE_ASSOCS; + if (strstr(temp_str, "2") || strstr(temp_str, "limits")) { + conf->accounting_storage_enforce + |= ACCOUNTING_ENFORCE_ASSOCS; + conf->accounting_storage_enforce + |= ACCOUNTING_ENFORCE_LIMITS; + } + if (strstr(temp_str, "wckeys")) { + conf->accounting_storage_enforce + |= ACCOUNTING_ENFORCE_ASSOCS; + conf->accounting_storage_enforce + |= ACCOUNTING_ENFORCE_WCKEYS; + conf->track_wckey = true; + } + if (strstr(temp_str, "all")) { + conf->accounting_storage_enforce = 0xffff; + conf->track_wckey = true; + } + + xfree(temp_str); + } if (!s_p_get_string(&conf->accounting_storage_host, "AccountingStorageHost", hashtbl)) { @@ -2040,10 +2065,6 @@ if (!s_p_get_uint16(&conf->wait_time, "WaitTime", hashtbl)) conf->wait_time = DEFAULT_WAIT_TIME; - if(!s_p_get_boolean((bool *)&conf->track_wckey, - "TrackWCKey", hashtbl)) - conf->track_wckey = false; - if (s_p_get_uint16(&conf->tree_width, "TreeWidth", hashtbl)) { if (conf->tree_width == 0) { error("TreeWidth=0 is invalid"); diff -Nru slurm-llnl-1.3.13/src/common/read_config.h slurm-llnl-1.3.15/src/common/read_config.h --- slurm-llnl-1.3.13/src/common/read_config.h 2008-12-03 18:41:12.000000000 +0100 +++ slurm-llnl-1.3.15/src/common/read_config.h 2009-02-03 00:02:44.000000000 +0100 @@ -50,13 +50,11 @@ extern char *default_plugin_path; extern char *default_plugstack; -enum { - ACCOUNTING_ENFORCE_NONE, - ACCOUNTING_ENFORCE_YES, - ACCOUNTING_ENFORCE_WITH_LIMITS -}; +#define ACCOUNTING_ENFORCE_ASSOCS 0x0001 +#define ACCOUNTING_ENFORCE_LIMITS 0x0002 +#define ACCOUNTING_ENFORCE_WCKEYS 0x0004 -#define DEFAULT_ACCOUNTING_ENFORCE ACCOUNTING_ENFORCE_NONE +#define DEFAULT_ACCOUNTING_ENFORCE 0 #define DEFAULT_ACCOUNTING_STORAGE_TYPE "accounting_storage/none" #define DEFAULT_AUTH_TYPE "auth/none" #define DEFAULT_BATCH_START_TIMEOUT 10 diff -Nru slurm-llnl-1.3.13/src/common/slurm_accounting_storage.c slurm-llnl-1.3.15/src/common/slurm_accounting_storage.c --- slurm-llnl-1.3.13/src/common/slurm_accounting_storage.c 2008-12-23 00:56:27.000000000 +0100 +++ slurm-llnl-1.3.15/src/common/slurm_accounting_storage.c 2009-03-24 22:57:43.000000000 +0100 @@ -326,13 +326,14 @@ */ static int _acct_storage_context_destroy(slurm_acct_storage_context_t *c) { + int rc = SLURM_SUCCESS; /* * Must check return code here because plugins might still * be loaded and active. */ if ( c->plugin_list ) { if ( plugrack_destroy( c->plugin_list ) != SLURM_SUCCESS ) { - return SLURM_ERROR; + rc = SLURM_ERROR; } } else { plugin_unload(c->cur_plugin); @@ -341,7 +342,7 @@ xfree( c->acct_storage_type ); xfree( c ); - return SLURM_SUCCESS; + return rc; } /* diff -Nru slurm-llnl-1.3.13/src/common/slurm_accounting_storage.h slurm-llnl-1.3.15/src/common/slurm_accounting_storage.h --- slurm-llnl-1.3.13/src/common/slurm_accounting_storage.h 2008-12-23 00:56:27.000000000 +0100 +++ slurm-llnl-1.3.15/src/common/slurm_accounting_storage.h 2009-02-11 23:42:39.000000000 +0100 @@ -845,7 +845,7 @@ * note List needs to be freed when called */ extern List acct_storage_g_get_clusters( - void *db_conn, uint32_t uid, acct_cluster_cond_t *cluster_cond); + void *db_conn, uint32_t uid, acct_cluster_cond_t *cluster_cond); /* * get info from the storage diff -Nru slurm-llnl-1.3.13/src/common/slurm_auth.c slurm-llnl-1.3.15/src/common/slurm_auth.c --- slurm-llnl-1.3.13/src/common/slurm_auth.c 2008-06-06 21:15:24.000000000 +0200 +++ slurm-llnl-1.3.15/src/common/slurm_auth.c 2009-03-24 22:57:43.000000000 +0100 @@ -280,13 +280,15 @@ static int _slurm_auth_context_destroy( slurm_auth_context_t c ) { + int rc = SLURM_SUCCESS; + /* * Must check return code here because plugins might still * be loaded and active. */ if ( c->plugin_list ) { if ( plugrack_destroy( c->plugin_list ) != SLURM_SUCCESS ) { - return SLURM_ERROR; + rc = SLURM_ERROR; } } else { plugin_unload(c->cur_plugin); @@ -295,7 +297,7 @@ xfree( c->auth_type ); xfree( c ); - return SLURM_SUCCESS; + return rc; } int inline diff -Nru slurm-llnl-1.3.13/src/common/slurm_cred.c slurm-llnl-1.3.15/src/common/slurm_cred.c --- slurm-llnl-1.3.13/src/common/slurm_cred.c 2008-11-25 02:07:17.000000000 +0100 +++ slurm-llnl-1.3.15/src/common/slurm_cred.c 2009-03-24 22:57:43.000000000 +0100 @@ -1,6 +1,6 @@ /*****************************************************************************\ * src/common/slurm_cred.c - SLURM job credential functions - * $Id: slurm_cred.c 15765 2008-11-25 01:07:17Z jette $ + * $Id: slurm_cred.c 17005 2009-03-24 21:57:43Z da $ ***************************************************************************** * Copyright (C) 2002-2007 The Regents of the University of California. * Copyright (C) 2008 Lawrence Livermore National Security. @@ -274,13 +274,14 @@ static int _slurm_crypto_context_destroy( slurm_crypto_context_t *c ) { + int rc = SLURM_SUCCESS; /* * Must check return code here because plugins might still * be loaded and active. */ if ( c->plugin_list ) { if ( plugrack_destroy( c->plugin_list ) != SLURM_SUCCESS ) { - return SLURM_ERROR; + rc = SLURM_ERROR; } } else { plugin_unload(c->cur_plugin); @@ -289,7 +290,7 @@ xfree( c->crypto_type ); xfree( c ); - return SLURM_SUCCESS; + return rc; } /* diff -Nru slurm-llnl-1.3.13/src/common/slurmdbd_defs.c slurm-llnl-1.3.15/src/common/slurmdbd_defs.c --- slurm-llnl-1.3.13/src/common/slurmdbd_defs.c 2008-12-22 22:46:26.000000000 +0100 +++ slurm-llnl-1.3.15/src/common/slurmdbd_defs.c 2009-03-23 21:37:54.000000000 +0100 @@ -145,10 +145,13 @@ tmp_errno = errno; } slurm_mutex_unlock(&slurmdbd_lock); - + slurm_mutex_lock(&agent_lock); if (make_agent && ((agent_tid == 0) || (agent_list == NULL))) _create_agent(); + else if(agent_list) + _load_dbd_state(); + slurm_mutex_unlock(&agent_lock); if(tmp_errno) { @@ -203,7 +206,8 @@ } else { /* resp->msg_type == DBD_RC */ dbd_rc_msg_t *msg = resp->data; *resp_code = msg->return_code; - if(msg->return_code != SLURM_SUCCESS) + if(msg->return_code != SLURM_SUCCESS + && msg->return_code != ACCOUNTING_FIRST_REG) error("slurmdbd(%d): from %u: %s", msg->return_code, msg->sent_type, msg->comment); slurmdbd_free_rc_msg(rpc_version, msg); @@ -244,9 +248,10 @@ rc = _send_msg(buffer); free_buf(buffer); if (rc != SLURM_SUCCESS) { - error("slurmdbd: Sending message type %u", req->msg_type); + error("slurmdbd: Sending message type %u: %d: %m", + req->msg_type, rc); slurm_mutex_unlock(&slurmdbd_lock); - return SLURM_ERROR; + return rc; } buffer = _recv_msg(read_timeout); @@ -1190,7 +1195,7 @@ rc = _send_msg(buffer); free_buf(buffer); if (rc != SLURM_SUCCESS) { - error("slurmdbd: Sending DBD_INIT message"); + error("slurmdbd: Sending DBD_INIT message: %d: %m", rc); return rc; } @@ -1519,6 +1524,9 @@ ****************************************************************************/ static void _create_agent(void) { + /* this needs to be set because the agent thread will do + nothing if the connection was closed and then opened again */ + agent_shutdown = 0; if (agent_list == NULL) { agent_list = list_create(_agent_queue_del); if (agent_list == NULL) @@ -1649,7 +1657,7 @@ slurm_mutex_unlock(&slurmdbd_lock); break; } - error("slurmdbd: Failure sending message"); + error("slurmdbd: Failure sending message: %d: %m", rc); } else { rc = _get_return_code(SLURMDBD_VERSION, read_timeout); if (rc == EAGAIN) { @@ -1658,7 +1666,7 @@ break; } error("slurmdbd: Failure with " - "message need to resend"); + "message need to resend: %d: %m", rc); } } slurm_mutex_unlock(&slurmdbd_lock); @@ -1741,12 +1749,18 @@ char *dbd_fname; Buf buffer; int fd, recovered = 0; - + dbd_fname = slurm_get_state_save_location(); xstrcat(dbd_fname, "/dbd.messages"); fd = open(dbd_fname, O_RDONLY); if (fd < 0) { - error("slurmdbd: Opening state save file %s", dbd_fname); + /* don't print an error message if there is no file */ + if(errno == ENOENT) + debug4("slurmdbd: There is no state save file to " + "open by name %s", dbd_fname); + else + error("slurmdbd: Opening state save file %s: %m", + dbd_fname); } else { while (1) { buffer = _load_dbd_rec(fd); @@ -1756,8 +1770,7 @@ fatal("slurmdbd: list_enqueue, no memory"); recovered++; } - } - if (fd >= 0) { + verbose("slurmdbd: recovered %d pending RPCs", recovered); (void) close(fd); (void) unlink(dbd_fname); /* clear save state */ diff -Nru slurm-llnl-1.3.13/src/common/slurm_errno.c slurm-llnl-1.3.15/src/common/slurm_errno.c --- slurm-llnl-1.3.13/src/common/slurm_errno.c 2008-10-07 02:16:53.000000000 +0200 +++ slurm-llnl-1.3.15/src/common/slurm_errno.c 2009-02-03 00:02:44.000000000 +0100 @@ -212,6 +212,8 @@ "Memory required by task is not available" }, { ESLURM_INVALID_ACCOUNT, "Job has invalid account" }, + { ESLURM_INVALID_WCKEY, + "Job has invalid wckey" }, { ESLURM_INVALID_LICENSES, "Job has invalid license specification" }, { ESLURM_NEED_RESTART, diff -Nru slurm-llnl-1.3.13/src/common/slurm_jobacct_gather.c slurm-llnl-1.3.15/src/common/slurm_jobacct_gather.c --- slurm-llnl-1.3.13/src/common/slurm_jobacct_gather.c 2008-06-06 21:15:24.000000000 +0200 +++ slurm-llnl-1.3.15/src/common/slurm_jobacct_gather.c 2009-03-24 22:57:43.000000000 +0100 @@ -142,13 +142,14 @@ static int _slurm_jobacct_gather_context_destroy( slurm_jobacct_gather_context_t *c ) { + int rc = SLURM_SUCCESS; /* * Must check return code here because plugins might still * be loaded and active. */ if ( c->plugin_list ) { if ( plugrack_destroy( c->plugin_list ) != SLURM_SUCCESS ) { - return SLURM_ERROR; + rc = SLURM_ERROR; } } else { plugin_unload(c->cur_plugin); @@ -157,7 +158,7 @@ xfree( c->jobacct_gather_type ); xfree( c ); - return SLURM_SUCCESS; + return rc; } /* diff -Nru slurm-llnl-1.3.13/src/common/slurm_jobcomp.c slurm-llnl-1.3.15/src/common/slurm_jobcomp.c --- slurm-llnl-1.3.13/src/common/slurm_jobcomp.c 2008-12-22 22:46:26.000000000 +0100 +++ slurm-llnl-1.3.15/src/common/slurm_jobcomp.c 2009-03-24 22:57:43.000000000 +0100 @@ -118,13 +118,14 @@ static int _slurm_jobcomp_context_destroy( slurm_jobcomp_context_t c ) { + int rc = SLURM_SUCCESS; /* * Must check return code here because plugins might still * be loaded and active. */ if ( c->plugin_list ) { if ( plugrack_destroy( c->plugin_list ) != SLURM_SUCCESS ) { - return SLURM_ERROR; + rc = SLURM_ERROR; } } else { plugin_unload(c->cur_plugin); @@ -133,7 +134,7 @@ xfree( c->jobcomp_type ); xfree( c ); - return SLURM_SUCCESS; + return rc; } /* diff -Nru slurm-llnl-1.3.13/src/common/slurm_protocol_api.c slurm-llnl-1.3.15/src/common/slurm_protocol_api.c --- slurm-llnl-1.3.13/src/common/slurm_protocol_api.c 2008-12-04 22:39:19.000000000 +0100 +++ slurm-llnl-1.3.15/src/common/slurm_protocol_api.c 2009-02-24 21:50:48.000000000 +0100 @@ -678,7 +678,7 @@ conf = slurm_conf_lock(); if(!strcasecmp(conf->accounting_storage_type, "accounting_storage/slurmdbd") - || strcasecmp(conf->accounting_storage_type, + || !strcasecmp(conf->accounting_storage_type, "accounting_storage/mysql")) enforce = 1; slurm_conf_unlock(); diff -Nru slurm-llnl-1.3.13/src/common/slurm_protocol_socket_implementation.c slurm-llnl-1.3.15/src/common/slurm_protocol_socket_implementation.c --- slurm-llnl-1.3.13/src/common/slurm_protocol_socket_implementation.c 2008-12-17 18:58:58.000000000 +0100 +++ slurm-llnl-1.3.15/src/common/slurm_protocol_socket_implementation.c 2009-02-20 00:57:13.000000000 +0100 @@ -77,7 +77,7 @@ #include "src/common/xmalloc.h" #include "src/common/util-net.h" -#define PORT_RETRIES 2 +#define PORT_RETRIES 3 #define MIN_USER_PORT (IPPORT_RESERVED + 1) #define MAX_USER_PORT 0xffff #define RANDOM_USER_PORT ((uint16_t) ((lrand48() % \ @@ -525,7 +525,7 @@ sizeof(*addr)); if (rc >= 0) /* success */ break; - if ((errno != ECONNREFUSED) || + if (((errno != ECONNREFUSED) && (errno != ETIMEDOUT)) || (!retry) || (retry_cnt >= PORT_RETRIES)) { slurm_seterrno(errno); goto error; diff -Nru slurm-llnl-1.3.13/src/common/switch.c slurm-llnl-1.3.15/src/common/switch.c --- slurm-llnl-1.3.13/src/common/switch.c 2008-06-06 21:15:24.000000000 +0200 +++ slurm-llnl-1.3.15/src/common/switch.c 2009-03-24 22:57:43.000000000 +0100 @@ -160,13 +160,14 @@ static int _slurm_switch_context_destroy( slurm_switch_context_t c ) { + int rc = SLURM_SUCCESS; /* * Must check return code here because plugins might still * be loaded and active. */ if ( c->plugin_list ) { if ( plugrack_destroy( c->plugin_list ) != SLURM_SUCCESS ) { - return SLURM_ERROR; + rc = SLURM_ERROR; } } else { plugin_unload(c->cur_plugin); @@ -175,7 +176,7 @@ xfree( c->switch_type ); xfree( c ); - return SLURM_SUCCESS; + return rc; } /* diff -Nru slurm-llnl-1.3.13/src/common/uid.c slurm-llnl-1.3.15/src/common/uid.c --- slurm-llnl-1.3.13/src/common/uid.c 2008-08-15 23:54:22.000000000 +0200 +++ slurm-llnl-1.3.15/src/common/uid.c 2009-04-07 20:09:43.000000000 +0200 @@ -1,6 +1,6 @@ /*****************************************************************************\ * src/common/uid.c - uid/gid lookup utility functions - * $Id: uid.c 14795 2008-08-15 21:54:22Z jette $ + * $Id: uid.c 17177 2009-04-07 18:09:43Z jette $ ***************************************************************************** * Copyright (C) 2002 The Regents of the University of California. * Produced at Lawrence Livermore National Laboratory (cf, DISCLAIMER). @@ -111,6 +111,24 @@ } gid_t +gid_from_uid (uid_t uid) +{ + struct passwd pwd, *result; + char buffer[PW_BUF_SIZE]; + gid_t gid; + int rc; + + rc = getpwuid_r(uid, &pwd, buffer, PW_BUF_SIZE, &result); + if (result == NULL) { + gid = (gid_t) -1; + } else { + gid = result->pw_gid; + } + + return gid; +} + +gid_t gid_from_string (char *name) { struct group grp, *result; diff -Nru slurm-llnl-1.3.13/src/common/uid.h slurm-llnl-1.3.15/src/common/uid.h --- slurm-llnl-1.3.13/src/common/uid.h 2008-08-15 23:54:22.000000000 +0200 +++ slurm-llnl-1.3.15/src/common/uid.h 2009-04-07 20:09:43.000000000 +0200 @@ -1,6 +1,6 @@ /*****************************************************************************\ * src/common/uid.h - uid/gid lookup utility functions - * $Id: uid.h 14795 2008-08-15 21:54:22Z jette $ + * $Id: uid.h 17177 2009-04-07 18:09:43Z jette $ ***************************************************************************** * Copyright (C) 2002 The Regents of the University of California. * Produced at Lawrence Livermore National Laboratory (cf, DISCLAIMER). @@ -60,6 +60,12 @@ uid_t uid_from_string (char *name); /* + * Return the primary group id for a given user id, or + * (gid_t) -1 on failure. + */ +gid_t gid_from_uid (uid_t uid); + +/* * Same as uid_from_name(), but for group name/id. */ gid_t gid_from_string (char *name); diff -Nru slurm-llnl-1.3.13/src/database/mysql_common.c slurm-llnl-1.3.15/src/database/mysql_common.c --- slurm-llnl-1.3.13/src/database/mysql_common.c 2008-12-09 22:37:22.000000000 +0100 +++ slurm-llnl-1.3.15/src/database/mysql_common.c 2009-03-17 18:50:44.000000000 +0100 @@ -66,7 +66,12 @@ mysql_errno(mysql_db), mysql_error(mysql_db)); } while (rc == 0); - + + if(rc > 0) { + errno = rc; + return SLURM_ERROR; + } + return SLURM_SUCCESS; } @@ -112,6 +117,7 @@ storage_field_t *fields, char *ending) { char *query = NULL; + char *correct_query = NULL; MYSQL_RES *result = NULL; MYSQL_ROW row; int i = 0; @@ -162,6 +168,7 @@ itr = list_iterator_create(columns); query = xstrdup_printf("alter table %s", table_name); + correct_query = xstrdup_printf("alter table %s", table_name); START_TIMER; while(fields[i].name) { int found = 0; @@ -171,6 +178,9 @@ xstrfmtcat(query, " modify %s %s,", fields[i].name, fields[i].options); + xstrfmtcat(correct_query, " modify %s %s,", + fields[i].name, + fields[i].options); list_delete_item(itr); found = 1; break; @@ -186,6 +196,9 @@ fields[i].name, fields[i].options, fields[i-1].name); + xstrfmtcat(correct_query, " modify %s %s,", + fields[i].name, + fields[i].options); } else { info("adding column %s at the beginning " "of table %s", @@ -195,6 +208,9 @@ xstrfmtcat(query, " add %s %s first,", fields[i].name, fields[i].options); + xstrfmtcat(correct_query, " modify %s %s,", + fields[i].name, + fields[i].options); } adding = 1; } @@ -228,9 +244,13 @@ if(temp[end]) { end++; primary_key = xstrndup(temp, end); - if(old_primary) + if(old_primary) { xstrcat(query, " drop primary key,"); + xstrcat(correct_query, " drop primary key,"); + } xstrfmtcat(query, " add %s,", primary_key); + xstrfmtcat(correct_query, " add %s,", primary_key); + xfree(primary_key); } } @@ -251,16 +271,21 @@ if(temp[end]) { end++; unique_index = xstrndup(temp, end); - if(old_index) + if(old_index) { xstrfmtcat(query, " drop index %s,", old_index); + xstrfmtcat(correct_query, " drop index %s,", + old_index); + } xstrfmtcat(query, " add %s,", unique_index); + xstrfmtcat(correct_query, " add %s,", unique_index); xfree(unique_index); } } xfree(old_index); query[strlen(query)-1] = ';'; + correct_query[strlen(correct_query)-1] = ';'; //info("%d query\n%s", __LINE__, query); /* see if we have already done this definition */ @@ -287,6 +312,7 @@ char *query2 = NULL; debug("Table %s has changed. Updating...", table_name); + if(mysql_db_query(mysql_db, query)) { xfree(query); return SLURM_ERROR; @@ -298,15 +324,17 @@ "on duplicate key update " "definition=\"%s\", mod_time=%d;", table_defs_table, now, now, - table_name, query, query, now); + table_name, correct_query, + correct_query, now); if(mysql_db_query(mysql_db, query2)) { xfree(query2); return SLURM_ERROR; } xfree(query2); } - + xfree(query); + xfree(correct_query); query = xstrdup_printf("make table current %s", table_name); END_TIMER2(query); xfree(query); @@ -377,17 +405,18 @@ { int rc = SLURM_SUCCESS; bool storage_init = false; - + if(!(*mysql_db = mysql_init(*mysql_db))) fatal("mysql_init failed: %s", mysql_error(*mysql_db)); else { + unsigned int my_timeout = 30; #ifdef MYSQL_OPT_RECONNECT -{ my_bool reconnect = 1; /* make sure reconnect is on */ mysql_options(*mysql_db, MYSQL_OPT_RECONNECT, &reconnect); -} #endif + mysql_options(*mysql_db, MYSQL_OPT_CONNECT_TIMEOUT, + (char *)&my_timeout); while(!storage_init) { if(!mysql_real_connect(*mysql_db, db_info->host, db_info->user, db_info->pass, @@ -544,6 +573,16 @@ return result; } +extern int mysql_db_query_check_after(MYSQL *mysql_db, char *query) +{ + int rc = SLURM_SUCCESS; + + if((rc = mysql_db_query(mysql_db, query)) != SLURM_ERROR) + rc = _clear_results(mysql_db); + + return rc; +} + extern int mysql_insert_ret_id(MYSQL *mysql_db, char *query) { int new_id = 0; diff -Nru slurm-llnl-1.3.13/src/database/mysql_common.h slurm-llnl-1.3.15/src/database/mysql_common.h --- slurm-llnl-1.3.13/src/database/mysql_common.h 2008-07-21 21:58:02.000000000 +0200 +++ slurm-llnl-1.3.15/src/database/mysql_common.h 2009-01-13 22:14:04.000000000 +0100 @@ -95,6 +95,7 @@ extern int mysql_db_rollback(MYSQL *mysql_db); extern MYSQL_RES *mysql_db_query_ret(MYSQL *mysql_db, char *query, bool last); +extern int mysql_db_query_check_after(MYSQL *mysql_db, char *query); extern int mysql_insert_ret_id(MYSQL *mysql_db, char *query); diff -Nru slurm-llnl-1.3.13/src/plugins/accounting_storage/filetxt/accounting_storage_filetxt.c slurm-llnl-1.3.15/src/plugins/accounting_storage/filetxt/accounting_storage_filetxt.c --- slurm-llnl-1.3.13/src/plugins/accounting_storage/filetxt/accounting_storage_filetxt.c 2008-12-22 22:46:26.000000000 +0100 +++ slurm-llnl-1.3.15/src/plugins/accounting_storage/filetxt/accounting_storage_filetxt.c 2009-03-25 20:26:12.000000000 +0100 @@ -488,8 +488,8 @@ { int i, rc=SLURM_SUCCESS; - char buf[BUFFER_SIZE], *jname, *account, *nodes; - char *wckey = NULL; + char buf[BUFFER_SIZE], *account, *nodes; + char *jname = NULL, *wckey = NULL; long priority; int track_steps = 0; @@ -509,7 +509,7 @@ } priority = (job_ptr->priority == NO_VAL) ? - -1L : (long) job_ptr->priority; + -1L : (long) job_ptr->priority; if (job_ptr->name && job_ptr->name[0]) { char *temp = NULL; diff -Nru slurm-llnl-1.3.13/src/plugins/accounting_storage/filetxt/filetxt_jobacct_process.c slurm-llnl-1.3.15/src/plugins/accounting_storage/filetxt/filetxt_jobacct_process.c --- slurm-llnl-1.3.13/src/plugins/accounting_storage/filetxt/filetxt_jobacct_process.c 2008-12-22 22:46:26.000000000 +0100 +++ slurm-llnl-1.3.15/src/plugins/accounting_storage/filetxt/filetxt_jobacct_process.c 2009-02-24 21:50:48.000000000 +0100 @@ -640,23 +640,6 @@ return job; } -static int _remove_job_record(List job_list, uint32_t jobnum) -{ - filetxt_job_rec_t *job = NULL; - int rc = SLURM_ERROR; - ListIterator itr = list_iterator_create(job_list); - - while((job = (filetxt_job_rec_t *)list_next(itr)) != NULL) { - if (job->header.jobnum == jobnum) { - list_remove(itr); - _destroy_filetxt_job_rec(job); - rc = SLURM_SUCCESS; - } - } - list_iterator_destroy(itr); - return rc; -} - static filetxt_step_rec_t *_find_step_record(filetxt_job_rec_t *job, long stepnum) { @@ -704,9 +687,12 @@ (*job)->priority = atoi(f[F_PRIORITY]); (*job)->ncpus = atoi(f[F_NCPUS]); (*job)->nodes = xstrdup(f[F_NODES]); + for (i=0; (*job)->nodes[i]; i++) { /* discard trailing */ - if (isspace((*job)->nodes[i])) + if (isspace((*job)->nodes[i])) { (*job)->nodes[i] = '\0'; + info("got here"); + } } if (!strcmp((*job)->nodes, "(null)")) { xfree((*job)->nodes); @@ -814,7 +800,7 @@ (*job)->exitcode = atoi(f[F_JOB_EXITCODE]); break; default: - printf("UNKOWN TYPE %d",i); + error("UNKOWN TYPE %d",i); break; } return SLURM_SUCCESS; @@ -828,19 +814,21 @@ _parse_line(f, (void **)&temp, len); job = _find_job_record(job_list, temp->header, JOB_START); - if (job) { /* Hmmm... that's odd */ - printf("job->header.job_submit = %d", - (int)job->header.job_submit); - if(job->header.job_submit == 0) - _remove_job_record(job_list, job->header.jobnum); - else { - fprintf(stderr, - "Conflicting JOB_START for job %u at" - " line %d -- ignoring it\n", - job->header.jobnum, lc); - _destroy_filetxt_job_rec(temp); - return; - } + if (job) { + /* in slurm we can get 2 start records one for submit + * and one for start, so look at the last one */ + xfree(job->jobname); + job->jobname = xstrdup(temp->jobname); + job->track_steps = temp->track_steps; + job->priority = temp->priority; + job->ncpus = temp->ncpus; + xfree(job->nodes); + job->nodes = xstrdup(temp->nodes); + xfree(job->account); + job->account = xstrdup(temp->account); + + _destroy_filetxt_job_rec(temp); + return; } job = temp; @@ -861,7 +849,7 @@ _parse_line(f, (void **)&temp, len); job = _find_job_record(job_list, temp->header, JOB_STEP); - + if (temp->stepnum == -2) { _destroy_filetxt_step_rec(temp); return; @@ -936,8 +924,7 @@ job->nodes = xstrdup(step->nodes); } -got_step: - +got_step: if (job->job_terminated_seen == 0) { /* If the job is still running, this is the most recent @@ -1053,6 +1040,8 @@ if(job_cond) { fdump_flag = job_cond->duplicates & FDUMP_FLAG; job_cond->duplicates &= (~FDUMP_FLAG); + if(!job_cond->duplicates) + itr2 = list_iterator_create(ret_job_list); } fd = _open_log_file(filein); @@ -1169,21 +1158,21 @@ switch(rec_type) { case JOB_START: if(i < F_JOB_ACCOUNT) { - printf("Bad data on a Job Start\n"); + error("Bad data on a Job Start\n"); _show_rec(f); } else _process_start(job_list, f, lc, show_full, i); break; case JOB_STEP: if(i < F_MAX_VSIZE) { - printf("Bad data on a Step entry\n"); + error("Bad data on a Step entry\n"); _show_rec(f); } else _process_step(job_list, f, lc, show_full, i); break; case JOB_SUSPEND: if(i < F_JOB_REQUID) { - printf("Bad data on a Suspend entry\n"); + error("Bad data on a Suspend entry\n"); _show_rec(f); } else _process_suspend(job_list, f, lc, @@ -1191,7 +1180,7 @@ break; case JOB_TERMINATED: if(i < F_JOB_REQUID) { - printf("Bad data on a Job Term\n"); + error("Bad data on a Job Term\n"); _show_rec(f); } else _process_terminated(job_list, f, lc, @@ -1211,30 +1200,28 @@ fclose(fd); itr = list_iterator_create(job_list); - if(!job_cond->duplicates) - itr2 = list_iterator_create(ret_job_list); + while((filetxt_job = list_next(itr))) { jobacct_job_rec_t *jobacct_job = _create_jobacct_job_rec(filetxt_job, job_cond); if(jobacct_job) { jobacct_job_rec_t *curr_job = NULL; - if(job_cond && !job_cond->duplicates) { + if(itr2) { + list_iterator_reset(itr2); while((curr_job = list_next(itr2))) { if (curr_job->jobid == jobacct_job->jobid) { list_delete_item(itr2); + info("removing job %d", jobacct_job->jobid); break; } } } list_append(ret_job_list, jobacct_job); - - if(!job_cond->duplicates) - list_iterator_reset(itr2); } } - if(!job_cond->duplicates) + if(itr2) list_iterator_destroy(itr2); list_iterator_destroy(itr); @@ -1302,12 +1289,12 @@ goto finished; } if ((statbuf.st_mode & S_IFLNK) == S_IFLNK) { - fprintf(stderr, "%s is a symbolic link; --expire requires " + error("%s is a symbolic link; --expire requires " "a hard-linked file name\n", filein); goto finished; } if (!(statbuf.st_mode & S_IFREG)) { - fprintf(stderr, "%s is not a regular file; --expire " + error("%s is not a regular file; --expire " "only works on accounting log files\n", filein); goto finished; @@ -1324,7 +1311,7 @@ goto finished; } } else { - fprintf(stderr, "Warning! %s exists -- please remove " + error("Warning! %s exists -- please remove " "or rename it before proceeding\n", old_logfile_name); goto finished; @@ -1386,7 +1373,7 @@ sprintf(logfile_name, "%s.expired", filein); new_file = stat(logfile_name, &statbuf); if ((expired_logfile = fopen(logfile_name, "a"))==NULL) { - fprintf(stderr, "Error while opening %s", + error("Error while opening %s", logfile_name); perror(""); xfree(logfile_name); @@ -1401,7 +1388,7 @@ logfile_name = _prefix_filename(filein, ".new."); if ((new_logfile = fopen(logfile_name, "w"))==NULL) { - fprintf(stderr, "Error while opening %s", + error("Error while opening %s", logfile_name); perror(""); fclose(expired_logfile); @@ -1421,16 +1408,16 @@ list_sort(keep_list, (ListCmpF) _cmp_jrec); /* if (params->opt_verbose > 2) { */ -/* fprintf(stderr, "--- contents of exp_list ---"); */ +/* error("--- contents of exp_list ---"); */ /* itr = list_iterator_create(exp_list); */ /* while((exp_rec = list_next(itr))) { */ /* if (!(i%5)) */ -/* fprintf(stderr, "\n"); */ +/* error("\n"); */ /* else */ -/* fprintf(stderr, "\t"); */ -/* fprintf(stderr, "%d", exp_rec->job); */ +/* error("\t"); */ +/* error("%d", exp_rec->job); */ /* } */ -/* fprintf(stderr, "\n---- end of exp_list ---\n"); */ +/* error("\n---- end of exp_list ---\n"); */ /* list_iterator_destroy(itr); */ /* } */ /* write the expired file */ @@ -1506,10 +1493,10 @@ perror("renaming new logfile"); /* undo it? */ if (!rename(old_logfile_name, filein)) - fprintf(stderr, "Please correct the problem " + error("Please correct the problem " "and try again"); else - fprintf(stderr, "SEVERE ERROR: Current accounting " + error("SEVERE ERROR: Current accounting " "log may have been renamed %s;\n" "please rename it to \"%s\" if necessary, " "and try again\n", @@ -1522,7 +1509,7 @@ file_err = slurm_reconfigure(); if (file_err) { file_err = 1; - fprintf(stderr, "Error: Attempt to reconfigure " + error("Error: Attempt to reconfigure " "SLURM failed.\n"); if (rename(old_logfile_name, filein)) { perror("renaming logfile from .old."); diff -Nru slurm-llnl-1.3.13/src/plugins/accounting_storage/mysql/accounting_storage_mysql.c slurm-llnl-1.3.15/src/plugins/accounting_storage/mysql/accounting_storage_mysql.c --- slurm-llnl-1.3.13/src/plugins/accounting_storage/mysql/accounting_storage_mysql.c 2009-01-12 22:49:26.000000000 +0100 +++ slurm-llnl-1.3.15/src/plugins/accounting_storage/mysql/accounting_storage_mysql.c 2009-03-23 20:30:53.000000000 +0100 @@ -163,6 +163,41 @@ uint32_t uid, acct_wckey_cond_t *wckey_cond); +static char *_get_cluster_from_associd(mysql_conn_t *mysql_conn, + uint32_t associd) +{ + char *cluster = NULL; + char *query = NULL; + MYSQL_RES *result = NULL; + MYSQL_ROW row; + + /* Just so we don't have to keep a + cache of the associations around we + will just query the db for the cluster + name of the association id. Since + this should sort of be a rare case + this isn't too bad. + */ + query = xstrdup_printf("select cluster from %s where id=%u", + assoc_table, associd); + + debug4("%d(%d) query\n%s", + mysql_conn->conn, __LINE__, query); + if(!(result = + mysql_db_query_ret(mysql_conn->db_conn, query, 0))) { + xfree(query); + return NULL; + } + xfree(query); + + if((row = mysql_fetch_row(result))) + cluster = xstrdup(row[0]); + + mysql_free_result(result); + + return cluster; +} + static char *_get_user_from_associd(mysql_conn_t *mysql_conn, uint32_t associd) { char *user = NULL; @@ -1131,7 +1166,7 @@ } else if(assoc_cond->user_list) { /* we want all the users, but no non-user associations */ set = 1; - xstrfmtcat(*extra, " && (%s.user!='')", prefix); + xstrfmtcat(*extra, " && (%s.user!='')", prefix); } if(assoc_cond->partition_list @@ -2359,6 +2394,224 @@ return db_index; } +/* checks should already be done before this to see if this is a valid + user or not. +*/ +static int _get_usage_for_list(mysql_conn_t *mysql_conn, + slurmdbd_msg_type_t type, List object_list, + time_t start, time_t end) +{ +#ifdef HAVE_MYSQL + int rc = SLURM_SUCCESS; + int i=0; + MYSQL_RES *result = NULL; + MYSQL_ROW row; + char *tmp = NULL; + char *my_usage_table = NULL; + char *query = NULL; + List usage_list = NULL; + char *id_str = NULL; + ListIterator itr = NULL, u_itr = NULL; + void *object = NULL; + acct_association_rec_t *assoc = NULL; + acct_wckey_rec_t *wckey = NULL; + acct_accounting_rec_t *accounting_rec = NULL; + + /* Since for id in association table we + use t3 and in wckey table we use t1 we can't define it here */ + char **usage_req_inx = NULL; + + enum { + USAGE_ID, + USAGE_START, + USAGE_ACPU, + USAGE_COUNT + }; + + + if(!object_list) { + error("We need an object to set data for getting usage"); + return SLURM_ERROR; + } + + if(_check_connection(mysql_conn) != SLURM_SUCCESS) + return SLURM_ERROR; + + switch (type) { + case DBD_GET_ASSOC_USAGE: + { + char *temp_usage[] = { + "t3.id", + "t1.period_start", + "t1.alloc_cpu_secs" + }; + usage_req_inx = temp_usage; + + itr = list_iterator_create(object_list); + while((assoc = list_next(itr))) { + if(id_str) + xstrfmtcat(id_str, " || t3.id=%d", assoc->id); + else + xstrfmtcat(id_str, "t3.id=%d", assoc->id); + } + list_iterator_destroy(itr); + + my_usage_table = assoc_day_table; + break; + } + case DBD_GET_WCKEY_USAGE: + { + char *temp_usage[] = { + "id", + "period_start", + "alloc_cpu_secs" + }; + usage_req_inx = temp_usage; + + itr = list_iterator_create(object_list); + while((wckey = list_next(itr))) { + if(id_str) + xstrfmtcat(id_str, " || id=%d", wckey->id); + else + xstrfmtcat(id_str, "id=%d", wckey->id); + } + list_iterator_destroy(itr); + + my_usage_table = wckey_day_table; + break; + } + default: + error("Unknown usage type %d", type); + return SLURM_ERROR; + break; + } + + if(_set_usage_information(&my_usage_table, type, &start, &end) + != SLURM_SUCCESS) { + xfree(id_str); + return SLURM_ERROR; + } + + xfree(tmp); + i=0; + xstrfmtcat(tmp, "%s", usage_req_inx[i]); + for(i=1; i= %d) " + "&& t1.id=t2.id && (%s) && " + "t2.lft between t3.lft and t3.rgt " + "order by t3.id, period_start;", + tmp, my_usage_table, assoc_table, assoc_table, + end, start, id_str); + break; + case DBD_GET_WCKEY_USAGE: + query = xstrdup_printf( + "select %s from %s " + "where (period_start < %d && period_start >= %d) " + "&& %s order by id, period_start;", + tmp, my_usage_table, end, start, id_str); + break; + default: + error("Unknown usage type %d", type); + xfree(id_str); + xfree(tmp); + return SLURM_ERROR; + break; + } + xfree(id_str); + xfree(tmp); + + debug4("%d(%d) query\n%s", mysql_conn->conn, __LINE__, query); + if(!(result = mysql_db_query_ret( + mysql_conn->db_conn, query, 0))) { + xfree(query); + return SLURM_ERROR; + } + xfree(query); + + usage_list = list_create(destroy_acct_accounting_rec); + + while((row = mysql_fetch_row(result))) { + acct_accounting_rec_t *accounting_rec = + xmalloc(sizeof(acct_accounting_rec_t)); + accounting_rec->id = atoi(row[USAGE_ID]); + accounting_rec->period_start = atoi(row[USAGE_START]); + accounting_rec->alloc_secs = atoll(row[USAGE_ACPU]); + list_append(usage_list, accounting_rec); + } + mysql_free_result(result); + + u_itr = list_iterator_create(usage_list); + itr = list_iterator_create(object_list); + while((object = list_next(itr))) { + int found = 0; + int id = 0; + List acct_list = NULL; + + switch (type) { + case DBD_GET_ASSOC_USAGE: + assoc = (acct_association_rec_t *)object; + if(!assoc->accounting_list) + assoc->accounting_list = list_create( + destroy_acct_accounting_rec); + acct_list = assoc->accounting_list; + id = assoc->id; + break; + case DBD_GET_WCKEY_USAGE: + wckey = (acct_wckey_rec_t *)object; + if(!wckey->accounting_list) + wckey->accounting_list = list_create( + destroy_acct_accounting_rec); + acct_list = wckey->accounting_list; + id = wckey->id; + break; + default: + continue; + break; + } + + while((accounting_rec = list_next(u_itr))) { + if(id == accounting_rec->id) { + list_append(acct_list, accounting_rec); + list_remove(u_itr); + found = 1; + } else if(found) { + /* here we know the + list is in id order so + if the next record + isn't the correct id + just continue since + there is no reason to + go through the rest of + the list when we know + it isn't going to be + the correct id */ + break; + } + } + list_iterator_reset(u_itr); + } + list_iterator_destroy(itr); + list_iterator_destroy(u_itr); + + if(list_count(usage_list)) + error("we have %d records not added " + "to the association list", + list_count(usage_list)); + list_destroy(usage_list); + + + return rc; +#else + return SLURM_ERROR; +#endif +} + static mysql_db_info_t *_mysql_acct_create_db_info() { mysql_db_info_t *db_info = xmalloc(sizeof(mysql_db_info_t)); @@ -2444,7 +2697,6 @@ { "control_host", "tinytext not null default ''" }, { "control_port", "mediumint not null default 0" }, { "rpc_version", "mediumint not null default 0" }, - { "valid_qos", "blob" }, { NULL, NULL} }; @@ -7250,14 +7502,12 @@ "control_host", "control_port", "rpc_version", - "valid_qos", }; enum { CLUSTER_REQ_NAME, CLUSTER_REQ_CH, CLUSTER_REQ_CP, CLUSTER_REQ_VERSION, - CLUSTER_REQ_VALID_QOS, CLUSTER_REQ_COUNT }; @@ -7344,13 +7594,6 @@ cluster->control_host = xstrdup(row[CLUSTER_REQ_CH]); cluster->control_port = atoi(row[CLUSTER_REQ_CP]); cluster->rpc_version = atoi(row[CLUSTER_REQ_VERSION]); - cluster->valid_qos_list = list_create(slurm_destroy_char); - - if(row[CLUSTER_REQ_VALID_QOS] && row[CLUSTER_REQ_VALID_QOS][0]) - slurm_addto_char_list(assoc->qos_list, - row[CLUSTER_REQ_VALID_QOS]+1); - else - list_append(cluster->valid_qos_list, xstrdup("all")); } mysql_free_result(result); @@ -7394,8 +7637,8 @@ list_iterator_destroy(itr); list_iterator_destroy(assoc_itr); if(list_count(assoc_list)) - info("I have %d left over associations", - list_count(assoc_list)); + error("I have %d left over associations", + list_count(assoc_list)); list_destroy(assoc_list); return cluster_list; @@ -7659,13 +7902,6 @@ else assoc->grp_cpu_mins = INFINITE; - /* get the usage if requested */ - if(with_usage) { - acct_storage_p_get_usage(mysql_conn, uid, assoc, - DBD_GET_ASSOC_USAGE, - assoc_cond->usage_start, - assoc_cond->usage_end); - } parent_acct = row[ASSOC_REQ_ACCT]; if(!without_parent_info && row[ASSOC_REQ_PARENT][0]) { @@ -7890,6 +8126,12 @@ //info("parent id is %d", assoc->parent_id); //log_assoc_rec(assoc); } + + if(with_usage && assoc_list) + _get_usage_for_list(mysql_conn, DBD_GET_ASSOC_USAGE, + assoc_list, assoc_cond->usage_start, + assoc_cond->usage_end); + mysql_free_result(result); list_destroy(delta_qos_list); @@ -8252,17 +8494,15 @@ wckey->name = xstrdup(""); wckey->cluster = xstrdup(row[WCKEY_REQ_CLUSTER]); - - /* get the usage if requested */ - if(with_usage) { - acct_storage_p_get_usage(mysql_conn, uid, wckey, - DBD_GET_WCKEY_USAGE, - wckey_cond->usage_start, - wckey_cond->usage_end); - } } mysql_free_result(result); + if(with_usage && wckey_list) + _get_usage_for_list(mysql_conn, DBD_GET_WCKEY_USAGE, + wckey_list, wckey_cond->usage_start, + wckey_cond->usage_end); + + //END_TIMER2("get_wckeys"); return wckey_list; #else @@ -8678,11 +8918,7 @@ List *my_list; uint32_t id = NO_VAL; - char *usage_req_inx[] = { - "t1.id", - "t1.period_start", - "t1.alloc_cpu_secs" - }; + char **usage_req_inx = NULL; enum { USAGE_ID, @@ -8693,17 +8929,35 @@ switch (type) { case DBD_GET_ASSOC_USAGE: + { + char *temp_usage[] = { + "t3.id", + "t1.period_start", + "t1.alloc_cpu_secs" + }; + usage_req_inx = temp_usage; + id = acct_assoc->id; username = acct_assoc->user; my_list = &acct_assoc->accounting_list; my_usage_table = assoc_day_table; break; + } case DBD_GET_WCKEY_USAGE: + { + char *temp_usage[] = { + "id", + "period_start", + "alloc_cpu_secs" + }; + usage_req_inx = temp_usage; + id = acct_wckey->id; username = acct_wckey->user; my_list = &acct_wckey->accounting_list; my_usage_table = wckey_day_table; break; + } default: error("Unknown usage type %d", type); return SLURM_ERROR; @@ -8762,7 +9016,7 @@ if(!acct_assoc->acct) { debug("No account name given " "in association."); - goto bad_user; + goto bad_user; } itr = list_iterator_create(user.coord_accts); @@ -8802,13 +9056,13 @@ "where (t1.period_start < %d && t1.period_start >= %d) " "&& t1.id=t2.id && t3.id=%d && " "t2.lft between t3.lft and t3.rgt " - "order by t1.id, period_start;", + "order by t3.id, period_start;", tmp, my_usage_table, assoc_table, assoc_table, end, start, id); break; case DBD_GET_WCKEY_USAGE: query = xstrdup_printf( - "select %s from %s as t1 " + "select %s from %s " "where (period_start < %d && period_start >= %d) " "&& id=%d order by id, period_start;", tmp, my_usage_table, end, start, id); @@ -9285,12 +9539,16 @@ } debug("%s has changed from %s cpus to %u", cluster, row[0], procs); + /* reset all the entries for this cluster since the procs + changed some of the downed nodes may have gone away. + Request them again with ACCOUNTING_FIRST_REG */ query = xstrdup_printf( "update %s set period_end=%d where cluster=\"%s\" " - "and period_end=0 and node_name=''", + "and period_end=0", event_table, event_time, cluster); rc = mysql_db_query(mysql_conn->db_conn, query); xfree(query); + first = 1; if(rc != SLURM_SUCCESS) goto end_it; add_it: @@ -9418,6 +9676,7 @@ int reinit = 0; time_t check_time = job_ptr->start_time; uint32_t wckeyid = 0; + int no_cluster = 0; if (!job_ptr->details || !job_ptr->details->submit_time) { error("jobacct_storage_p_job_start: " @@ -9446,6 +9705,12 @@ xfree(query); } else slurm_mutex_unlock(&rollup_lock); + + if(!cluster_name && job_ptr->assoc_id) { + no_cluster = 1; + cluster_name = _get_cluster_from_associd(mysql_conn, + job_ptr->assoc_id); + } priority = (job_ptr->priority == NO_VAL) ? @@ -9458,12 +9723,20 @@ /* then grep for " since that is the delimiter for the wckey */ if((temp = strchr(jname, '\"'))) { - /* if we have a wckey set the " to NULL to - * end the jname */ - temp[0] = '\0'; - /* increment and copy the remainder */ - temp++; - wckey = xstrdup(temp); + if(strrchr(jname, '\"') != temp) { + error("job %u has quotes in it's name '%s', " + "no way to get correct wckey", + job_ptr->job_id, jname); + xfree(jname); + jname = _fix_double_quotes(job_ptr->name); + } else { + /* if we have a wckey set the " to NULL to + * end the jname */ + temp[0] = '\0'; + /* increment and copy the remainder */ + temp++; + wckey = xstrdup(temp); + } } } @@ -9498,6 +9771,7 @@ wckeyid = _get_wckeyid(mysql_conn, &wckey, job_ptr->user_id, cluster_name, job_ptr->assoc_id); + /* We need to put a 0 for 'end' incase of funky job state * files from a hot start of the controllers we call @@ -9615,7 +9889,8 @@ xfree(wckey); xfree(query); - + if(no_cluster) + xfree(cluster_name); return rc; #else return SLURM_ERROR; @@ -9631,6 +9906,7 @@ #ifdef HAVE_MYSQL char *query = NULL, *nodes = NULL; int rc=SLURM_SUCCESS; + time_t start_time = job_ptr->start_time; if (!job_ptr->db_index && (!job_ptr->details || !job_ptr->details->submit_time)) { @@ -9649,7 +9925,8 @@ if (job_ptr->end_time == 0) { debug("mysql_jobacct: job %u never started", job_ptr->job_id); return SLURM_SUCCESS; - } + } else if(start_time > job_ptr->end_time) + start_time = 0; slurm_mutex_lock(&rollup_lock); if(job_ptr->end_time < global_last_rollup) { @@ -9692,7 +9969,7 @@ query = xstrdup_printf("update %s set start=%d, end=%d, state=%d, " "nodelist=\"%s\", comp_code=%d, " "kill_requid=%d where id=%d", - job_table, (int)job_ptr->start_time, + job_table, (int)start_time, (int)job_ptr->end_time, job_ptr->job_state & (~JOB_COMPLETING), nodes, job_ptr->exit_code, diff -Nru slurm-llnl-1.3.13/src/plugins/accounting_storage/mysql/mysql_jobacct_process.c slurm-llnl-1.3.15/src/plugins/accounting_storage/mysql/mysql_jobacct_process.c --- slurm-llnl-1.3.13/src/plugins/accounting_storage/mysql/mysql_jobacct_process.c 2009-01-12 20:24:20.000000000 +0100 +++ slurm-llnl-1.3.15/src/plugins/accounting_storage/mysql/mysql_jobacct_process.c 2009-03-16 21:47:03.000000000 +0100 @@ -230,7 +230,7 @@ while((object = list_next(itr))) { if(set) xstrcat(*extra, " || "); - xstrfmtcat(*extra, "t1.gid=", object); + xstrfmtcat(*extra, "t1.gid='%s'", object); set = 1; } list_iterator_destroy(itr); @@ -283,6 +283,13 @@ "(t1.eligible < %d " "&& (t1.end >= %d || t1.end = 0)))", job_cond->usage_end, job_cond->usage_start); + } else if(job_cond->usage_end) { + if(*extra) + xstrcat(*extra, " && ("); + else + xstrcat(*extra, " where ("); + xstrfmtcat(*extra, + "(t1.eligible < %d))", job_cond->usage_end); } if(job_cond->state_list && list_count(job_cond->state_list)) { @@ -600,7 +607,10 @@ easy to look for duplicates */ if(job_cond && !job_cond->duplicates) - xstrcat(query, " order by jobid, submit desc"); + xstrcat(query, " order by t1.cluster, jobid, submit desc"); + else + xstrcat(query, " order by t1.cluster, submit desc"); + debug3("%d(%d) query\n%s", mysql_conn->conn, __LINE__, query); if(!(result = mysql_db_query_ret( @@ -735,6 +745,9 @@ job->elapsed -= job->suspended; } + if((int)job->elapsed < 0) + job->elapsed = 0; + job->jobid = curr_id; job->jobname = xstrdup(row[JOB_REQ_NAME]); job->gid = atoi(row[JOB_REQ_GID]); @@ -851,6 +864,10 @@ step->elapsed = step->end - step->start; } step->elapsed -= step->suspended; + + if((int)step->elapsed < 0) + step->elapsed = 0; + step->user_cpu_sec = atoi(step_row[STEP_REQ_USER_SEC]); step->user_cpu_usec = atoi(step_row[STEP_REQ_USER_USEC]); @@ -915,6 +932,8 @@ job->track_steps = 1; } } + /* need to reset here to make the above test valid */ + step = NULL; } mysql_free_result(result); @@ -1246,10 +1265,11 @@ if (rc) (void) unlink(new_file); else { /* file shuffle */ + int ign; /* avoid warning */ (void) unlink(old_file); - (void) link(reg_file, old_file); + ign = link(reg_file, old_file); (void) unlink(reg_file); - (void) link(new_file, reg_file); + ign = link(new_file, reg_file); (void) unlink(new_file); } xfree(old_file); @@ -1432,10 +1452,11 @@ if (rc) (void) unlink(new_file); else { /* file shuffle */ + int ign; /* avoid warning */ (void) unlink(old_file); - (void) link(reg_file, old_file); + ign = link(reg_file, old_file); (void) unlink(reg_file); - (void) link(new_file, reg_file); + ign = link(new_file, reg_file); (void) unlink(new_file); } xfree(old_file); @@ -1520,7 +1541,7 @@ } debug3("%d(%d) query\n%s", mysql_conn->conn, __LINE__, data); - error_code = mysql_db_query(mysql_conn->db_conn, data); + error_code = mysql_db_query_check_after(mysql_conn->db_conn, data); xfree(data); if(error_code != SLURM_SUCCESS) { error("Couldn't load old data"); diff -Nru slurm-llnl-1.3.13/src/plugins/accounting_storage/mysql/mysql_rollup.c slurm-llnl-1.3.15/src/plugins/accounting_storage/mysql/mysql_rollup.c --- slurm-llnl-1.3.13/src/plugins/accounting_storage/mysql/mysql_rollup.c 2009-01-13 18:31:44.000000000 +0100 +++ slurm-llnl-1.3.15/src/plugins/accounting_storage/mysql/mysql_rollup.c 2009-02-09 23:43:51.000000000 +0100 @@ -304,7 +304,7 @@ "|| end = 0)) " "order by associd, eligible", job_str, job_table, - curr_end, curr_start, curr_start); + curr_end, curr_start); debug3("%d(%d) query\n%s", mysql_conn->conn, __LINE__, query); if(!(result = mysql_db_query_ret( diff -Nru slurm-llnl-1.3.13/src/plugins/accounting_storage/pgsql/accounting_storage_pgsql.c slurm-llnl-1.3.15/src/plugins/accounting_storage/pgsql/accounting_storage_pgsql.c --- slurm-llnl-1.3.13/src/plugins/accounting_storage/pgsql/accounting_storage_pgsql.c 2008-12-22 22:46:26.000000000 +0100 +++ slurm-llnl-1.3.15/src/plugins/accounting_storage/pgsql/accounting_storage_pgsql.c 2009-03-25 20:26:12.000000000 +0100 @@ -1164,7 +1164,7 @@ { #ifdef HAVE_PGSQL int rc=SLURM_SUCCESS; - char *jname, *nodes; + char *jname = NULL, *nodes; long priority; int track_steps = 0; char *block_id = NULL; diff -Nru slurm-llnl-1.3.13/src/plugins/accounting_storage/pgsql/pgsql_jobacct_process.c slurm-llnl-1.3.15/src/plugins/accounting_storage/pgsql/pgsql_jobacct_process.c --- slurm-llnl-1.3.13/src/plugins/accounting_storage/pgsql/pgsql_jobacct_process.c 2008-12-22 22:46:26.000000000 +0100 +++ slurm-llnl-1.3.15/src/plugins/accounting_storage/pgsql/pgsql_jobacct_process.c 2009-01-15 00:59:25.000000000 +0100 @@ -700,6 +700,8 @@ job->track_steps = 1; } } + /* need to reset here to make the above test valid */ + step = NULL; } PQclear(result); diff -Nru slurm-llnl-1.3.13/src/plugins/accounting_storage/slurmdbd/accounting_storage_slurmdbd.c slurm-llnl-1.3.15/src/plugins/accounting_storage/slurmdbd/accounting_storage_slurmdbd.c --- slurm-llnl-1.3.13/src/plugins/accounting_storage/slurmdbd/accounting_storage_slurmdbd.c 2008-12-22 22:46:26.000000000 +0100 +++ slurm-llnl-1.3.15/src/plugins/accounting_storage/slurmdbd/accounting_storage_slurmdbd.c 2009-03-23 21:37:54.000000000 +0100 @@ -1279,6 +1279,7 @@ { slurmdbd_msg_t msg; dbd_cluster_procs_msg_t req; + int rc = SLURM_ERROR; debug2("Sending info for cluster %s", cluster); req.cluster_name = cluster; @@ -1287,10 +1288,9 @@ msg.msg_type = DBD_CLUSTER_PROCS; msg.data = &req; - if (slurm_send_slurmdbd_msg(SLURMDBD_VERSION, &msg) < 0) - return SLURM_ERROR; + slurm_send_slurmdbd_recv_rc_msg(SLURMDBD_VERSION, &msg, &rc); - return SLURM_SUCCESS; + return rc; } extern int clusteracct_storage_p_register_ctld(void *db_conn, diff -Nru slurm-llnl-1.3.13/src/plugins/proctrack/rms/proctrack_rms.c slurm-llnl-1.3.15/src/plugins/proctrack/rms/proctrack_rms.c --- slurm-llnl-1.3.13/src/plugins/proctrack/rms/proctrack_rms.c 2008-03-20 00:10:58.000000000 +0100 +++ slurm-llnl-1.3.15/src/plugins/proctrack/rms/proctrack_rms.c 2009-03-02 18:53:23.000000000 +0100 @@ -93,7 +93,7 @@ /* * Return a handle to an existing prgid or create a new one */ - if (rms_getprgid (job->jmgr_pid, &prgid) < 0) { + if (rms_getprgid ((int) job->jmgr_pid, &prgid) < 0) { int fd = _prg_destructor_fork(); /* Use slurmd job-step manager's pid as a unique identifier */ prgid = job->jmgr_pid; diff -Nru slurm-llnl-1.3.13/src/plugins/sched/backfill/backfill_wrapper.c slurm-llnl-1.3.15/src/plugins/sched/backfill/backfill_wrapper.c --- slurm-llnl-1.3.13/src/plugins/sched/backfill/backfill_wrapper.c 2008-09-09 19:26:00.000000000 +0200 +++ slurm-llnl-1.3.15/src/plugins/sched/backfill/backfill_wrapper.c 2009-03-25 19:22:05.000000000 +0100 @@ -75,7 +75,7 @@ } slurm_attr_init( &attr ); - pthread_attr_setdetachstate( &attr, PTHREAD_CREATE_DETACHED ); + /* since we do a join on this later we don't make it detached */ if (pthread_create( &backfill_thread, &attr, backfill_agent, NULL)) error("Unable to start backfill thread: %m"); pthread_mutex_unlock( &thread_flag_mutex ); @@ -93,6 +93,7 @@ if ( backfill_thread ) { verbose( "Backfill scheduler plugin shutting down" ); stop_backfill_agent(); + pthread_join(backfill_thread, NULL); backfill_thread = 0; } pthread_mutex_unlock( &thread_flag_mutex ); diff -Nru slurm-llnl-1.3.13/src/plugins/sched/wiki/job_modify.c slurm-llnl-1.3.15/src/plugins/sched/wiki/job_modify.c --- slurm-llnl-1.3.13/src/plugins/sched/wiki/job_modify.c 2008-09-03 21:12:46.000000000 +0200 +++ slurm-llnl-1.3.15/src/plugins/sched/wiki/job_modify.c 2009-03-20 17:07:52.000000000 +0100 @@ -40,6 +40,7 @@ #include "src/slurmctld/job_scheduler.h" #include "src/slurmctld/locks.h" #include "src/slurmctld/slurmctld.h" +#include "src/common/slurm_accounting_storage.h" extern void null_term(char *str) { @@ -60,6 +61,7 @@ uint32_t new_time_limit) { struct job_record *job_ptr; + bool update_accounting = false; job_ptr = find_job_record(jobid); if (job_ptr == NULL) { @@ -96,9 +98,12 @@ last_job_update = time(NULL); } - if (bank_ptr && - (update_job_account("wiki", job_ptr, bank_ptr) != SLURM_SUCCESS)) { - return EINVAL; + if (bank_ptr) { + if(update_job_account("wiki", job_ptr, bank_ptr) + != SLURM_SUCCESS) + return EINVAL; + else + update_accounting = true; } if (new_hostlist) { @@ -147,11 +152,14 @@ } host_fini: if (rc) { - info("wiki: change job %u invalid hostlist %s", jobid, new_hostlist); + info("wiki: change job %u invalid hostlist %s", + jobid, new_hostlist); xfree(job_ptr->details->req_nodes); return EINVAL; } else { - info("wiki: change job %u hostlist %s", jobid, new_hostlist); + info("wiki: change job %u hostlist %s", + jobid, new_hostlist); + update_accounting = true; } } @@ -169,6 +177,7 @@ job_ptr->partition = xstrdup(part_name_ptr); job_ptr->part_ptr = part_ptr; last_job_update = time(NULL); + update_accounting = true; } if (new_node_cnt) { if (IS_JOB_PENDING(job_ptr) && job_ptr->details) { @@ -179,6 +188,7 @@ info("wiki: change job %u min_nodes to %u", jobid, new_node_cnt); last_job_update = time(NULL); + update_accounting = true; } else { error("wiki: MODIFYJOB node count of non-pending " "job %u", jobid); @@ -186,6 +196,14 @@ } } + if(update_accounting) { + if (job_ptr->details && job_ptr->details->begin_time) { + /* Update job record in accounting to reflect changes */ + jobacct_storage_g_job_start( + acct_db_conn, slurmctld_cluster_name, job_ptr); + } + } + return SLURM_SUCCESS; } diff -Nru slurm-llnl-1.3.13/src/plugins/sched/wiki/msg.c slurm-llnl-1.3.15/src/plugins/sched/wiki/msg.c --- slurm-llnl-1.3.13/src/plugins/sched/wiki/msg.c 2008-12-10 20:44:46.000000000 +0100 +++ slurm-llnl-1.3.15/src/plugins/sched/wiki/msg.c 2009-03-16 22:52:29.000000000 +0100 @@ -474,7 +474,7 @@ return NULL; } - debug("wiki msg recv:%s", buf); + debug2("wiki msg recv:%s", buf); return buf; } @@ -488,7 +488,7 @@ char header[10]; size_t data_sent; - debug("wiki msg send:%s", buf); + debug2("wiki msg send:%s", buf); (void) sprintf(header, "%08lu\n", (unsigned long) size); if (_write_bytes((int) new_fd, header, 9) != 9) { diff -Nru slurm-llnl-1.3.13/src/plugins/sched/wiki/start_job.c slurm-llnl-1.3.15/src/plugins/sched/wiki/start_job.c --- slurm-llnl-1.3.13/src/plugins/sched/wiki/start_job.c 2008-04-21 20:24:03.000000000 +0200 +++ slurm-llnl-1.3.15/src/plugins/sched/wiki/start_job.c 2009-02-13 19:04:40.000000000 +0100 @@ -185,6 +185,20 @@ goto fini; } + if (!bit_super_set(new_bitmap, avail_node_bitmap)) { + /* Selected node is UP and not responding + * or it just went DOWN */ + *err_code = -700; + *err_msg = "TASKLIST includes non-responsive node"; + error("wiki: Attempt to use non-responsive nodes for " + "job %u, %s", + jobid, hostlist); + xfree(new_node_list); + bit_free(new_bitmap); + rc = -1; + goto fini; + } + /* User excluded node list incompatable with Wiki * Exclude all nodes not explicitly requested */ FREE_NULL_BITMAP(job_ptr->details->exc_node_bitmap); diff -Nru slurm-llnl-1.3.13/src/plugins/sched/wiki2/get_jobs.c slurm-llnl-1.3.15/src/plugins/sched/wiki2/get_jobs.c --- slurm-llnl-1.3.13/src/plugins/sched/wiki2/get_jobs.c 2008-12-15 17:58:00.000000000 +0100 +++ slurm-llnl-1.3.15/src/plugins/sched/wiki2/get_jobs.c 2009-03-20 17:06:08.000000000 +0100 @@ -439,6 +439,13 @@ field_sep = "?"; } + /* JOBFLAGS:RESTARTABLE */ + if (job_ptr->details && job_ptr->details->requeue) { + size += snprintf((buffer + size), (buf_size - size), + "%sJOBFLAGS:RESTARTABLE", field_sep); + field_sep = "?"; + } + /* COMMENT SET BY MOAB */ if (job_ptr->comment && job_ptr->comment[0]) { size += snprintf((buffer + size), (buf_size - size), @@ -487,7 +494,7 @@ return max_nodes; /* should never reach here */ if (job_ptr->details->max_nodes) { - max_nodes = job_ptr->details->max_nodes; + max_nodes = job_ptr->details->max_nodes; if (job_ptr->part_ptr->max_nodes != INFINITE) { max_nodes = MIN(max_nodes, job_ptr->part_ptr->max_nodes); diff -Nru slurm-llnl-1.3.13/src/plugins/sched/wiki2/job_modify.c slurm-llnl-1.3.15/src/plugins/sched/wiki2/job_modify.c --- slurm-llnl-1.3.13/src/plugins/sched/wiki2/job_modify.c 2008-12-02 23:22:43.000000000 +0100 +++ slurm-llnl-1.3.15/src/plugins/sched/wiki2/job_modify.c 2009-03-20 17:07:52.000000000 +0100 @@ -40,6 +40,7 @@ #include "src/slurmctld/job_scheduler.h" #include "src/slurmctld/locks.h" #include "src/slurmctld/slurmctld.h" +#include "src/common/slurm_accounting_storage.h" /* Given a string, replace the first space found with '\0' */ extern void null_term(char *str) @@ -63,6 +64,7 @@ { struct job_record *job_ptr; time_t now = time(NULL); + bool update_accounting = false; job_ptr = find_job_record(jobid); if (job_ptr == NULL) { @@ -127,6 +129,7 @@ jobid, begin_time); job_ptr->details->begin_time = begin_time; last_job_update = now; + update_accounting = true; } else { error("wiki: MODIFYJOB begin_time of non-pending " "job %u", jobid); @@ -140,6 +143,7 @@ xfree(job_ptr->name); job_ptr->name = xstrdup(name_ptr); last_job_update = now; + update_accounting = true; } else { error("wiki: MODIFYJOB name of non-pending job %u", jobid); @@ -200,6 +204,7 @@ } else { info("wiki: change job %u hostlist %s", jobid, new_hostlist); + update_accounting = true; } } @@ -217,6 +222,7 @@ job_ptr->partition = xstrdup(part_name_ptr); job_ptr->part_ptr = part_ptr; last_job_update = now; + update_accounting = true; } if (new_node_cnt) { @@ -228,6 +234,7 @@ info("wiki: change job %u min_nodes to %u", jobid, new_node_cnt); last_job_update = now; + update_accounting = true; } else { error("wiki: MODIFYJOB node count of non-pending " "job %u", jobid); @@ -235,6 +242,14 @@ } } + if(update_accounting) { + if (job_ptr->details && job_ptr->details->begin_time) { + /* Update job record in accounting to reflect changes */ + jobacct_storage_g_job_start( + acct_db_conn, slurmctld_cluster_name, job_ptr); + } + } + return SLURM_SUCCESS; } diff -Nru slurm-llnl-1.3.13/src/plugins/sched/wiki2/job_requeue.c slurm-llnl-1.3.15/src/plugins/sched/wiki2/job_requeue.c --- slurm-llnl-1.3.13/src/plugins/sched/wiki2/job_requeue.c 2008-03-20 00:10:58.000000000 +0100 +++ slurm-llnl-1.3.15/src/plugins/sched/wiki2/job_requeue.c 2009-03-27 20:35:49.000000000 +0100 @@ -44,6 +44,7 @@ { char *arg_ptr, *tmp_char; uint32_t jobid; + struct job_record *job_ptr; static char reply_msg[128]; int slurm_rc; /* Write lock on job and node info */ @@ -67,24 +68,23 @@ lock_slurmctld(job_write_lock); slurm_rc = job_requeue(0, jobid, -1); - if (slurm_rc == SLURM_SUCCESS) { - /* We need to clear the required node list here. - * If the job was submitted with srun and a - * required node list, it gets lost here. */ - struct job_record *job_ptr; - job_ptr = find_job_record(jobid); - if (job_ptr && job_ptr->details) { - xfree(job_ptr->details->req_nodes); - FREE_NULL_BITMAP(job_ptr->details-> - req_node_bitmap); - } - info("wiki: requeued job %u", jobid); - } else { + if (slurm_rc != SLURM_SUCCESS) { + unlock_slurmctld(job_write_lock); *err_code = -700; *err_msg = slurm_strerror(slurm_rc); error("wiki: Failed to requeue job %u (%m)", jobid); return -1; } + + /* We need to clear the required node list here. + * If the job was submitted with srun and a + * required node list, it gets lost here. */ + job_ptr = find_job_record(jobid); + if (job_ptr && job_ptr->details) { + xfree(job_ptr->details->req_nodes); + FREE_NULL_BITMAP(job_ptr->details->req_node_bitmap); + } + info("wiki: requeued job %u", jobid); unlock_slurmctld(job_write_lock); snprintf(reply_msg, sizeof(reply_msg), "job %u requeued successfully", jobid); diff -Nru slurm-llnl-1.3.13/src/plugins/sched/wiki2/job_will_run.c slurm-llnl-1.3.15/src/plugins/sched/wiki2/job_will_run.c --- slurm-llnl-1.3.13/src/plugins/sched/wiki2/job_will_run.c 2008-05-15 22:48:32.000000000 +0200 +++ slurm-llnl-1.3.15/src/plugins/sched/wiki2/job_will_run.c 2009-01-20 19:49:55.000000000 +0100 @@ -234,7 +234,8 @@ } if (job_ptr->details->exc_node_bitmap) { bit_not(job_ptr->details->exc_node_bitmap); - bit_and(avail_bitmap, job_ptr->details->exc_node_bitmap); + bit_and(avail_bitmap, + job_ptr->details->exc_node_bitmap); bit_not(job_ptr->details->exc_node_bitmap); } if ((job_ptr->details->req_node_bitmap) && diff -Nru slurm-llnl-1.3.13/src/plugins/sched/wiki2/msg.c slurm-llnl-1.3.15/src/plugins/sched/wiki2/msg.c --- slurm-llnl-1.3.13/src/plugins/sched/wiki2/msg.c 2008-12-10 20:44:46.000000000 +0100 +++ slurm-llnl-1.3.15/src/plugins/sched/wiki2/msg.c 2009-03-19 20:01:45.000000000 +0100 @@ -43,6 +43,11 @@ #define _DEBUG 0 +/* When a remote socket closes on AIX, we have seen poll() return EAGAIN + * indefinitely for a pending write request. Rather than locking up + * slurmctld's wiki interface, abort after MAX_RETRIES poll() failures. */ +#define MAX_RETRIES 10 + static bool thread_running = false; static bool thread_shutdown = false; static pthread_mutex_t thread_flag_mutex = PTHREAD_MUTEX_INITIALIZER; @@ -433,8 +438,8 @@ rc = poll(&ufds, 1, 10000); /* 10 sec timeout */ if (rc == 0) /* timed out */ break; - if ((rc == -1) /* some error */ - && ((errno== EINTR) || (errno == EAGAIN))) + if ((rc == -1) && /* some error */ + ((errno== EINTR) || (errno == EAGAIN))) continue; if ((ufds.revents & POLLIN) == 0) /* some poll error */ break; @@ -455,7 +460,7 @@ size_t bytes_remaining, bytes_written; char *ptr; struct pollfd ufds; - int rc; + int rc, retry_cnt = 0; bytes_remaining = size; size = 0; @@ -466,9 +471,15 @@ rc = poll(&ufds, 1, 10000); /* 10 sec timeout */ if (rc == 0) /* timed out */ break; - if ((rc == -1) /* some error */ - && ((errno== EINTR) || (errno == EAGAIN))) + if ((rc == -1) && /* some error */ + ((errno== EINTR) || (errno == EAGAIN))) { + if ((retry_cnt++) >= MAX_RETRIES) { + error("wiki: repeated poll errors for " + "write: %m"); + break; + } continue; + } if ((ufds.revents & POLLOUT) == 0) /* some poll error */ break; @@ -518,7 +529,7 @@ return NULL; } - debug("wiki msg recv:%s", buf); + debug2("wiki msg recv:%s", buf); return buf; } @@ -532,7 +543,7 @@ char header[10]; size_t data_sent; - debug("wiki msg send:%s", buf); + debug2("wiki msg send:%s", buf); (void) sprintf(header, "%08lu\n", (unsigned long) size); if (_write_bytes((int) new_fd, header, 9) != 9) { diff -Nru slurm-llnl-1.3.13/src/plugins/sched/wiki2/start_job.c slurm-llnl-1.3.15/src/plugins/sched/wiki2/start_job.c --- slurm-llnl-1.3.13/src/plugins/sched/wiki2/start_job.c 2008-04-21 19:54:48.000000000 +0200 +++ slurm-llnl-1.3.15/src/plugins/sched/wiki2/start_job.c 2009-02-13 19:04:40.000000000 +0100 @@ -239,6 +239,20 @@ goto fini; } + if (!bit_super_set(new_bitmap, avail_node_bitmap)) { + /* Selected node is UP and not responding + * or it just went DOWN */ + *err_code = -700; + *err_msg = "TASKLIST includes non-responsive node"; + error("wiki: Attempt to use non-responsive nodes for " + "job %u, %s", + jobid, hostlist); + xfree(new_node_list); + bit_free(new_bitmap); + rc = -1; + goto fini; + } + /* User excluded node list incompatable with Wiki * Exclude all nodes not explicitly requested */ FREE_NULL_BITMAP(job_ptr->details->exc_node_bitmap); diff -Nru slurm-llnl-1.3.13/src/plugins/select/bluegene/block_allocator/block_allocator.c slurm-llnl-1.3.15/src/plugins/select/bluegene/block_allocator/block_allocator.c --- slurm-llnl-1.3.13/src/plugins/select/bluegene/block_allocator/block_allocator.c 2008-12-29 22:56:17.000000000 +0100 +++ slurm-llnl-1.3.15/src/plugins/select/bluegene/block_allocator/block_allocator.c 2009-04-10 21:25:52.000000000 +0200 @@ -1,7 +1,7 @@ /*****************************************************************************\ * block_allocator.c - Assorted functions for layout of bluegene blocks, * wiring, mapping for smap, etc. - * $Id: block_allocator.c 16088 2008-12-29 21:56:17Z jette $ + * $Id: block_allocator.c 17225 2009-04-10 19:25:52Z da $ ***************************************************************************** * Copyright (C) 2004 The Regents of the University of California. * Produced at Lawrence Livermore National Laboratory (cf, DISCLAIMER). @@ -62,16 +62,20 @@ List best_path = NULL; int best_count; int color_count = 0; -bool *passthrough = NULL; +uint16_t *deny_pass = NULL; /* extern Global */ +my_bluegene_t *bg = NULL; +uint16_t ba_deny_pass = 0; List bp_map_list = NULL; char letters[62]; char colors[6]; #ifdef HAVE_3D int DIM_SIZE[BA_SYSTEM_DIMENSIONS] = {0,0,0}; +int REAL_DIM_SIZE[BA_SYSTEM_DIMENSIONS] = {0,0,0}; #else int DIM_SIZE[BA_SYSTEM_DIMENSIONS] = {0}; +int REAL_DIM_SIZE[BA_SYSTEM_DIMENSIONS] = {0}; #endif s_p_options_t bg_conf_file_options[] = { @@ -88,6 +92,7 @@ {"AltCnloadImage", S_P_ARRAY, parse_image, NULL}, {"AltIoloadImage", S_P_ARRAY, parse_image, NULL}, #endif + {"DenyPassthrough", S_P_STRING}, {"LayoutMode", S_P_STRING}, {"MloaderImage", S_P_STRING}, {"BridgeAPILogFile", S_P_STRING}, @@ -206,6 +211,7 @@ /* */ static void _destroy_geo(void *object); + extern char *bg_block_state_string(rm_partition_state_t state) { static char tmp[16]; @@ -238,6 +244,28 @@ return tmp; } +extern char *ba_passthroughs_string(uint16_t passthrough) +{ + char *pass = NULL; + if(passthrough & PASS_FOUND_X) + xstrcat(pass, "X"); + if(passthrough & PASS_FOUND_Y) { + if(pass) + xstrcat(pass, ",Y"); + else + xstrcat(pass, "Y"); + } + if(passthrough & PASS_FOUND_Z) { + if(pass) + xstrcat(pass, ",Z"); + else + xstrcat(pass, "Z"); + } + + return pass; +} + + extern int parse_blockreq(void **dest, slurm_parser_enum_t type, const char *key, const char *value, const char *line, char **leftover) @@ -485,8 +513,11 @@ geo[X] = ba_request->geometry[X]; geo[Y] = ba_request->geometry[Y]; geo[Z] = ba_request->geometry[Z]; - passthrough = &ba_request->passthrough; - + if(ba_request->deny_pass == (uint16_t)NO_VAL) + ba_request->deny_pass = ba_deny_pass; + + deny_pass = &ba_request->deny_pass; + if(geo[X] != (uint16_t)NO_VAL) { for (i=0; ielongate_geos, ba_request->rotate); } - startagain: +// startagain: picked=0; for(i=0;i<8;i++) checked[i]=0; @@ -617,9 +648,21 @@ break; } } + /* This size can not be made into a + block return. If you want to try + until we find the next largest block + uncomment the code below and the goto + above. If a user specifies a max + node count the job will never + run. + */ if(i2==1) { - ba_request->size +=1; - goto startagain; + error("Can't make a block of " + "%d into a cube.", + ba_request->size); + return 0; +/* ba_request->size +=1; */ +/* goto startagain; */ } } else { @@ -856,7 +899,6 @@ int end[BA_SYSTEM_DIMENSIONS]; #ifdef HAVE_BG_FILES - my_bluegene_t *bg = NULL; rm_size3D_t bp_size; int rc = 0; #endif /* HAVE_BG_FILES */ @@ -943,6 +985,10 @@ DIM_SIZE[X]++; DIM_SIZE[Y]++; DIM_SIZE[Z]++; + /* this will probably be reset below */ + REAL_DIM_SIZE[X] = DIM_SIZE[X]; + REAL_DIM_SIZE[Y] = DIM_SIZE[Y]; + REAL_DIM_SIZE[Z] = DIM_SIZE[Z]; #else DIM_SIZE[X] = node_info_ptr->record_count; #endif @@ -951,28 +997,6 @@ #ifdef HAVE_3D node_info_error: -#ifdef HAVE_BG_FILES - if (have_db2 - && ((DIM_SIZE[X]==0) || (DIM_SIZE[Y]==0) || (DIM_SIZE[Z]==0))) { - if ((rc = bridge_get_bg(&bg)) != STATUS_OK) { - error("bridge_get_BG(): %d", rc); - return; - } - - if ((bg != NULL) - && ((rc = bridge_get_data(bg, RM_Msize, &bp_size)) - == STATUS_OK)) { - DIM_SIZE[X]=bp_size.X; - DIM_SIZE[Y]=bp_size.Y; - DIM_SIZE[Z]=bp_size.Z; - } else { - error("bridge_get_data(RM_Msize): %d", rc); - } - if ((rc = bridge_free_bg(bg)) != STATUS_OK) - error("bridge_free_BG(): %d", rc); - } -#endif - if ((DIM_SIZE[X]==0) || (DIM_SIZE[Y]==0) || (DIM_SIZE[Z]==0)) { debug("Setting dimensions from slurm.conf file"); count = slurm_conf_nodename_array(&ptr_array); @@ -1025,9 +1049,58 @@ DIM_SIZE[X]++; DIM_SIZE[Y]++; DIM_SIZE[Z]++; + /* this will probably be reset below */ + REAL_DIM_SIZE[X] = DIM_SIZE[X]; + REAL_DIM_SIZE[Y] = DIM_SIZE[Y]; + REAL_DIM_SIZE[Z] = DIM_SIZE[Z]; } - debug("DIM_SIZE = %c%c%c\n", alpha_num[DIM_SIZE[X]], - alpha_num[DIM_SIZE[Y]], alpha_num[DIM_SIZE[Z]]); +#ifdef HAVE_BG_FILES + /* sanity check. We can only request part of the system, but + we don't want to allow more than we have. */ + if (have_db2) { + verbose("Attempting to contact MMCS"); + if ((rc = bridge_get_bg(&bg)) != STATUS_OK) { + fatal("bridge_get_BG() failed. This usually means " + "there is something wrong with the database. " + "You might want to run slurmctld in daemon " + "mode (-D) to see what the real error from " + "the api was. The return code was %d", rc); + return; + } + + if ((bg != NULL) + && ((rc = bridge_get_data(bg, RM_Msize, &bp_size)) + == STATUS_OK)) { + verbose("BlueGene configured with " + "%d x %d x %d base blocks", + bp_size.X, bp_size.Y, bp_size.Z); + REAL_DIM_SIZE[X] = bp_size.X; + REAL_DIM_SIZE[Y] = bp_size.Y; + REAL_DIM_SIZE[Z] = bp_size.Z; + if((DIM_SIZE[X] > bp_size.X) + || (DIM_SIZE[Y] > bp_size.Y) + || (DIM_SIZE[Z] > bp_size.Z)) { + fatal("You requested a %c%c%c system, " + "but we only have a system of %c%c%c. " + "Change your slurm.conf.", + alpha_num[DIM_SIZE[X]], + alpha_num[DIM_SIZE[Y]], + alpha_num[DIM_SIZE[Z]], + alpha_num[bp_size.X], + alpha_num[bp_size.Y], + alpha_num[bp_size.Z]); + } + } else { + error("bridge_get_data(RM_Msize): %d", rc); + } + } +#endif + + + debug("We are using %c x %c x %c of the system.", + alpha_num[DIM_SIZE[X]], + alpha_num[DIM_SIZE[Y]], + alpha_num[DIM_SIZE[Z]]); #else if (DIM_SIZE[X]==0) { @@ -1120,6 +1193,9 @@ best_path = NULL; } #ifdef HAVE_BG_FILES + if(bg) + bridge_free_bg(bg); + if (bp_map_list) { list_destroy(bp_map_list); bp_map_list = NULL; @@ -1529,13 +1605,23 @@ /* This midplane should have already been checked if it was in use or not */ list_append(results, ba_node); - if(conn_type == SELECT_SMALL) { + if(conn_type >= SELECT_SMALL) { /* adding the ba_node and ending */ ba_node->used = true; name = xstrdup_printf("%c%c%c", alpha_num[ba_node->coord[X]], alpha_num[ba_node->coord[Y]], alpha_num[ba_node->coord[Z]]); + if(ba_node->letter == '.') { + ba_node->letter = letters[color_count%62]; + ba_node->color = colors[color_count%6]; + debug3("count %d setting letter = %c " + "color = %d", + color_count, + ba_node->letter, + ba_node->color); + color_count++; + } goto end_it; } found = _find_x_path(results, ba_node, @@ -1959,7 +2045,6 @@ extern int set_bp_map(void) { #ifdef HAVE_BG_FILES - static my_bluegene_t *bg = NULL; int rc; rm_BP_t *my_bp = NULL; ba_bp_map_t *bp_map = NULL; @@ -1978,15 +2063,19 @@ return -1; } +#ifdef HAVE_BGL if (!getenv("DB2INSTANCE") || !getenv("VWSPATH")) { - fatal("Missing DB2INSTANCE or VWSPATH env var." - "Execute 'db2profile'"); + fatal("Missing DB2INSTANCE or VWSPATH env var. " + "Execute 'db2profile'"); return -1; } +#endif - if ((rc = bridge_get_bg(&bg)) != STATUS_OK) { - error("bridge_get_BG(): %d", rc); - return -1; + if (!bg) { + if((rc = bridge_get_bg(&bg)) != STATUS_OK) { + error("bridge_get_BG(): %d", rc); + return -1; + } } if ((rc = bridge_get_data(bg, RM_BPNum, &bp_num)) != STATUS_OK) { @@ -2052,10 +2141,6 @@ free(bp_id); } - - if ((rc = bridge_free_bg(bg)) != STATUS_OK) - error("bridge_free_BG(): %s", rc); - #endif _bp_map_initialized = true; return 1; @@ -2070,17 +2155,40 @@ #ifdef HAVE_BG_FILES ba_bp_map_t *bp_map = NULL; ListIterator itr; - + char *check = NULL; + if(!bp_map_list) { if(set_bp_map() == -1) return NULL; } + + check = xstrdup(bp_id); + /* with BGP they changed the names of the rack midplane action from + * R000 to R00-M0 so we now support both formats for each of the + * systems */ +#ifdef HAVE_BGL + if(check[3] == '-') { + if(check[5]) { + check[3] = check[5]; + check[4] = '\0'; + } + } +#else + if(check[3] != '-') { + xfree(check); + check = xstrdup_printf("R%c%c-M%c", + bp_id[1], bp_id[2], bp_id[3]); + } +#endif + itr = list_iterator_create(bp_map_list); - while ((bp_map = list_next(itr)) != NULL) - if (!strcasecmp(bp_map->bp_id, bp_id)) + while ((bp_map = list_next(itr))) + if (!strcasecmp(bp_map->bp_id, check)) break; /* we found it */ - list_iterator_destroy(itr); + + xfree(check); + if(bp_map != NULL) return bp_map->coord; else @@ -2599,6 +2707,42 @@ } +/* */ +extern int validate_coord(int *coord) +{ +#ifdef HAVE_BG_FILES + if(coord[X]>=REAL_DIM_SIZE[X] + || coord[Y]>=REAL_DIM_SIZE[Y] + || coord[Z]>=REAL_DIM_SIZE[Z]) { + error("got coord %c%c%c greater than system dims " + "%c%c%c", + alpha_num[coord[X]], + alpha_num[coord[Y]], + alpha_num[coord[Z]], + alpha_num[REAL_DIM_SIZE[X]], + alpha_num[REAL_DIM_SIZE[Y]], + alpha_num[REAL_DIM_SIZE[Z]]); + return 0; + } + + if(coord[X]>=DIM_SIZE[X] + || coord[Y]>=DIM_SIZE[Y] + || coord[Z]>=DIM_SIZE[Z]) { + debug4("got coord %c%c%c greater than what we are using " + "%c%c%c", + alpha_num[coord[X]], + alpha_num[coord[Y]], + alpha_num[coord[Z]], + alpha_num[DIM_SIZE[X]], + alpha_num[DIM_SIZE[Y]], + alpha_num[DIM_SIZE[Z]]); + return 0; + } +#endif + return 1; +} + + /********************* Local Functions *********************/ #ifdef HAVE_BG @@ -2727,7 +2871,9 @@ if(rotate) { for (i = (BA_SYSTEM_DIMENSIONS - 1); i >= 0; i--) { for (j = 1; j <= i; j++) { - if (geometry[j-1] > geometry[j]) { + if ((geometry[j-1] > geometry[j]) + && (geometry[j] <= DIM_SIZE[j-i]) + && (geometry[j-1] <= DIM_SIZE[j])) { temp_geo = geometry[j-1]; geometry[j-1] = geometry[j]; geometry[j] = temp_geo; @@ -2760,7 +2906,7 @@ /* * Fill in the paths and extra midplanes we need for the block. - * Basically copy the x path sent in with the start_list in each Y anx + * Basically copy the x path sent in with the start_list in each Y and * Z dimension filling in every midplane for the block and then * completing the Y and Z wiring, tying the whole block together. * @@ -2795,14 +2941,12 @@ curr_switch = &check_node->axis_switch[X]; for(y=0; ycoord[Y]+y) - >= DIM_SIZE[Y]) { + if((check_node->coord[Y]+y) >= DIM_SIZE[Y]) { rc = 0; goto failed; } for(z=0; zcoord[Z]+z) - >= DIM_SIZE[Z]) { + if((check_node->coord[Z]+z) >= DIM_SIZE[Z]) { rc = 0; goto failed; } @@ -2853,6 +2997,19 @@ goto failed; } } + + if(deny_pass) { + if((*deny_pass & PASS_DENY_Y) + && (*deny_pass & PASS_FOUND_Y)) { + debug("We don't allow Y passthoughs"); + rc = 0; + } else if((*deny_pass & PASS_DENY_Z) + && (*deny_pass & PASS_FOUND_Z)) { + debug("We don't allow Z passthoughs"); + rc = 0; + } + } + failed: list_iterator_destroy(itr); @@ -3056,7 +3213,16 @@ dim_curr_switch->int_wire[2].port_tar = 0; dim_curr_switch = dim_next_switch; - + + if(deny_pass + && (node_tar[i2] != first[i2])) { + if(i2 == 1) + *deny_pass |= + PASS_FOUND_Y; + else + *deny_pass |= + PASS_FOUND_Z; + } while(node_tar[i2] != first[i2]) { debug3("on dim %d at %d " "looking for %d", @@ -3070,6 +3236,7 @@ "here 3"); return 0; } + dim_curr_switch-> int_wire[2].used = 1; dim_curr_switch-> @@ -3147,7 +3314,7 @@ geometry[i2], i2, count); return 0; } - } else if(geometry[i2] == 1) { + } else if((geometry[i2] == 1) && (conn_type == SELECT_TORUS)) { /* FIX ME: This is put here because we got into a state where the Y dim was not being processed correctly. This will set up the @@ -3643,26 +3810,40 @@ ba_node_t* target) { #ifdef HAVE_BG_FILES - my_bluegene_t *bg = NULL; +#ifdef HAVE_BGL + +#define UNDER_POS 7 +#define NODE_LEN 5 +#define VAL_NAME_LEN 12 + +#else + +#define UNDER_POS 9 +#define NODE_LEN 7 +#define VAL_NAME_LEN 16 + +#endif int rc; int i; rm_wire_t *my_wire = NULL; rm_port_t *my_port = NULL; char *wire_id = NULL; - char from_node[5]; - char to_node[5]; int from_port, to_port; int wire_num; int *coord; + char from_node[NODE_LEN]; + char to_node[NODE_LEN]; if (!have_db2) { error("Can't access DB2 library, run from service node"); return -1; } - if ((rc = bridge_get_bg(&bg)) != STATUS_OK) { - error("bridge_get_BG(): %d", rc); - return -1; + if (!bg) { + if((rc = bridge_get_bg(&bg)) != STATUS_OK) { + error("bridge_get_BG(): %d", rc); + return -1; + } } if (bg == NULL) @@ -3700,7 +3881,7 @@ continue; } - if(wire_id[7] != '_') + if(wire_id[UNDER_POS] != '_') continue; switch(wire_id[0]) { case 'X': @@ -3713,17 +3894,17 @@ dim = Z; break; } - if(strlen(wire_id)<12) { + if(strlen(wire_id) < VAL_NAME_LEN) { error("Wire_id isn't correct %s",wire_id); continue; } - strncpy(from_node, wire_id+2, 4); - strncpy(to_node, wire_id+8, 4); + memset(&from_node, 0, sizeof(from_node)); + memset(&to_node, 0, sizeof(to_node)); + strncpy(from_node, wire_id+2, NODE_LEN-1); + strncpy(to_node, wire_id+UNDER_POS+1, NODE_LEN-1); free(wire_id); - - from_node[4] = '\0'; - to_node[4] = '\0'; + if ((rc = bridge_get_data(my_wire, RM_WireFromPort, &my_port)) != STATUS_OK) { error("bridge_get_data(RM_FirstWire): %d", rc); @@ -3750,20 +3931,9 @@ error("1 find_bp_loc: bpid %s not known", from_node); continue; } - - if(coord[X]>=DIM_SIZE[X] - || coord[Y]>=DIM_SIZE[Y] - || coord[Z]>=DIM_SIZE[Z]) { - error("got coord %c%c%c greater than system dims " - "%c%c%c", - alpha_num[coord[X]], - alpha_num[coord[Y]], - alpha_num[coord[Z]], - alpha_num[DIM_SIZE[X]], - alpha_num[DIM_SIZE[Y]], - alpha_num[DIM_SIZE[Z]]); + if(!validate_coord(coord)) continue; - } + source = &ba_system_ptr-> grid[coord[X]][coord[Y]][coord[Z]]; coord = find_bp_loc(to_node); @@ -3771,19 +3941,9 @@ error("2 find_bp_loc: bpid %s not known", to_node); continue; } - if(coord[X]>=DIM_SIZE[X] - || coord[Y]>=DIM_SIZE[Y] - || coord[Z]>=DIM_SIZE[Z]) { - error("got coord %c%c%c greater than system dims " - "%c%c%c", - alpha_num[coord[X]], - alpha_num[coord[Y]], - alpha_num[coord[Z]], - alpha_num[DIM_SIZE[X]], - alpha_num[DIM_SIZE[Y]], - alpha_num[DIM_SIZE[Z]]); + if(!validate_coord(coord)) continue; - } + target = &ba_system_ptr-> grid[coord[X]][coord[Y]][coord[Z]]; _switch_config(source, @@ -3803,9 +3963,6 @@ alpha_num[target->coord[Z]], _port_enum(to_port)); } - if ((rc = bridge_free_bg(bg)) != STATUS_OK) - error("bridge_free_BG(): %s", rc); - #else _switch_config(source, source, dim, 0, 0); @@ -4100,10 +4257,16 @@ /* we don't need to go any further */ if(x_size == 1) { - curr_switch->int_wire[source_port].used = 1; - curr_switch->int_wire[source_port].port_tar = target_port; - curr_switch->int_wire[target_port].used = 1; - curr_switch->int_wire[target_port].port_tar = source_port; + /* Only set this if Torus since mesh doesn't have any + * connections in this path */ + if(conn_type == SELECT_TORUS) { + curr_switch->int_wire[source_port].used = 1; + curr_switch->int_wire[source_port].port_tar = + target_port; + curr_switch->int_wire[target_port].used = 1; + curr_switch->int_wire[target_port].port_tar = + source_port; + } return 1; } @@ -4204,6 +4367,11 @@ } else if(found == x_size) { debug2("Algo(%d) finishing the torus!", algo); + if(deny_pass && (*deny_pass & PASS_DENY_X)) { + info("we don't allow passthroughs 1"); + return 0; + } + if(best_path) list_flush(best_path); else @@ -4358,7 +4526,13 @@ debug2("Algo(%d) yes found next free %d", algo, best_count); node_tar = _set_best_path(); - + + if(deny_pass && (*deny_pass & PASS_DENY_X) + && (*deny_pass & PASS_FOUND_X)) { + debug("We don't allow X passthoughs."); + return 0; + } + next_node = &ba_system_ptr->grid[node_tar[X]] #ifdef HAVE_3D [node_tar[Y]] @@ -4776,9 +4950,9 @@ itr = list_iterator_create(best_path); while((path_switch = (ba_path_switch_t*) list_next(itr))) { - if(passthrough && path_switch->in > 1 && path_switch->out > 1) { - *passthrough = true; - debug2("got a passthrough"); + if(deny_pass && path_switch->in > 1 && path_switch->out > 1) { + *deny_pass |= PASS_FOUND_X; + debug2("got a passthrough in X"); } #ifdef HAVE_3D debug3("mapping %c%c%c %d->%d", @@ -4837,7 +5011,8 @@ return 1; } -static void _destroy_geo(void *object) { +static void _destroy_geo(void *object) +{ int *geo_ptr = (int *)object; xfree(geo_ptr); } diff -Nru slurm-llnl-1.3.13/src/plugins/select/bluegene/block_allocator/block_allocator.h slurm-llnl-1.3.15/src/plugins/select/bluegene/block_allocator/block_allocator.h --- slurm-llnl-1.3.13/src/plugins/select/bluegene/block_allocator/block_allocator.h 2008-12-29 22:56:17.000000000 +0100 +++ slurm-llnl-1.3.15/src/plugins/select/bluegene/block_allocator/block_allocator.h 2009-02-11 18:42:28.000000000 +0100 @@ -52,6 +52,16 @@ #define BA_SYSTEM_DIMENSIONS 1 #endif +#define PASS_DENY_X 0x0001 +#define PASS_DENY_Y 0x0002 +#define PASS_DENY_Z 0x0004 +#define PASS_DENY_ALL 0x00ff + +#define PASS_FOUND_X 0x0100 +#define PASS_FOUND_Y 0x0200 +#define PASS_FOUND_Z 0x0400 +#define PASS_FOUND_ANY 0xff00 + extern bool _initialized; enum {X, Y, Z}; @@ -92,8 +102,11 @@ int geometry[BA_SYSTEM_DIMENSIONS]; /* size of block in geometry */ char *linuximage; /* LinuxImage for this block */ char *mloaderimage; /* mloaderImage for this block */ - bool passthrough; /* filled in if there are - passthroughs in the block created */ + uint16_t deny_pass; /* PASSTHROUGH_FOUND is set if there are + passthroughs in the block + created you can deny + passthroughs by setting the + appropriate bits*/ int procs; /* Number of Real processors in block */ char *ramdiskimage; /* RamDiskImage for this block */ @@ -249,6 +262,7 @@ } ba_bp_map_t; /* Global */ +extern my_bluegene_t *bg; extern List bp_map_list; /* list used for conversion from XYZ to Rack * midplane */ extern char letters[62]; /* complete list of letters used in smap */ @@ -257,10 +271,14 @@ * each dimension */ extern s_p_options_t bg_conf_file_options[]; /* used to parse the * bluegene.conf file. */ +extern uint16_t ba_deny_pass; /* Translate a state enum to a readable string */ extern char *bg_block_state_string(rm_partition_state_t state); +/* must xfree return of this */ +extern char *ba_passthroughs_string(uint16_t passthrough); + /* Parse a block request from the bluegene.conf file */ extern int parse_blockreq(void **dest, slurm_parser_enum_t type, const char *key, const char *value, @@ -517,4 +535,8 @@ */ extern List get_and_set_block_wiring(char *bg_block_id); +/* make sure a node is in the system return 1 if it is 0 if not */ +extern int validate_coord(int *coord); + + #endif /* _BLOCK_ALLOCATOR_H_ */ diff -Nru slurm-llnl-1.3.13/src/plugins/select/bluegene/plugin/bg_block_info.c slurm-llnl-1.3.15/src/plugins/select/bluegene/plugin/bg_block_info.c --- slurm-llnl-1.3.13/src/plugins/select/bluegene/plugin/bg_block_info.c 2008-11-18 00:20:37.000000000 +0100 +++ slurm-llnl-1.3.15/src/plugins/select/bluegene/plugin/bg_block_info.c 2009-04-09 18:56:23.000000000 +0200 @@ -1,7 +1,7 @@ /*****************************************************************************\ * bg_block_info.c - bluegene block information from the db2 database. * - * $Id: bg_block_info.c 15717 2008-11-17 23:20:37Z da $ + * $Id: bg_block_info.c 17202 2009-04-09 16:56:23Z da $ ***************************************************************************** * Copyright (C) 2004-2006 The Regents of the University of California. * Produced at Lawrence Livermore National Laboratory (cf, DISCLAIMER). @@ -64,6 +64,8 @@ #include "src/common/xstring.h" #include "src/slurmctld/proc_req.h" #include "src/api/job_info.h" +#include "src/slurmctld/trigger_mgr.h" +#include "src/slurmctld/locks.h" #include "bluegene.h" #define _DEBUG 0 @@ -87,7 +89,6 @@ if(bg_record->modifying) return SLURM_SUCCESS; - user_name = xstrdup(bg_slurm_user_name); if(remove_all_users(bg_record->bg_block_id, NULL) @@ -95,13 +96,12 @@ error("Something happened removing " "users from block %s", bg_record->bg_block_id); - } - + } if(bg_record->target_name && bg_record->user_name) { if(!strcmp(bg_record->target_name, user_name)) { if(strcmp(bg_record->target_name, bg_record->user_name) - || (jobid > -1)) { + || (jobid > NO_JOB_RUNNING)) { kill_job_struct_t *freeit = xmalloc(sizeof(freeit)); freeit->jobid = jobid; @@ -113,13 +113,6 @@ bg_record->bg_block_id, bg_record->user_name, jobid); - - if(remove_from_bg_list(bg_job_block_list, - bg_record) - == SLURM_SUCCESS) { - num_unused_cpus += bg_record->bp_count - * bg_record->cpus_per_bp; - } } else { debug("Block %s was in a ready state " "but is being freed. No job running.", @@ -130,7 +123,6 @@ "for block %s.", bg_record->bg_block_id); } - remove_from_bg_list(bg_booted_block_list, bg_record); } else if(bg_record->user_name) { error("Target Name was not set " "not set for block %s.", @@ -144,6 +136,10 @@ bg_record->target_name = xstrdup(bg_record->user_name); } + if(remove_from_bg_list(bg_job_block_list, bg_record) == SLURM_SUCCESS) + num_unused_cpus += bg_record->cpu_cnt; + remove_from_bg_list(bg_booted_block_list, bg_record); + xfree(user_name); return SLURM_SUCCESS; @@ -201,6 +197,8 @@ xfree(block_id); } else rc = READY_JOB_ERROR; +/* info("returning %d for job %u %d %d", */ +/* rc, job_ptr->job_id, READY_JOB_ERROR, READY_JOB_FATAL); */ return rc; } @@ -215,8 +213,8 @@ pack16((uint16_t)bg_record->conn_type, buffer); #ifdef HAVE_BGL pack16((uint16_t)bg_record->node_use, buffer); - pack16((uint16_t)bg_record->quarter, buffer); - pack16((uint16_t)bg_record->nodecard, buffer); + pack16((uint16_t)0, buffer); + pack16((uint16_t)0, buffer); #endif pack32((uint32_t)bg_record->node_cnt, buffer); pack_bit_fmt(bg_record->bitmap, buffer); @@ -244,6 +242,8 @@ time_t now; kill_job_struct_t *freeit = NULL; ListIterator itr = NULL; + slurmctld_lock_t job_write_lock = { + NO_LOCK, WRITE_LOCK, WRITE_LOCK, NO_LOCK }; if(!kill_job_list) kill_job_list = list_create(_destroy_kill_struct); @@ -300,6 +300,50 @@ bg_record->node_use = node_use; updated = 1; } +#else + if((bg_record->node_cnt < bluegene_bp_node_cnt) + || (bluegene_bp_node_cnt == bluegene_nodecard_node_cnt)) { + char *mode = NULL; + uint16_t conn_type = SELECT_SMALL; + if ((rc = bridge_get_data(block_ptr, + RM_PartitionOptions, + &mode)) + != STATUS_OK) { + error("bridge_get_data(RM_PartitionOptions): " + "%s", bg_err_str(rc)); + updated = -1; + goto next_block; + } else if(mode) { + switch(mode[0]) { + case 's': + conn_type = SELECT_HTC_S; + break; + case 'd': + conn_type = SELECT_HTC_D; + break; + case 'v': + conn_type = SELECT_HTC_V; + break; + case 'l': + conn_type = SELECT_HTC_L; + break; + default: + conn_type = SELECT_SMALL; + break; + } + free(mode); + } + + if(bg_record->conn_type != conn_type) { + debug("mode of small Block %s was %u " + "and now is %u", + bg_record->bg_block_id, + bg_record->conn_type, + conn_type); + bg_record->conn_type = conn_type; + updated = 1; + } + } #endif if ((rc = bridge_get_data(block_ptr, RM_PartitionState, &state)) @@ -320,18 +364,56 @@ check to make sure block went through freeing correctly */ - if(bg_record->state != RM_PARTITION_DEALLOCATING + if((bg_record->state != RM_PARTITION_DEALLOCATING + && bg_record->state != RM_PARTITION_ERROR) && state == RM_PARTITION_FREE) skipped_dealloc = 1; + else if((bg_record->state == RM_PARTITION_READY) + && (state == RM_PARTITION_CONFIGURING)) { + /* This means the user did a reboot through + mpirun but we missed the state + change */ + debug("Block %s skipped rebooting, " + "but it really is. " + "Setting target_name back to %s", + bg_record->bg_block_id, + bg_record->user_name); + xfree(bg_record->target_name); + bg_record->target_name = + xstrdup(bg_record->user_name); + } bg_record->state = state; if(bg_record->state == RM_PARTITION_DEALLOCATING - || skipped_dealloc) { + || skipped_dealloc) _block_is_deallocating(bg_record); - } else if(bg_record->state == RM_PARTITION_CONFIGURING) +#ifndef HAVE_BGL + else if(bg_record->state == RM_PARTITION_REBOOTING) { + /* This means the user did a reboot through + mpirun */ + debug("Block %s rebooting. " + "Setting target_name back to %s", + bg_record->bg_block_id, + bg_record->user_name); + xfree(bg_record->target_name); + bg_record->target_name = + xstrdup(bg_record->user_name); + } +#endif + else if(bg_record->state == RM_PARTITION_CONFIGURING) bg_record->boot_state = 1; + else if(bg_record->state == RM_PARTITION_FREE) { + if(remove_from_bg_list(bg_job_block_list, + bg_record) + == SLURM_SUCCESS) { + num_unused_cpus += bg_record->cpu_cnt; + } + remove_from_bg_list(bg_booted_block_list, + bg_record); + } updated = 1; + } /* check the boot state */ @@ -350,7 +432,33 @@ break; case RM_PARTITION_ERROR: - error("block in an error state"); + bg_record->boot_state = 0; + bg_record->boot_count = 0; + if(bg_record->job_running > NO_JOB_RUNNING) { + error("Block %s in an error " + "state while booting. " + "Failing job %u.", + bg_record->bg_block_id, + bg_record->job_running); + freeit = xmalloc( + sizeof(kill_job_struct_t)); + freeit->jobid = bg_record->job_running; + list_push(kill_job_list, freeit); + if(remove_from_bg_list( + bg_job_block_list, + bg_record) + == SLURM_SUCCESS) { + num_unused_cpus += + bg_record->cpu_cnt; + } + } else + error("block %s in an error " + "state while booting.", + bg_record->bg_block_id); + remove_from_bg_list(bg_booted_block_list, + bg_record); + trigger_block_error(); + break; case RM_PARTITION_FREE: if(bg_record->boot_count < RETRY_BOOT_COUNT) { slurm_mutex_unlock(&block_state_mutex); @@ -386,6 +494,16 @@ slurm_mutex_lock(&block_state_mutex); bg_record->boot_state = 0; bg_record->boot_count = 0; + if(remove_from_bg_list( + bg_job_block_list, + bg_record) + == SLURM_SUCCESS) { + num_unused_cpus += + bg_record->cpu_cnt; + } + remove_from_bg_list( + bg_booted_block_list, + bg_record); } break; case RM_PARTITION_READY: @@ -431,8 +549,15 @@ /* kill all the jobs from unexpectedly freed blocks */ while((freeit = list_pop(kill_job_list))) { - debug2("killing job %d", freeit->jobid); - (void) slurm_fail_job(freeit->jobid); + debug2("Trying to requeue job %d", freeit->jobid); + lock_slurmctld(job_write_lock); + if((rc = job_requeue(0, freeit->jobid, -1))) { + error("couldn't requeue job %u, failing it: %s", + freeit->jobid, + slurm_strerror(rc)); + (void) job_fail(freeit->jobid); + } + unlock_slurmctld(job_write_lock); _destroy_kill_struct(freeit); } @@ -501,6 +626,7 @@ state); bg_record->state = state; + updated = 1; } next_block: if ((rc = bridge_free_block(block_ptr)) diff -Nru slurm-llnl-1.3.13/src/plugins/select/bluegene/plugin/bg_job_place.c slurm-llnl-1.3.15/src/plugins/select/bluegene/plugin/bg_job_place.c --- slurm-llnl-1.3.13/src/plugins/select/bluegene/plugin/bg_job_place.c 2008-11-22 00:38:34.000000000 +0100 +++ slurm-llnl-1.3.15/src/plugins/select/bluegene/plugin/bg_job_place.c 2009-04-09 19:24:11.000000000 +0200 @@ -2,7 +2,7 @@ * bg_job_place.c - blue gene job placement (e.g. base block selection) * functions. * - * $Id: bg_job_place.c 15759 2008-11-21 23:38:34Z da $ + * $Id: bg_job_place.c 17205 2009-04-09 17:24:11Z da $ ***************************************************************************** * Copyright (C) 2004-2007 The Regents of the University of California. * Produced at Lawrence Livermore National Laboratory (cf, DISCLAIMER). @@ -137,13 +137,16 @@ */ static int _bg_record_sort_aval_inc(bg_record_t* rec_a, bg_record_t* rec_b) { - int size_a = rec_a->node_cnt; - int size_b = rec_b->node_cnt; - - if(rec_a->job_ptr && !rec_b->job_ptr) + if((rec_a->job_running == BLOCK_ERROR_STATE) + && (rec_b->job_running != BLOCK_ERROR_STATE)) + return 1; + else if((rec_a->job_running != BLOCK_ERROR_STATE) + && (rec_b->job_running == BLOCK_ERROR_STATE)) return -1; else if(!rec_a->job_ptr && rec_b->job_ptr) return 1; + else if(rec_a->job_ptr && !rec_b->job_ptr) + return -1; else if(rec_a->job_ptr && rec_b->job_ptr) { if(rec_a->job_ptr->start_time > rec_b->job_ptr->start_time) return 1; @@ -151,29 +154,7 @@ return -1; } - if (size_a < size_b) - return -1; - else if (size_a > size_b) - return 1; - if(rec_a->nodes && rec_b->nodes) { - size_a = strcmp(rec_a->nodes, rec_b->nodes); - if (size_a < 0) - return -1; - else if (size_a > 0) - return 1; - } -#ifdef HAVE_BGL - if (rec_a->quarter < rec_b->quarter) - return -1; - else if (rec_a->quarter > rec_b->quarter) - return 1; - - if(rec_a->nodecard < rec_b->nodecard) - return -1; - else if(rec_a->nodecard > rec_b->nodecard) - return 1; -#endif - return 0; + return bg_record_cmpf_inc(rec_a, rec_b); } /* @@ -184,10 +165,16 @@ */ static int _bg_record_sort_aval_dec(bg_record_t* rec_a, bg_record_t* rec_b) { - if(rec_a->job_ptr && !rec_b->job_ptr) + if((rec_a->job_running == BLOCK_ERROR_STATE) + && (rec_b->job_running != BLOCK_ERROR_STATE)) + return -1; + else if((rec_a->job_running != BLOCK_ERROR_STATE) + && (rec_b->job_running == BLOCK_ERROR_STATE)) return 1; else if(!rec_a->job_ptr && rec_b->job_ptr) return -1; + else if(rec_a->job_ptr && !rec_b->job_ptr) + return 1; else if(rec_a->job_ptr && rec_b->job_ptr) { if(rec_a->job_ptr->start_time > rec_b->job_ptr->start_time) return -1; @@ -363,7 +350,6 @@ { bg_record_t *bg_record = NULL; ListIterator itr = NULL; - uint32_t proc_cnt = 0; char tmp_char[256]; debug("number of blocks to check: %d state %d", @@ -377,8 +363,10 @@ */ debug3("%s job_running = %d", bg_record->bg_block_id, bg_record->job_running); - /*block is messed up some how (BLOCK_ERROR_STATE) ignore it*/ - if(bg_record->job_running == BLOCK_ERROR_STATE) { + /*block is messed up some how (BLOCK_ERROR_STATE) + * ignore it or if state == RM_PARTITION_ERROR */ + if((bg_record->job_running == BLOCK_ERROR_STATE) + || (bg_record->state == RM_PARTITION_ERROR)) { debug("block %s is in an error state (can't use)", bg_record->bg_block_id); continue; @@ -395,15 +383,15 @@ } /* Check processor count */ - proc_cnt = bg_record->bp_count * bg_record->cpus_per_bp; debug3("asking for %u-%u looking at %d", - request->procs, max_procs, proc_cnt); - if ((proc_cnt < request->procs) - || ((max_procs != NO_VAL) && (proc_cnt > max_procs))) { + request->procs, max_procs, bg_record->cpu_cnt); + if ((bg_record->cpu_cnt < request->procs) + || ((max_procs != NO_VAL) + && (bg_record->cpu_cnt > max_procs))) { /* We use the proccessor count per block here mostly to see if we can run on a smaller block. */ - convert_num_unit((float)proc_cnt, tmp_char, + convert_num_unit((float)bg_record->cpu_cnt, tmp_char, sizeof(tmp_char), UNIT_NONE); debug("block %s CPU count (%s) not suitable", bg_record->bg_block_id, @@ -477,6 +465,21 @@ /***********************************************/ if ((request->conn_type != bg_record->conn_type) && (request->conn_type != SELECT_NAV)) { +#ifndef HAVE_BGL + if(request->conn_type >= SELECT_SMALL) { + /* we only want to reboot blocks if + they have to be so skip booted + blocks if in small state + */ + if(check_image + && (bg_record->state + == RM_PARTITION_READY)) { + *allow = 1; + continue; + } + goto good_conn_type; + } +#endif debug("bg block %s conn-type not usable asking for %s " "bg_record is %s", bg_record->bg_block_id, @@ -484,7 +487,9 @@ convert_conn_type(bg_record->conn_type)); continue; } - +#ifndef HAVE_BGL + good_conn_type: +#endif /*****************************************/ /* match up geometry as "best" possible */ /*****************************************/ @@ -611,9 +616,12 @@ } } - if(found_record->job_running != NO_JOB_RUNNING) { - if(found_record->job_running - == BLOCK_ERROR_STATE) + if((found_record->job_running != NO_JOB_RUNNING) + || (found_record->state == RM_PARTITION_ERROR)) { + if((found_record->job_running + == BLOCK_ERROR_STATE) + || (found_record->state + == RM_PARTITION_ERROR)) error("can't use %s, " "overlapping block %s " "is in an error state.", @@ -651,12 +659,26 @@ } destroy_bg_record(bg_record); if(!found_record) { - debug2("This record wasn't " - "found in the bg_list, " - "no big deal, it " - "probably wasn't added"); + /* There may be a bug + here where on a real + system we don't go + destroy this block + in the real system. + If that is the case we + need to add the + bg_record to the + free_block_list + instead of destroying + it like above. + */ + debug("This record wasn't " + "found in the bg_list, " + "no big deal, it " + "probably wasn't added"); //rc = SLURM_ERROR; } else { + debug("removing the block " + "from the system"); List temp_list = list_create(NULL); list_push(temp_list, @@ -728,7 +750,8 @@ */ debug("trying with %d", create_try); if((new_blocks = create_dynamic_block(block_list, - request, temp_list))) { + request, temp_list, + true))) { bg_record_t *bg_record = NULL; while((bg_record = list_pop(new_blocks))) { if(block_exist_in_list(block_list, bg_record)) @@ -942,12 +965,15 @@ *found_bg_record = NULL; allow = 0; + memset(&request, 0, sizeof(ba_request_t)); + for(i=0; ibg_block_id); - bg_record->job_running = - BLOCK_ERROR_STATE; - bg_record->state = RM_PARTITION_ERROR; - trigger_block_error(); + put_block_in_error_state( + bg_record, BLOCK_ERROR_STATE); continue; } } @@ -1087,20 +1111,46 @@ slurm_mutex_unlock(&block_state_mutex); list_sort(job_list, (ListCmpF)_bg_record_sort_aval_inc); while(1) { + bool track_down_nodes = true; /* this gets altered in * create_dynamic_block so we reset it */ for(i=0; ijob_running, - bg_record->bg_block_id, - bg_record->job_ptr->start_time, - bg_record->job_ptr->end_time); + if(bg_record) { + if(bg_record->job_ptr) + debug2("taking off %d(%s) " + "started at %d " + "ends at %d", + bg_record->job_running, + bg_record->bg_block_id, + bg_record->job_ptr-> + start_time, + bg_record->job_ptr-> + end_time); + else if(bg_record->job_running + == BLOCK_ERROR_STATE) + debug2("taking off (%s) " + "which is in an error " + "state", + bg_record->job_running, + bg_record->bg_block_id, + bg_record->job_ptr-> + start_time, + bg_record->job_ptr-> + end_time); + } else + /* This means we didn't have + any jobs to take off + anymore so we are making + sure we can look at every + node on the system. + */ + track_down_nodes = false; if(!(new_blocks = create_dynamic_block( - block_list, &request, job_list))) { + block_list, &request, job_list, + track_down_nodes))) { destroy_bg_record(bg_record); if(errno == ESLURM_INTERCONNECT_FAILURE || !list_count(job_list)) { @@ -1239,7 +1289,7 @@ #ifdef HAVE_BG bg_record_t* bg_record = NULL; char buf[100]; - uint16_t tmp16 = (uint16_t)NO_VAL; + uint16_t conn_type = (uint16_t)NO_VAL; List block_list = NULL; int blocks_added = 0; time_t starttime = time(NULL); @@ -1257,6 +1307,33 @@ job_block_test_list = bg_job_block_list; + select_g_get_jobinfo(job_ptr->select_jobinfo, + SELECT_DATA_CONN_TYPE, &conn_type); + if(conn_type == SELECT_NAV) { + uint32_t max_procs = (uint32_t)NO_VAL; + if(bluegene_bp_node_cnt == bluegene_nodecard_node_cnt) + conn_type = SELECT_SMALL; + else if(min_nodes > 1) { + conn_type = SELECT_TORUS; + /* make sure the max procs are set to NO_VAL */ + select_g_set_jobinfo(job_ptr->select_jobinfo, + SELECT_DATA_MAX_PROCS, + &max_procs); + + } else { + select_g_get_jobinfo(job_ptr->select_jobinfo, + SELECT_DATA_MAX_PROCS, + &max_procs); + if((max_procs > procs_per_node) + || (max_procs == NO_VAL)) + conn_type = SELECT_TORUS; + else + conn_type = SELECT_SMALL; + } + select_g_set_jobinfo(job_ptr->select_jobinfo, + SELECT_DATA_CONN_TYPE, + &conn_type); + } select_g_sprint_jobinfo(job_ptr->select_jobinfo, buf, sizeof(buf), SELECT_PRINT_MIXED); debug("bluegene:submit_job: %s nodes=%u-%u-%u", @@ -1311,7 +1388,8 @@ else starttime = bg_record->job_ptr->end_time; - } + } else if(bg_record->job_running == BLOCK_ERROR_STATE) + starttime = INFINITE; job_ptr->start_time = starttime; @@ -1325,22 +1403,14 @@ if(!bg_record->bg_block_id) { uint16_t geo[BA_SYSTEM_DIMENSIONS]; - debug2("%d can start job at " - "%u on %s on unmade block", - test_only, starttime, + debug2("%d can start unassigned job %u at " + "%u on %s", + test_only, job_ptr->job_id, starttime, bg_record->nodes); select_g_set_jobinfo(job_ptr->select_jobinfo, SELECT_DATA_BLOCK_ID, "unassigned"); - /* if(job_ptr->num_procs < bluegene_bp_node_cnt */ -/* && job_ptr->num_procs > 0) { */ -/* i = procs_per_node/job_ptr->num_procs; */ -/* debug2("divide by %d", i); */ -/* } else */ -/* i = 1; */ -/* min_nodes *= bluegene_bp_node_cnt/i; */ - /* this seems to do the same thing as - * above */ + min_nodes = bg_record->node_cnt; select_g_set_jobinfo(job_ptr->select_jobinfo, SELECT_DATA_NODE_CNT, @@ -1360,10 +1430,11 @@ error("Small block used in " "non-shared partition"); - debug2("%d can start job at %u on %s", - test_only, starttime, + debug2("%d can start job %u at %u on %s(%s)", + test_only, job_ptr->job_id, starttime, + bg_record->bg_block_id, bg_record->nodes); - + select_g_set_jobinfo(job_ptr->select_jobinfo, SELECT_DATA_BLOCK_ID, bg_record->bg_block_id); @@ -1374,10 +1445,10 @@ SELECT_DATA_GEOMETRY, &bg_record->geo); - tmp16 = bg_record->conn_type; - select_g_set_jobinfo(job_ptr->select_jobinfo, - SELECT_DATA_CONN_TYPE, - &tmp16); + /* tmp16 = bg_record->conn_type; */ +/* select_g_set_jobinfo(job_ptr->select_jobinfo, */ +/* SELECT_DATA_CONN_TYPE, */ +/* &tmp16); */ } } else { error("we got a success, but no block back"); @@ -1424,12 +1495,41 @@ itr = list_iterator_create(req_list); while((will_run = list_next(itr))) { + uint16_t conn_type = (uint16_t)NO_VAL; + if(!will_run->job_ptr) { error("test_job_list: you need to give me a job_ptr"); rc = SLURM_ERROR; break; } + select_g_get_jobinfo(will_run->job_ptr->select_jobinfo, + SELECT_DATA_CONN_TYPE, &conn_type); + if(conn_type == SELECT_NAV) { + uint32_t max_procs = (uint32_t)NO_VAL; + if(will_run->min_nodes > 1) { + conn_type = SELECT_TORUS; + /* make sure the max procs are set to NO_VAL */ + select_g_set_jobinfo( + will_run->job_ptr->select_jobinfo, + SELECT_DATA_MAX_PROCS, + &max_procs); + + } else { + select_g_get_jobinfo( + will_run->job_ptr->select_jobinfo, + SELECT_DATA_MAX_PROCS, + &max_procs); + if((max_procs > procs_per_node) + || (max_procs == NO_VAL)) + conn_type = SELECT_TORUS; + else + conn_type = SELECT_SMALL; + } + select_g_set_jobinfo(will_run->job_ptr->select_jobinfo, + SELECT_DATA_CONN_TYPE, + &conn_type); + } select_g_sprint_jobinfo(will_run->job_ptr->select_jobinfo, buf, sizeof(buf), SELECT_PRINT_MIXED); diff -Nru slurm-llnl-1.3.13/src/plugins/select/bluegene/plugin/bg_job_run.c slurm-llnl-1.3.15/src/plugins/select/bluegene/plugin/bg_job_run.c --- slurm-llnl-1.3.13/src/plugins/select/bluegene/plugin/bg_job_run.c 2009-01-06 19:20:48.000000000 +0100 +++ slurm-llnl-1.3.15/src/plugins/select/bluegene/plugin/bg_job_run.c 2009-04-09 18:56:23.000000000 +0200 @@ -2,7 +2,7 @@ * bg_job_run.c - blue gene job execution (e.g. initiation and termination) * functions. * - * $Id: bg_job_run.c 16146 2009-01-06 18:20:48Z da $ + * $Id: bg_job_run.c 17202 2009-04-09 16:56:23Z da $ ***************************************************************************** * Copyright (C) 2004-2006 The Regents of the University of California. * Produced at Lawrence Livermore National Laboratory (cf, DISCLAIMER). @@ -76,6 +76,10 @@ struct job_record *job_ptr; /* pointer to job running on * block or NULL if no job */ uint16_t reboot; /* reboot block before starting job */ +#ifndef HAVE_BGL + uint16_t conn_type; /* needed to boot small blocks into + HTC mode or not */ +#endif pm_partition_id_t bg_block_id; char *blrtsimage; /* BlrtsImage for this block */ char *linuximage; /* LinuxImage for this block */ @@ -87,6 +91,7 @@ static pthread_mutex_t agent_cnt_mutex = PTHREAD_MUTEX_INITIALIZER; static pthread_mutex_t job_start_mutex = PTHREAD_MUTEX_INITIALIZER; +static pthread_cond_t agent_cond = PTHREAD_COND_INITIALIZER; static int agent_cnt = 0; #ifdef HAVE_BG_FILES @@ -109,16 +114,17 @@ /* Kill a job and remove its record from MMCS */ static int _remove_job(db_job_id_t job_id) { - int i, rc; + int rc; + int count = 0; rm_job_t *job_rec = NULL; rm_job_state_t job_state; debug("removing job %d from MMCS", job_id); - for (i=0; i 0) + while(1) { + if (count) sleep(POLL_INTERVAL); + count++; - /* Find the job */ if ((rc = bridge_get_job(job_id, &job_rec)) != STATUS_OK) { @@ -153,36 +159,45 @@ /* check the state and process accordingly */ if(job_state == RM_JOB_TERMINATED) return STATUS_OK; - else if(job_state == RM_JOB_DYING) + else if(job_state == RM_JOB_DYING) { + /* start sending sigkills for the last 5 tries */ + if(count > MAX_POLL_RETRIES) + error("Job %d isn't dying, trying for " + "%d seconds", count*POLL_INTERVAL); continue; - else if(job_state == RM_JOB_ERROR) { + } else if(job_state == RM_JOB_ERROR) { error("job %d is in a error state.", job_id); //free_bg_block(); return STATUS_OK; } - (void) bridge_signal_job(job_id, SIGKILL); - rc = bridge_cancel_job(job_id); - /* it doesn't appear that this does anything. */ - // rc = bridge_remove_job(job_id); + /* we have been told the next 2 lines do the same + * thing, but I don't believe it to be true. In most + * cases when you do a signal of SIGTERM the mpirun + * process gets killed with a SIGTERM. In the case of + * bridge_cancel_job it always gets killed with a + * SIGKILL. From IBM's point of view that is a bad + * deally, so we are going to use signal ;). + */ + +// rc = bridge_cancel_job(job_id); + rc = bridge_signal_job(job_id, SIGTERM); if (rc != STATUS_OK) { if (rc == JOB_NOT_FOUND) { debug("job %d removed from MMCS", job_id); return STATUS_OK; - } + } if(rc == INCOMPATIBLE_STATE) debug("job %d is in an INCOMPATIBLE_STATE", job_id); else - error("bridge_cancel_job(%d): %s", job_id, + error("bridge_cancel_job(%d): %s", job_id, bg_err_str(rc)); } } - /* try once more... */ - /* it doesn't appear that this does anything. */ - // (void) bridge_remove_job(job_id); + error("Failed to remove job %d from MMCS", job_id); return INTERNAL_ERROR; } @@ -197,20 +212,17 @@ bg_record->job_running = NO_JOB_RUNNING; bg_record->job_ptr = NULL; } - /* remove user from list */ - + /* remove user from list */ if(bg_record->target_name) { - if(strcmp(bg_record->target_name, - bg_slurm_user_name)) { + if(strcmp(bg_record->target_name, bg_slurm_user_name)) { xfree(bg_record->target_name); bg_record->target_name = xstrdup(bg_slurm_user_name); } update_block_user(bg_record, 1); } else { - bg_record->target_name = - xstrdup(bg_slurm_user_name); + bg_record->target_name = xstrdup(bg_slurm_user_name); } @@ -220,8 +232,7 @@ last_bg_update = time(NULL); if(remove_from_bg_list(bg_job_block_list, bg_record) == SLURM_SUCCESS) { - num_unused_cpus += - bg_record->bp_count*bg_record->cpus_per_bp; + num_unused_cpus += bg_record->cpu_cnt; } } else { error("No block given to reset"); @@ -261,11 +272,11 @@ bg_record->job_running = bg_update_ptr->job_ptr->job_id; bg_record->job_ptr = bg_update_ptr->job_ptr; - if(!block_exist_in_list(bg_job_block_list, bg_record)) { + if(!block_ptr_exist_in_list(bg_job_block_list, bg_record)) { list_push(bg_job_block_list, bg_record); - num_unused_cpus -= bg_record->bp_count*bg_record->cpus_per_bp; + num_unused_cpus -= bg_record->cpu_cnt; } - if(!block_exist_in_list(bg_booted_block_list, bg_record)) + if(!block_ptr_exist_in_list(bg_booted_block_list, bg_record)) list_push(bg_booted_block_list, bg_record); slurm_mutex_unlock(&block_state_mutex); @@ -314,8 +325,7 @@ slurm_mutex_lock(&job_start_mutex); - bg_record = - find_bg_record_in_list(bg_list, bg_update_ptr->bg_block_id); + bg_record = find_bg_record_in_list(bg_list, bg_update_ptr->bg_block_id); if(!bg_record) { error("block %s not found in bg_list", @@ -339,6 +349,7 @@ } slurm_mutex_lock(&block_state_mutex); if(bg_record->job_running <= NO_JOB_RUNNING) { + // _reset_block(bg_record); should already happened slurm_mutex_unlock(&block_state_mutex); slurm_mutex_unlock(&job_start_mutex); debug("job %u finished during the queueing job " @@ -350,6 +361,9 @@ slurm_mutex_unlock(&block_state_mutex); debug("Block is in Deallocating state, waiting for free."); bg_free_block(bg_record); + /* no reason to reboot here since we are already + deallocating */ + bg_update_ptr->reboot = 0; } else slurm_mutex_unlock(&block_state_mutex); @@ -357,8 +371,7 @@ delete_list = list_create(NULL); slurm_mutex_lock(&block_state_mutex); itr = list_iterator_create(bg_list); - while ((found_record = (bg_record_t*) - list_next(itr)) != NULL) { + while ((found_record = list_next(itr))) { if ((!found_record) || (bg_record == found_record)) continue; @@ -438,6 +451,7 @@ slurm_mutex_lock(&block_state_mutex); if(bg_record->job_running <= NO_JOB_RUNNING) { + // _reset_block(bg_record); should already happened slurm_mutex_unlock(&block_state_mutex); slurm_mutex_unlock(&job_start_mutex); debug("job %u already finished before boot", @@ -455,6 +469,13 @@ bg_record->blrtsimage = xstrdup(bg_update_ptr->blrtsimage); rc = 1; } +#else + if((bg_update_ptr->conn_type >= SELECT_SMALL) + && (bg_update_ptr->conn_type != bg_record->conn_type)) { + debug3("changing small block mode from %u to %u", + bg_record->conn_type, bg_update_ptr->conn_type); + rc = 1; + } #endif if(bg_update_ptr->linuximage && strcasecmp(bg_update_ptr->linuximage, bg_record->linuximage)) { @@ -538,6 +559,33 @@ error("bridge_modify_block(RM_MODIFY_IoloadImg)", bg_err_str(rc)); + if(bg_update_ptr->conn_type > SELECT_SMALL) { + char *conn_type = NULL; + switch(bg_update_ptr->conn_type) { + case SELECT_HTC_S: + conn_type = "s"; + break; + case SELECT_HTC_D: + conn_type = "d"; + break; + case SELECT_HTC_V: + conn_type = "v"; + break; + case SELECT_HTC_L: + conn_type = "l"; + break; + default: + break; + } + /* the option has to be set before the pool can be + set */ + if ((rc = bridge_modify_block( + bg_record->bg_block_id, + RM_MODIFY_Options, + conn_type)) != STATUS_OK) + error("bridge_set_data(RM_MODIFY_Options)", + bg_err_str(rc)); + } #endif if ((rc = bridge_modify_block(bg_record->bg_block_id, RM_MODIFY_MloaderImg, @@ -550,22 +598,22 @@ slurm_mutex_lock(&block_state_mutex); bg_record->modifying = 0; slurm_mutex_unlock(&block_state_mutex); - } else if(bg_update_ptr->reboot) -#ifdef HAVE_BGL + } else if(bg_update_ptr->reboot) { + slurm_mutex_lock(&block_state_mutex); + bg_record->modifying = 1; + slurm_mutex_unlock(&block_state_mutex); + bg_free_block(bg_record); -#else - bg_reboot_block(bg_record); -#endif + + slurm_mutex_lock(&block_state_mutex); + bg_record->modifying = 0; + slurm_mutex_unlock(&block_state_mutex); + } if(bg_record->state == RM_PARTITION_FREE) { if((rc = boot_block(bg_record)) != SLURM_SUCCESS) { slurm_mutex_lock(&block_state_mutex); _reset_block(bg_record); - if (remove_from_bg_list(bg_job_block_list, bg_record) - == SLURM_SUCCESS) { - num_unused_cpus += bg_record->bp_count - *bg_record->cpus_per_bp; - } slurm_mutex_unlock(&block_state_mutex); sleep(2); /* wait for the slurmd to begin @@ -626,13 +674,11 @@ slurm_mutex_lock(&block_state_mutex); if (remove_from_bg_list(bg_job_block_list, bg_record) == SLURM_SUCCESS) { - num_unused_cpus += bg_record->bp_count - *bg_record->cpus_per_bp; + num_unused_cpus += bg_record->cpu_cnt; } slurm_mutex_unlock(&block_state_mutex); } slurm_mutex_unlock(&job_start_mutex); - } /* Perform job termination work */ @@ -732,8 +778,7 @@ #endif /* remove the block's users */ - bg_record = - find_bg_record_in_list(bg_list, bg_update_ptr->bg_block_id); + bg_record = find_bg_record_in_list(bg_list, bg_update_ptr->bg_block_id); if(bg_record) { debug("got the record %s user is %s", bg_record->bg_block_id, @@ -813,6 +858,8 @@ if (agent_cnt == 0) { list_destroy(bg_update_list); bg_update_list = NULL; + pthread_cond_signal(&agent_cond); + } slurm_mutex_unlock(&agent_cnt_mutex); return NULL; @@ -998,7 +1045,12 @@ SELECT_DATA_BLRTS_IMAGE, bg_update_ptr->blrtsimage); } +#else + select_g_get_jobinfo(job_ptr->select_jobinfo, + SELECT_DATA_CONN_TYPE, + &(bg_update_ptr->conn_type)); #endif + select_g_get_jobinfo(job_ptr->select_jobinfo, SELECT_DATA_LINUX_IMAGE, &(bg_update_ptr->linuximage)); @@ -1030,16 +1082,14 @@ find_bg_record_in_list(bg_list, bg_update_ptr->bg_block_id); if (bg_record) { slurm_mutex_lock(&block_state_mutex); - job_ptr->num_procs = (bg_record->cpus_per_bp * - bg_record->bp_count); + job_ptr->num_procs = bg_record->cpu_cnt; bg_record->job_running = bg_update_ptr->job_ptr->job_id; bg_record->job_ptr = bg_update_ptr->job_ptr; - if(!block_exist_in_list(bg_job_block_list, bg_record)) { + if(!block_ptr_exist_in_list(bg_job_block_list, bg_record)) { list_push(bg_job_block_list, bg_record); - num_unused_cpus -= - bg_record->bp_count*bg_record->cpus_per_bp; + num_unused_cpus -= bg_record->cpu_cnt; } - if(!block_exist_in_list(bg_booted_block_list, bg_record)) + if(!block_ptr_exist_in_list(bg_booted_block_list, bg_record)) list_push(bg_booted_block_list, bg_record); slurm_mutex_unlock(&block_state_mutex); } else { @@ -1210,7 +1260,7 @@ if ((rc = bridge_set_block_owner(bg_record->bg_block_id, bg_slurm_user_name)) != STATUS_OK) { - error("bridge_set_part_owner(%s,%s): %s", + error("bridge_set_block_owner(%s,%s): %s", bg_record->bg_block_id, bg_slurm_user_name, bg_err_str(rc)); @@ -1241,7 +1291,7 @@ } slurm_mutex_lock(&block_state_mutex); - if(!block_exist_in_list(bg_booted_block_list, bg_record)) + if(!block_ptr_exist_in_list(bg_booted_block_list, bg_record)) list_push(bg_booted_block_list, bg_record); slurm_mutex_unlock(&block_state_mutex); @@ -1267,7 +1317,7 @@ slurm_mutex_unlock(&block_state_mutex); #else slurm_mutex_lock(&block_state_mutex); - if(!block_exist_in_list(bg_booted_block_list, bg_record)) + if(!block_ptr_exist_in_list(bg_booted_block_list, bg_record)) list_push(bg_booted_block_list, bg_record); bg_record->state = RM_PARTITION_READY; last_bg_update = time(NULL); @@ -1277,3 +1327,9 @@ return SLURM_SUCCESS; } + +extern void waitfor_block_agents() +{ + if(agent_cnt) + pthread_cond_wait(&agent_cond, &agent_cnt_mutex); +} diff -Nru slurm-llnl-1.3.13/src/plugins/select/bluegene/plugin/bg_job_run.h slurm-llnl-1.3.15/src/plugins/select/bluegene/plugin/bg_job_run.h --- slurm-llnl-1.3.13/src/plugins/select/bluegene/plugin/bg_job_run.h 2008-04-25 21:45:26.000000000 +0200 +++ slurm-llnl-1.3.15/src/plugins/select/bluegene/plugin/bg_job_run.h 2009-03-25 18:20:43.000000000 +0100 @@ -86,4 +86,6 @@ */ extern int term_jobs_on_block(pm_partition_id_t bg_block_id); +extern void waitfor_block_agents(); + #endif /* _BG_JOB_RUN_H_ */ diff -Nru slurm-llnl-1.3.13/src/plugins/select/bluegene/plugin/bg_record_functions.c slurm-llnl-1.3.15/src/plugins/select/bluegene/plugin/bg_record_functions.c --- slurm-llnl-1.3.13/src/plugins/select/bluegene/plugin/bg_record_functions.c 2008-11-21 00:07:34.000000000 +0100 +++ slurm-llnl-1.3.15/src/plugins/select/bluegene/plugin/bg_record_functions.c 2009-04-09 18:56:23.000000000 +0200 @@ -41,6 +41,7 @@ #include "src/common/uid.h" #include "src/slurmctld/trigger_mgr.h" +#include "src/slurmctld/locks.h" /* some local functions */ #ifdef HAVE_BG @@ -62,7 +63,7 @@ info("\tsize: %d BPs %u Nodes %d cpus", bg_record->bp_count, bg_record->node_cnt, - bg_record->cpus_per_bp * bg_record->bp_count); + bg_record->cpu_cnt); info("\tgeo: %ux%ux%u", bg_record->geo[X], bg_record->geo[Y], bg_record->geo[Z]); info("\tconn_type: %s", convert_conn_type(bg_record->conn_type)); @@ -111,25 +112,26 @@ } } +/* see if a record already of like bitmaps exists in a list */ extern int block_exist_in_list(List my_list, bg_record_t *bg_record) { ListIterator itr = list_iterator_create(my_list); bg_record_t *found_record = NULL; int rc = 0; - while ((found_record = (bg_record_t *) list_next(itr)) != NULL) { + while ((found_record = list_next(itr))) { /* check for full node bitmap compare */ if(bit_equal(bg_record->bitmap, found_record->bitmap) && bit_equal(bg_record->ionode_bitmap, found_record->ionode_bitmap)) { if(bg_record->ionodes) - debug3("This block %s[%s] " + debug("This block %s[%s] " "is already in the list %s", bg_record->nodes, bg_record->ionodes, found_record->bg_block_id); else - debug3("This block %s " + debug("This block %s " "is already in the list %s", bg_record->nodes, found_record->bg_block_id); @@ -142,6 +144,23 @@ return rc; } +/* see if the exact record already exists in a list */ +extern int block_ptr_exist_in_list(List my_list, bg_record_t *bg_record) +{ + ListIterator itr = list_iterator_create(my_list); + bg_record_t *found_record = NULL; + int rc = 0; + + while ((found_record = list_next(itr))) { + if(bg_record == found_record) { + rc = 1; + break; + } + } + list_iterator_destroy(itr); + return rc; +} + extern void process_nodes(bg_record_t *bg_record, bool startup) { #ifdef HAVE_BG @@ -162,7 +181,7 @@ } memset(&best_start, 0, sizeof(best_start)); bg_record->bp_count = 0; - if((bg_record->conn_type == SELECT_SMALL) && (!startup)) + if((bg_record->conn_type >= SELECT_SMALL) && (!startup)) error("We shouldn't be here there could be some " "badness if we use this logic %s", bg_record->nodes); @@ -296,12 +315,19 @@ alpha_num[bg_record->geo[Y]], alpha_num[bg_record->geo[Z]], bg_record->bp_count); - - if ((bg_record->geo[X] == DIM_SIZE[X]) - && (bg_record->geo[Y] == DIM_SIZE[Y]) - && (bg_record->geo[Z] == DIM_SIZE[Z])) { - bg_record->full_block = 1; - } + /* This check is for sub midplane systems to figure out what + the largest block can be. + */ + if((DIM_SIZE[X] > 1) || (DIM_SIZE[Y] > 1) || (DIM_SIZE[Z] > 1)) { + /* means we have more than 1 base partition */ + if ((bg_record->geo[X] == DIM_SIZE[X]) + && (bg_record->geo[Y] == DIM_SIZE[Y]) + && (bg_record->geo[Z] == DIM_SIZE[Z])) { + bg_record->full_block = 1; + } + } else if(bg_record->node_cnt == bluegene_bp_node_cnt) + bg_record->full_block = 1; + /* #ifndef HAVE_BG_FILES */ /* max_dim[X] = MAX(max_dim[X], end[X]); */ @@ -420,28 +446,26 @@ } sec_record->job_running = fir_record->job_running; sec_record->job_ptr = fir_record->job_ptr; - sec_record->cpus_per_bp = fir_record->cpus_per_bp; + sec_record->cpu_cnt = fir_record->cpu_cnt; sec_record->node_cnt = fir_record->node_cnt; -#ifdef HAVE_BGL - sec_record->quarter = fir_record->quarter; - sec_record->nodecard = fir_record->nodecard; -#endif } /* * Comparator used for sorting blocks smallest to largest * - * returns: -1: rec_a >rec_b 0: rec_a == rec_b 1: rec_a < rec_b + * returns: -1: rec_a > rec_b 0: rec_a == rec_b 1: rec_a < rec_b * */ extern int bg_record_cmpf_inc(bg_record_t* rec_a, bg_record_t* rec_b) { int size_a = rec_a->node_cnt; int size_b = rec_b->node_cnt; + if (size_a < size_b) return -1; else if (size_a > size_b) return 1; + if(rec_a->nodes && rec_b->nodes) { size_a = strcmp(rec_a->nodes, rec_b->nodes); if (size_a < 0) @@ -449,22 +473,15 @@ else if (size_a > 0) return 1; } -#ifdef HAVE_BGL - if (rec_a->quarter < rec_b->quarter) - return -1; - else if (rec_a->quarter > rec_b->quarter) - return 1; - if(rec_a->nodecard < rec_b->nodecard) - return -1; - else if(rec_a->nodecard > rec_b->nodecard) - return 1; -#else + if(!rec_a->ionode_bitmap || !rec_b->ionode_bitmap) + return 0; + if(bit_ffs(rec_a->ionode_bitmap) < bit_ffs(rec_b->ionode_bitmap)) return -1; else return 1; -#endif + return 0; } @@ -473,29 +490,25 @@ ListIterator itr; bg_record_t *bg_record = NULL; + xassert(my_list); + if(!bg_block_id) return NULL; - if(my_list) { - slurm_mutex_lock(&block_state_mutex); - itr = list_iterator_create(my_list); - while ((bg_record = (bg_record_t *) list_next(itr)) != NULL) { - if(bg_record->bg_block_id) - if (!strcmp(bg_record->bg_block_id, - bg_block_id)) - break; - } - list_iterator_destroy(itr); - slurm_mutex_unlock(&block_state_mutex); - if(bg_record) - return bg_record; - else - return NULL; - } else { - error("find_bg_record_in_list: no list"); - return NULL; + slurm_mutex_lock(&block_state_mutex); + itr = list_iterator_create(my_list); + while ((bg_record = (bg_record_t *) list_next(itr)) != NULL) { + if(bg_record->bg_block_id) + if (!strcmp(bg_record->bg_block_id, + bg_block_id)) + break; } - + list_iterator_destroy(itr); + slurm_mutex_unlock(&block_state_mutex); + if(bg_record) + return bg_record; + else + return NULL; } /* All changes to the bg_list target_name must @@ -573,13 +586,30 @@ bool needed = true; hostlist_t hl; char *host = NULL; - char bg_down_node[128]; - if(bg_record->job_running > NO_JOB_RUNNING) - slurm_fail_job(bg_record->job_running); + if(bg_record->job_running > NO_JOB_RUNNING) { + int rc; + slurmctld_lock_t job_write_lock = { + NO_LOCK, WRITE_LOCK, WRITE_LOCK, NO_LOCK }; + lock_slurmctld(job_write_lock); + debug2("Trying to requeue job %d", bg_record->job_running); + if((rc = job_requeue(0, bg_record->job_running, -1))) { + error("couldn't requeue job %u, failing it: %s", + bg_record->job_running, + slurm_strerror(rc)); + job_fail(bg_record->job_running); + } + unlock_slurmctld(job_write_lock); + slurm_mutex_lock(&block_state_mutex); + if(remove_from_bg_list(bg_job_block_list, bg_record) + == SLURM_SUCCESS) { + num_unused_cpus += bg_record->cpu_cnt; + } + slurm_mutex_unlock(&block_state_mutex); + } /* small blocks */ - if(bg_record->cpus_per_bp != procs_per_node) { + if(bg_record->cpu_cnt < procs_per_node) { debug2("small block"); goto end_it; } @@ -591,7 +621,7 @@ return; } while ((host = hostlist_shift(hl))) { - if (node_already_down(bg_down_node)) { + if (node_already_down(host)) { needed = false; free(host); break; @@ -610,59 +640,27 @@ sleep(1); } - slurm_mutex_lock(&block_state_mutex); - error("Setting Block %s to ERROR state.", bg_record->bg_block_id); - bg_record->job_running = BLOCK_ERROR_STATE; - bg_record->state = RM_PARTITION_ERROR; - slurm_mutex_unlock(&block_state_mutex); - trigger_block_error(); + put_block_in_error_state(bg_record, BLOCK_ERROR_STATE); return; } -#ifdef HAVE_BGL - -extern int set_ionodes(bg_record_t *bg_record) +extern int set_ionodes(bg_record_t *bg_record, int io_start, int io_nodes) { - int i = 0; - int start_bit = 0; - int size = 0; char bitstring[BITSIZE]; - - if(!bg_record) - return SLURM_ERROR; - /* set the bitmap blank here if it is a full node we don't - want anything set we also don't want the bg_record->ionodes set. - */ - bg_record->ionode_bitmap = bit_alloc(bluegene_numpsets); - if(bg_record->quarter == (uint16_t)NO_VAL) { - return SLURM_SUCCESS; - } - - start_bit = bluegene_quarter_ionode_cnt*bg_record->quarter; - - if(bg_record->nodecard != (uint16_t)NO_VAL - && bluegene_nodecard_ionode_cnt) { - start_bit += bluegene_nodecard_ionode_cnt*bg_record->nodecard; - size = bluegene_nodecard_ionode_cnt; - } else - size = bluegene_quarter_ionode_cnt; - size += start_bit; - if(size == start_bit) { - error("start bit is the same as the end bit %d", size); + if(!bg_record) return SLURM_ERROR; - } - for(i=start_bit; iionode_bitmap, i); + bg_record->ionode_bitmap = bit_alloc(bluegene_numpsets); + /* Set the correct ionodes being used in this block */ + bit_nset(bg_record->ionode_bitmap, io_start, io_start+io_nodes); bit_fmt(bitstring, BITSIZE, bg_record->ionode_bitmap); bg_record->ionodes = xstrdup(bitstring); - return SLURM_SUCCESS; } -#endif -extern int add_bg_record(List records, List used_nodes, blockreq_t *blockreq) +extern int add_bg_record(List records, List used_nodes, blockreq_t *blockreq, + bool no_check, bitoff_t io_start) { bg_record_t *bg_record = NULL; ba_node_t *ba_node = NULL; @@ -670,23 +668,15 @@ uid_t pw_uid; int i, len; int small_count = 0; -#ifdef HAVE_BGL - int node_cnt = 0; - uint16_t quarter = 0; - uint16_t nodecard = 0; - int small_size = 0; - bg_record_t *found_record = NULL; -#endif + if(!records) { fatal("add_bg_record: no records list given"); } bg_record = (bg_record_t*) xmalloc(sizeof(bg_record_t)); - bg_record->user_name = - xstrdup(bg_slurm_user_name); - bg_record->target_name = - xstrdup(bg_slurm_user_name); + bg_record->user_name = xstrdup(bg_slurm_user_name); + bg_record->target_name = xstrdup(bg_slurm_user_name); pw_uid = uid_from_string(bg_record->user_name); if(pw_uid == (uid_t) -1) { @@ -705,10 +695,8 @@ /* bg_record->boot_state = 0; Implicit */ /* bg_record->state = 0; Implicit */ #ifdef HAVE_BGL - bg_record->quarter = (uint16_t)NO_VAL; - bg_record->nodecard = (uint16_t)NO_VAL; debug2("asking for %s %d %d %s", - blockreq->block, blockreq->small128, blockreq->small32, + blockreq->block, blockreq->small32, blockreq->small128, convert_conn_type(blockreq->conn_type)); #else debug2("asking for %s %d %d %d %d %d %s", @@ -746,7 +734,7 @@ bg_record->node_use = SELECT_COPROCESSOR_MODE; #endif bg_record->conn_type = blockreq->conn_type; - bg_record->cpus_per_bp = procs_per_node; + bg_record->cpu_cnt = procs_per_node * bg_record->bp_count; bg_record->node_cnt = bluegene_bp_node_cnt * bg_record->bp_count; bg_record->job_running = NO_JOB_RUNNING; @@ -785,79 +773,8 @@ } } else { debug("adding a small block"); -#ifdef HAVE_BGL // remove this clause when other works. Only here to - // perserve old code - - /* if the ionode cnt for small32 is 0 then don't - allow a nodecard allocation - */ - if(!bluegene_nodecard_ionode_cnt) { - if(blockreq->small32) - fatal("There is an error in your " - "bluegene.conf file.\n" - "Can't create a 32 node block with " - "Numpsets=%u. (Try setting it to 64)", - bluegene_numpsets); - } - - if(blockreq->small32==0 && blockreq->small128==0) { - info("No specs given for this small block, " - "I am spliting this block into 4 128CnBlocks"); - blockreq->small128=4; - } - - i = (blockreq->small32*bluegene_nodecard_node_cnt) + - (blockreq->small128*bluegene_quarter_node_cnt); - if(i != bluegene_bp_node_cnt) - fatal("There is an error in your bluegene.conf file.\n" - "I am unable to request %d nodes consisting of " - "%u 32CnBlocks and\n%u 128CnBlocks in one " - "base partition with %u nodes.", - i, bluegene_bp_node_cnt, - blockreq->small32, blockreq->small128); - small_count = blockreq->small32+blockreq->small128; - /* Automatically create 4-way split if - * conn_type == SELECT_SMALL in bluegene.conf - * Here we go through each node listed and do the same thing - * for each node. - */ - itr = list_iterator_create(bg_record->bg_block_list); - while ((ba_node = list_next(itr)) != NULL) { - /* break base partition up into 16 parts */ - small_size = bluegene_bp_nodecard_cnt; - node_cnt = 0; - quarter = 0; - nodecard = 0; - for(i=0; ismall32) { - /* break base partition - up into 4 parts */ - small_size = 4; - } - - if(small_size == 4) - nodecard = (uint16_t)NO_VAL; - else - nodecard = i%4; - found_record = create_small_record(bg_record, - quarter, - nodecard); - - /* this needs to be an append so we - keep things in the order we got - them, they will be sorted later */ - list_append(records, found_record); - node_cnt += bluegene_bp_node_cnt/small_size; - if(node_cnt == 128) { - node_cnt = 0; - quarter++; - } - } - } - list_iterator_destroy(itr); - destroy_bg_record(bg_record); -#else // remove this when testing. Only here to perserve old code the - // code below is already for bgl + if(no_check) + goto no_check; /* if the ionode cnt for small32 is 0 then don't allow a sub quarter allocation */ @@ -934,6 +851,7 @@ + blockreq->small128 + blockreq->small256; #endif + no_check: /* Automatically create 2-way split if * conn_type == SELECT_SMALL in bluegene.conf * Here we go through each node listed and do the same thing @@ -942,17 +860,15 @@ itr = list_iterator_create(bg_record->bg_block_list); while ((ba_node = list_next(itr)) != NULL) { handle_small_record_request(records, blockreq, - bg_record, 0); + bg_record, io_start); } list_iterator_destroy(itr); destroy_bg_record(bg_record); -#endif // remove this when done testing } return SLURM_SUCCESS; } -#ifndef HAVE_BGL extern int handle_small_record_request(List records, blockreq_t *blockreq, bg_record_t *bg_record, bitoff_t start) { @@ -964,6 +880,9 @@ xassert(blockreq); xassert(bg_record); + xassert(start >= 0); + xassert(start < bluegene_numpsets); + #ifndef HAVE_BGL for(i=0; ismall16; i++) { bit_nset(ionodes, start, start); @@ -1036,7 +955,6 @@ return SLURM_SUCCESS; } -#endif extern int format_node_name(bg_record_t *bg_record, char *buf, int buf_size) { @@ -1050,6 +968,421 @@ return SLURM_SUCCESS; } +extern int down_nodecard(char *bp_name, bitoff_t io_start) +{ + List requests = NULL; + List delete_list = NULL; + ListIterator itr = NULL; + bg_record_t *bg_record = NULL, *found_record = NULL, tmp_record; + bg_record_t *smallest_bg_record = NULL; + struct node_record *node_ptr = NULL; + int bp_bit = 0; + static int io_cnt = NO_VAL; + static int create_size = NO_VAL; + static blockreq_t blockreq; + int rc = SLURM_SUCCESS; + + xassert(bp_name); + + if(io_cnt == NO_VAL) { + io_cnt = 1; + /* Translate 1 nodecard count to ionode count */ + if((io_cnt *= bluegene_io_ratio)) + io_cnt--; + /* make sure we create something that is able to be + created */ + if(bluegene_smallest_block < bluegene_nodecard_node_cnt) + create_size = bluegene_nodecard_node_cnt; + else + create_size = bluegene_smallest_block; + } + + node_ptr = find_node_record(bp_name); + if (!node_ptr) { + error ("down_sub_node_blocks: invalid node specified '%s'", + bp_name); + return EINVAL; + } + bp_bit = (node_ptr - node_record_table_ptr); + + memset(&blockreq, 0, sizeof(blockreq_t)); + + blockreq.conn_type = SELECT_SMALL; + blockreq.block = bp_name; + + debug3("here setting %d of %d and %d-%d of %d", + bp_bit, node_record_count, io_start, + io_start+io_cnt, bluegene_numpsets); + + memset(&tmp_record, 0, sizeof(bg_record_t)); + tmp_record.bp_count = 1; + tmp_record.node_cnt = bluegene_nodecard_node_cnt; + tmp_record.bitmap = bit_alloc(node_record_count); + bit_set(tmp_record.bitmap, bp_bit); + + tmp_record.ionode_bitmap = bit_alloc(bluegene_numpsets); + bit_nset(tmp_record.ionode_bitmap, io_start, io_start+io_cnt); + + slurm_mutex_lock(&block_state_mutex); + itr = list_iterator_create(bg_list); + while ((bg_record = list_next(itr))) { + if(!bit_test(bg_record->bitmap, bp_bit)) + continue; + + if(!blocks_overlap(bg_record, &tmp_record)) + continue; + + if(bg_record->job_running > NO_JOB_RUNNING) + slurm_fail_job(bg_record->job_running); + + /* mark every one of these in an error state */ + if(bluegene_layout_mode != LAYOUT_DYNAMIC) { + if(!delete_list) + delete_list = list_create(NULL); + list_append(delete_list, bg_record); + continue; + } + + /* below is only for dynamic modes since there are + never overlapping blocks there */ + /* if the block is smaller than the create size just + continue on. + */ + if(bg_record->node_cnt < create_size) + continue; + + if(!smallest_bg_record || + (smallest_bg_record->node_cnt > bg_record->node_cnt)) + smallest_bg_record = bg_record; + } + list_iterator_destroy(itr); + slurm_mutex_unlock(&block_state_mutex); + + if(bluegene_layout_mode != LAYOUT_DYNAMIC) { + debug3("running non-dynamic mode"); + if(delete_list) { + int cnt_set = 0; + /* don't lock here since it is handled inside + the put_block_in_error_state + */ + itr = list_iterator_create(delete_list); + while ((bg_record = list_next(itr))) { + /* we already handled this */ + if(bg_record->state == RM_PARTITION_ERROR) { + rc = SLURM_NO_CHANGE_IN_DATA; + continue; + } + + rc = put_block_in_error_state( + bg_record, BLOCK_ERROR_STATE); + cnt_set++; + } + if(cnt_set) + rc = SLURM_SUCCESS; + list_iterator_destroy(itr); + list_destroy(delete_list); + goto cleanup; + } + + debug("didn't get a smallest block"); + if(!node_already_down(bp_name)) { + time_t now = time(NULL); + char reason[128], time_str[32]; + slurm_make_time_str(&now, time_str, + sizeof(time_str)); + snprintf(reason, sizeof(reason), + "select_bluegene: " + "nodecard down [SLURM@%s]", + time_str); + slurm_drain_nodes(bp_name, reason); + } + rc = SLURM_SUCCESS; + goto cleanup; + } + + + if(smallest_bg_record) { + debug2("smallest dynamic block is %s", + smallest_bg_record->bg_block_id); + if(smallest_bg_record->state == RM_PARTITION_ERROR) { + rc = SLURM_NO_CHANGE_IN_DATA; + goto cleanup; + } + + while(smallest_bg_record->job_running > NO_JOB_RUNNING) + sleep(1); + + if(smallest_bg_record->node_cnt == create_size) { + rc = put_block_in_error_state( + smallest_bg_record, BLOCK_ERROR_STATE); + goto cleanup; + } + + if(create_size > smallest_bg_record->node_cnt) { + /* we should never get here. This means we + * have a create_size that is bigger than a + * block that is already made. + */ + rc = put_block_in_error_state( + smallest_bg_record, BLOCK_ERROR_STATE); + goto cleanup; + } + debug3("node count is %d", smallest_bg_record->node_cnt); + switch(smallest_bg_record->node_cnt) { +#ifndef HAVE_BGL + case 64: + blockreq.small32 = 2; + break; + case 256: + blockreq.small32 = 8; + break; +#endif + case 128: + blockreq.small32 = 4; + break; + case 512: + default: + blockreq.small32 = 16; + break; + } + + if(create_size != bluegene_nodecard_node_cnt) { + blockreq.small128 = blockreq.small32 / 4; + blockreq.small32 = 0; + } + /* set the start to be the same as the start of the + ionode_bitmap. If no ionodes set (not a small + block) set io_start = 0. */ + if((io_start = bit_ffs(smallest_bg_record->ionode_bitmap)) + == -1) + io_start = 0; + } else { + switch(create_size) { +#ifndef HAVE_BGL + case 64: + blockreq.small64 = 8; + break; + case 256: + blockreq.small256 = 2; +#endif + case 32: + blockreq.small32 = 16; + break; + case 128: + blockreq.small128 = 4; + break; + case 512: + if(!node_already_down(bp_name)) { + time_t now = time(NULL); + char reason[128], time_str[32]; + slurm_make_time_str(&now, time_str, + sizeof(time_str)); + snprintf(reason, sizeof(reason), + "select_bluegene: " + "nodecard down [SLURM@%s]", + time_str); + slurm_drain_nodes(bp_name, reason); + } + rc = SLURM_SUCCESS; + goto cleanup; + break; + default: + error("Unknown create size of %d", create_size); + break; + } + /* since we don't have a block in this midplane + we need to start at the beginning. */ + io_start = 0; + /* we also need a bg_block to pretend to be the + smallest block that takes up the entire midplane. */ + } + + + /* Here we need to add blocks that take up nodecards on this + midplane. Since Slurm only keeps track of midplanes + natively this is the only want to handle this case. + */ + requests = list_create(destroy_bg_record); + add_bg_record(requests, NULL, &blockreq, 1, io_start); + + delete_list = list_create(NULL); + while((bg_record = list_pop(requests))) { + slurm_mutex_lock(&block_state_mutex); + itr = list_iterator_create(bg_list); + while((found_record = list_next(itr))) { + if(!blocks_overlap(bg_record, found_record)) + continue; + list_push(delete_list, found_record); + list_remove(itr); + num_block_to_free++; + } + list_iterator_destroy(itr); + slurm_mutex_unlock(&block_state_mutex); + + /* we need to add this record since it doesn't exist */ + if(configure_block(bg_record) == SLURM_ERROR) { + destroy_bg_record(bg_record); + error("down_sub_node_blocks: " + "unable to configure block in api"); + continue; + } + + debug("adding block %s to fill in small blocks " + "around bad nodecards", + bg_record->bg_block_id); + print_bg_record(bg_record); + slurm_mutex_lock(&block_state_mutex); + list_append(bg_list, bg_record); + slurm_mutex_unlock(&block_state_mutex); + if(bit_overlap(bg_record->ionode_bitmap, + tmp_record.ionode_bitmap)) { + /* here we know the error block doesn't exist + so just set the state here */ + rc = put_block_in_error_state( + bg_record, BLOCK_ERROR_STATE); + } + } + list_destroy(requests); + + slurm_mutex_lock(&block_state_mutex); + free_block_list(delete_list); + list_destroy(delete_list); + sort_bg_record_inc_size(bg_list); + slurm_mutex_unlock(&block_state_mutex); + last_bg_update = time(NULL); + +cleanup: + FREE_NULL_BITMAP(tmp_record.bitmap); + FREE_NULL_BITMAP(tmp_record.ionode_bitmap); + + return rc; + +} + +extern int up_nodecard(char *bp_name, bitstr_t *ionode_bitmap) +{ + ListIterator itr = NULL; + bg_record_t *bg_record = NULL; + struct node_record *node_ptr = NULL; + int bp_bit = 0; + int ret = 0; + + xassert(bp_name); + xassert(ionode_bitmap); + + node_ptr = find_node_record(bp_name); + if (!node_ptr) { + error ("down_sub_node_blocks: invalid node specified %s", + bp_name); + return EINVAL; + } + bp_bit = (node_ptr - node_record_table_ptr); + + slurm_mutex_lock(&block_state_mutex); + itr = list_iterator_create(bg_list); + while((bg_record = list_next(itr))) { + if(bg_record->job_running != BLOCK_ERROR_STATE) + continue; + if(!bit_test(bg_record->bitmap, bp_bit)) + continue; + + if(!bit_overlap(bg_record->ionode_bitmap, ionode_bitmap)) { + continue; + } + resume_block(bg_record); + } + list_iterator_destroy(itr); + slurm_mutex_unlock(&block_state_mutex); + + /* FIX ME: This needs to call the opposite of + slurm_drain_nodes which does not yet exist. + */ + if((ret = node_already_down(bp_name))) { + /* means it was drained */ + if(ret == 2) { + /* debug("node %s put back into service after " */ +/* "being in an error state", */ +/* bp_name); */ + } + } + + return SLURM_SUCCESS; +} + +extern int put_block_in_error_state(bg_record_t *bg_record, int state) +{ + uid_t pw_uid; + + xassert(bg_record); + + /* only check this if the blocks are created, meaning this + isn't at startup. + */ + if(blocks_are_created) { + /* Since we are putting this block in an error state we need + to wait for the job to be removed. We don't really + need to free the block though since we may just + want it to be in an error state for some reason. */ + while(bg_record->job_running > NO_JOB_RUNNING) + sleep(1); + } + + info("Setting Block %s to ERROR state.", bg_record->bg_block_id); + /* we add the block to these lists so we don't try to schedule + on them. */ + if(!block_ptr_exist_in_list(bg_job_block_list, bg_record)) { + list_push(bg_job_block_list, bg_record); + num_unused_cpus -= bg_record->cpu_cnt; + } + if(!block_ptr_exist_in_list(bg_booted_block_list, bg_record)) + list_push(bg_booted_block_list, bg_record); + + slurm_mutex_lock(&block_state_mutex); + bg_record->job_running = state; + bg_record->state = RM_PARTITION_ERROR; + + xfree(bg_record->user_name); + xfree(bg_record->target_name); + bg_record->user_name = xstrdup(bg_slurm_user_name); + bg_record->target_name = xstrdup(bg_slurm_user_name); + + pw_uid = uid_from_string(bg_record->user_name); + if(pw_uid == (uid_t) -1) { + error("No such user: %s", bg_record->user_name); + } else { + bg_record->user_uid = pw_uid; + } + slurm_mutex_unlock(&block_state_mutex); + + trigger_block_error(); + last_bg_update = time(NULL); + + return SLURM_SUCCESS; +} + +/* block_state_mutex should be locked before calling */ +extern int resume_block(bg_record_t *bg_record) +{ + xassert(bg_record); + + if(bg_record->job_running >= NO_JOB_RUNNING) + return SLURM_SUCCESS; + + info("Block %s put back into service after " + "being in an error state.", + bg_record->bg_block_id); + + if(remove_from_bg_list(bg_job_block_list, bg_record) == SLURM_SUCCESS) + num_unused_cpus += bg_record->cpu_cnt; + remove_from_bg_list(bg_booted_block_list, bg_record); + + bg_record->job_running = NO_JOB_RUNNING; + bg_record->state = RM_PARTITION_FREE; + last_bg_update = time(NULL); + + return SLURM_SUCCESS; +} + /************************* local functions ***************************/ #ifdef HAVE_BG @@ -1125,6 +1458,8 @@ alpha_num[node_a->coord[Z]]); return 0; } + + #endif //HAVE_BG diff -Nru slurm-llnl-1.3.13/src/plugins/select/bluegene/plugin/bg_record_functions.h slurm-llnl-1.3.15/src/plugins/select/bluegene/plugin/bg_record_functions.h --- slurm-llnl-1.3.13/src/plugins/select/bluegene/plugin/bg_record_functions.h 2008-11-18 00:20:37.000000000 +0100 +++ slurm-llnl-1.3.15/src/plugins/select/bluegene/plugin/bg_record_functions.h 2009-04-01 19:20:31.000000000 +0200 @@ -98,13 +98,9 @@ int job_running; /* job id of job running of if * block is in an error state * BLOCK_ERROR_STATE */ - int cpus_per_bp; /* count of cpus per base part */ - uint32_t node_cnt; /* count of nodes per block */ + uint32_t cpu_cnt; /* count of cpus per block */ + uint32_t node_cnt; /* count of cnodes per block */ #ifdef HAVE_BGL - uint16_t quarter; /* used for small blocks - determine quarter of BP */ - uint16_t nodecard; /* used for small blocks - determine nodecard of quarter */ char *blrtsimage; /* BlrtsImage for this block */ #endif char *linuximage; /* LinuxImage/CnloadImage for @@ -120,6 +116,7 @@ extern void print_bg_record(bg_record_t *record); extern void destroy_bg_record(void *object); extern int block_exist_in_list(List my_list, bg_record_t *bg_record); +extern int block_ptr_exist_in_list(List my_list, bg_record_t *bg_record); extern void process_nodes(bg_record_t *bg_reord, bool startup); extern List copy_bg_list(List in_list); extern void copy_bg_record(bg_record_t *fir_record, bg_record_t *sec_record); @@ -134,14 +131,17 @@ extern int update_block_user(bg_record_t *bg_block_id, int set); extern void drain_as_needed(bg_record_t *bg_record, char *reason); -#ifdef HAVE_BGL -extern int set_ionodes(bg_record_t *bg_record); -#endif +extern int set_ionodes(bg_record_t *bg_record, int io_start, int io_nodes); -extern int add_bg_record(List records, List used_nodes, blockreq_t *blockreq); +extern int add_bg_record(List records, List used_nodes, blockreq_t *blockreq, + bool no_check, bitoff_t io_start); extern int handle_small_record_request(List records, blockreq_t *blockreq, bg_record_t *bg_record, bitoff_t start); extern int format_node_name(bg_record_t *bg_record, char *buf, int buf_size); +extern int down_nodecard(char *bp_name, bitoff_t io_start); +extern int up_nodecard(char *bp_name, bitstr_t *ionode_bitmap); +extern int put_block_in_error_state(bg_record_t *bg_record, int state); +extern int resume_block(bg_record_t *bg_record); #endif /* _BLUEGENE_BG_RECORD_FUNCTIONS_H_ */ diff -Nru slurm-llnl-1.3.13/src/plugins/select/bluegene/plugin/bg_switch_connections.c slurm-llnl-1.3.15/src/plugins/select/bluegene/plugin/bg_switch_connections.c --- slurm-llnl-1.3.13/src/plugins/select/bluegene/plugin/bg_switch_connections.c 2008-12-10 20:13:46.000000000 +0100 +++ slurm-llnl-1.3.15/src/plugins/select/bluegene/plugin/bg_switch_connections.c 2009-04-01 19:20:31.000000000 +0200 @@ -2,7 +2,7 @@ * bg_switch_connections.c - Blue Gene switch management functions, * establish switch connections * - * $Id: bg_switch_connections.c 15919 2008-12-10 19:13:46Z da $ + * $Id: bg_switch_connections.c 17104 2009-04-01 17:20:31Z da $ ***************************************************************************** * Copyright (C) 2004 The Regents of the University of California. * Produced at Lawrence Livermore National Laboratory (cf, DISCLAIMER). @@ -127,7 +127,7 @@ for (i=0; iused && ba_conn->port_tar != source) { switch(ba_conn->port_tar) { case 0: - conn.p2 = RM_PORT_S0; + conn[i].p2 = RM_PORT_S0; break; case 3: - conn.p2 = RM_PORT_S3; + conn[i].p2 = RM_PORT_S3; break; case 5: - conn.p2 = RM_PORT_S5; + conn[i].p2 = RM_PORT_S5; break; default: error("we are trying to connection %d -> %d " @@ -218,45 +220,15 @@ source, ba_conn->port_tar); break; } - conn.part_state = RM_PARTITION_READY; - - if(firstconnect) { - if ((rc = bridge_set_data( - curr_switch, - RM_SwitchFirstConnection, - &conn)) - != STATUS_OK) { - list_iterator_destroy(itr); - - fatal("bridge_set_data" - "(RM_SwitchFirstConnection): " - "%s", - bg_err_str(rc)); - return SLURM_ERROR; - } - firstconnect=0; - } else { - if ((rc = bridge_set_data( - curr_switch, - RM_SwitchNextConnection, - &conn)) - != STATUS_OK) { - list_iterator_destroy(itr); - - fatal("bridge_set_data" - "(RM_SwitchNextConnection): %s", - bg_err_str(rc)); - return SLURM_ERROR; - } - } - - conn_num++; + conn[i].part_state = RM_PARTITION_READY; debug2("adding %d -> %d", source, ba_conn->port_tar); + list_push(conn_list, &conn[i]); } } - if(conn_num) { - if ((rc = bridge_set_data(curr_switch, RM_SwitchConnNum, - &conn_num)) + + i = list_count(conn_list); + if(i) { + if ((rc = bridge_set_data(curr_switch, RM_SwitchConnNum, &i)) != STATUS_OK) { fatal("bridge_set_data: RM_SwitchConnNum: %s", bg_err_str(rc)); @@ -265,9 +237,43 @@ } } else { debug("we got a switch with no connections"); - return SLURM_ERROR; + list_destroy(conn_list); + return SLURM_ERROR; + } + + /* Now we can add them to the mix */ + while((conn_ptr = list_pop(conn_list))) { + if(firstconnect) { + if ((rc = bridge_set_data( + curr_switch, + RM_SwitchFirstConnection, + conn_ptr)) + != STATUS_OK) { + fatal("bridge_set_data" + "(RM_SwitchFirstConnection): " + "%s", + bg_err_str(rc)); + list_destroy(conn_list); + return SLURM_ERROR; + } + firstconnect=0; + } else { + if ((rc = bridge_set_data( + curr_switch, + RM_SwitchNextConnection, + conn_ptr)) + != STATUS_OK) { + fatal("bridge_set_data" + "(RM_SwitchNextConnection): %s", + bg_err_str(rc)); + list_destroy(conn_list); + return SLURM_ERROR; + } + } } + list_destroy(conn_list); + return SLURM_SUCCESS; } #endif @@ -316,207 +322,21 @@ return switch_count; } -#ifdef HAVE_BGL extern int configure_small_block(bg_record_t *bg_record) { int rc = SLURM_SUCCESS; #ifdef HAVE_BG_FILES bool small = true; - ListIterator itr; - ba_node_t* ba_node = NULL; - rm_BP_t *curr_bp = NULL; - rm_bp_id_t bp_id = NULL; - int num_ncards = 0; - rm_nodecard_t *ncard; - rm_nodecard_list_t *ncard_list = NULL; - rm_quarter_t quarter; - int num, i; -#endif - if(bg_record->bp_count != 1) { - error("Requesting small block with %d bps, needs to be 1.", - bg_record->bp_count); - return SLURM_ERROR; - } - -#ifdef HAVE_BG_FILES - /* set that we are doing a small block */ - - if ((rc = bridge_set_data(bg_record->bg_block, RM_PartitionSmall, - &small)) != STATUS_OK) { - - fatal("bridge_set_data(RM_PartitionPsetsPerBP)", - bg_err_str(rc)); - } - - num_ncards = bg_record->node_cnt/bluegene_nodecard_node_cnt; - if(num_ncards < 1) - num_ncards = 1; - - if ((rc = bridge_set_data(bg_record->bg_block, - RM_PartitionNodeCardNum, - &num_ncards)) - != STATUS_OK) { - - fatal("bridge_set_data: RM_PartitionBPNum: %s", - bg_err_str(rc)); - } - - - itr = list_iterator_create(bg_record->bg_block_list); - ba_node = list_next(itr); - list_iterator_destroy(itr); - - if (_get_bp_by_location(bg, ba_node->coord, &curr_bp) - == SLURM_ERROR) { - fatal("_get_bp_by_location()"); - } - - /* Set the one BP */ - - if ((rc = bridge_set_data(bg_record->bg_block, - RM_PartitionBPNum, - &bg_record->bp_count)) - != STATUS_OK) { - - fatal("bridge_set_data: RM_PartitionBPNum: %s", - bg_err_str(rc)); - return SLURM_ERROR; - } - if ((rc = bridge_set_data(bg_record->bg_block, - RM_PartitionFirstBP, - curr_bp)) - != STATUS_OK) { - - fatal("bridge_set_data(" - "BRIDGE_PartitionFirstBP): %s", - bg_err_str(rc)); - return SLURM_ERROR; - } - - - /* find the bp_id of the bp to get the small32 */ - if ((rc = bridge_get_data(curr_bp, RM_BPID, &bp_id)) - != STATUS_OK) { - error("bridge_get_data(): %d", rc); - return SLURM_ERROR; - } - - - if(!bp_id) { - error("No BP ID was returned from database"); - return SLURM_ERROR; - } - - if ((rc = bridge_get_nodecards(bp_id, &ncard_list)) - != STATUS_OK) { - error("bridge_get_nodecards(%s): %d", - bp_id, rc); - free(bp_id); - return SLURM_ERROR; - } - free(bp_id); - - - if((rc = bridge_get_data(ncard_list, RM_NodeCardListSize, &num)) - != STATUS_OK) { - error("bridge_get_data(RM_NodeCardListSize): %s", - bg_err_str(rc)); - return SLURM_ERROR; - } - num_ncards = 0; - for(i=0; iquarter != quarter) - continue; - if(bg_record->nodecard != (uint16_t) NO_VAL) { - if(bg_record->nodecard != (i%4)) - continue; - } - - - if (num_ncards) { - if ((rc = bridge_set_data(bg_record->bg_block, - RM_PartitionNextNodeCard, - ncard)) - != STATUS_OK) { - - fatal("bridge_set_data(" - "RM_PartitionNextNodeCard): %s", - bg_err_str(rc)); - } - } else { - if ((rc = bridge_set_data(bg_record->bg_block, - RM_PartitionFirstNodeCard, - ncard)) - != STATUS_OK) { - - fatal("bridge_set_data(" - "RM_PartitionFirstNodeCard): %s", - bg_err_str(rc)); - } - } - - num_ncards++; - if(num_ncards == 4) - break; - } -cleanup: - if ((rc = bridge_free_nodecard_list(ncard_list)) != STATUS_OK) { - error("bridge_free_nodecard_list(): %s", bg_err_str(rc)); - return SLURM_ERROR; - } -#endif - debug2("making the small block"); - return rc; -} - -#else - -extern int configure_small_block(bg_record_t *bg_record) -{ - int rc = SLURM_SUCCESS; -#ifdef HAVE_BG_FILES - bool small = true; - ListIterator itr; ba_node_t* ba_node = NULL; rm_BP_t *curr_bp = NULL; rm_bp_id_t bp_id = NULL; +#ifndef HAVE_BGL rm_nodecard_id_t nc_char = NULL; +#endif int nc_id = 0; - int num_ncards = 0, sub_nodecard = 0, ionode_card = 0; + int num_ncards = 0, sub_nodecard = 0, ionode_card = 0, nc_count = 0; rm_nodecard_t *ncard; rm_nodecard_list_t *ncard_list = NULL; -#ifdef HAVE_BGL - rm_quarter_t quarter; -#endif int num, i; int use_nc[bluegene_bp_nodecard_cnt]; double nc_pos = 0; @@ -527,7 +347,8 @@ bg_record->bp_count); return SLURM_ERROR; } - +/* info("configuring small block on ionodes %s out of %d ncs", */ +/* bg_record->ionodes, bluegene_bp_nodecard_cnt); */ #ifdef HAVE_BG_FILES /* set that we are doing a small block */ if ((rc = bridge_set_data(bg_record->bg_block, RM_PartitionSmall, @@ -570,10 +391,7 @@ bg_err_str(rc)); } - - itr = list_iterator_create(bg_record->bg_block_list); - ba_node = list_next(itr); - list_iterator_destroy(itr); + ba_node = list_peek(bg_record->bg_block_list); if (_get_bp_by_location(bg, ba_node->coord, &curr_bp) == SLURM_ERROR) { @@ -632,7 +450,12 @@ bg_err_str(rc)); return SLURM_ERROR; } - num_ncards = 0; + if(num_ncards > num) { + error("You requested more (%d > %d) nodecards " + "than are available on this block %s", + num_ncards, num, bg_record->nodes); + } + for(i=0; ibg_block, RM_PartitionNextNodeCard, ncard)) @@ -778,8 +610,8 @@ } } - num_ncards++; - + nc_count++; +#ifndef HAVE_BGL if(sub_nodecard) { if((rc = bridge_free_nodecard(ncard)) != STATUS_OK) { error("bridge_free_nodecard(): %s", @@ -788,6 +620,9 @@ goto cleanup; } } +#endif + if(nc_count == num_ncards) + break; } cleanup: if ((rc = bridge_free_nodecard_list(ncard_list)) != STATUS_OK) { @@ -798,7 +633,6 @@ debug2("making the small block"); return rc; } -#endif /** * connect the given switch up with the given connections @@ -825,7 +659,7 @@ bg_record->bp_count = 0; itr = list_iterator_create(bg_record->bg_block_list); - while ((ba_node = (ba_node_t *) list_next(itr)) != NULL) { + while ((ba_node = list_next(itr))) { if(ba_node->used) { bg_record->bp_count++; } @@ -857,7 +691,7 @@ debug3("switch count %d", bg_record->switch_count); list_iterator_reset(itr); - while ((ba_node = (ba_node_t *) list_next(itr)) != NULL) { + while ((ba_node = list_next(itr))) { #ifdef HAVE_BG_FILES if (_get_bp_by_location(bg, ba_node->coord, &curr_bp) == SLURM_ERROR) { diff -Nru slurm-llnl-1.3.13/src/plugins/select/bluegene/plugin/block_sys.c slurm-llnl-1.3.15/src/plugins/select/bluegene/plugin/block_sys.c --- slurm-llnl-1.3.13/src/plugins/select/bluegene/plugin/block_sys.c 2009-01-06 19:20:48.000000000 +0100 +++ slurm-llnl-1.3.15/src/plugins/select/bluegene/plugin/block_sys.c 2009-04-06 22:18:23.000000000 +0200 @@ -1,7 +1,7 @@ /*****************************************************************************\ * block_sys.c - component used for wiring up the blocks * - * $Id: block_sys.c 16146 2009-01-06 18:20:48Z da $ + * $Id: block_sys.c 17162 2009-04-06 20:18:23Z da $ ***************************************************************************** * Copyright (C) 2004 The Regents of the University of California. * Produced at Lawrence Livermore National Laboratory (cf, DISCLAIMER). @@ -119,6 +119,9 @@ error("bridge_set_data(RM_PartitionRamdiskImg)", bg_err_str(rc)); #else + struct tm my_tm; + struct timeval my_tv; + if ((rc = bridge_set_data(bg_record->bg_block, RM_PartitionCnloadImg, bg_record->linuximage)) != STATUS_OK) @@ -130,6 +133,15 @@ error("bridge_set_data(RM_PartitionIoloadImg)", bg_err_str(rc)); + gettimeofday(&my_tv, NULL); + localtime_r(&my_tv.tv_sec, &my_tm); + bg_record->bg_block_id = xstrdup_printf( + "RMP%2.2d%2.2s%2.2d%2.2d%2.2d%3.3d", + my_tm.tm_mday, mon_abbr(my_tm.tm_mon), + my_tm.tm_hour, my_tm.tm_min, my_tm.tm_sec, my_tv.tv_usec/1000); + if ((rc = bridge_set_data(bg_record->bg_block, RM_PartitionID, + bg_record->bg_block_id)) != STATUS_OK) + error("bridge_set_data(RM_PartitionID)", bg_err_str(rc)); #endif if ((rc = bridge_set_data(bg_record->bg_block, RM_PartitionMloaderImg, bg_record->mloaderimage)) != STATUS_OK) @@ -155,11 +167,6 @@ != STATUS_OK) error("bridge_set_data(RM_PartitionUserName)", bg_err_str(rc)); -/* info("setting it here"); */ -/* bg_record->bg_block_id = "RMP101"; */ -/* if ((rc = bridge_set_data(bg_record->bg_block, RM_PartitionID, */ -/* &bg_record->bg_block_id)) != STATUS_OK) */ -/* error("bridge_set_data(RM_PartitionID)", bg_err_str(rc)); */ #endif } @@ -262,8 +269,7 @@ #ifdef HAVE_BG_FILES #ifdef HAVE_BGL -static int _find_nodecard(bg_record_t *bg_record, - rm_partition_t *block_ptr) +static int _find_nodecard(rm_partition_t *block_ptr, int *nc_id) { char *my_card_name = NULL; char *card_name = NULL; @@ -275,6 +281,9 @@ rm_nodecard_t *ncard = NULL; rm_BP_t *curr_bp = NULL; + xassert(block_ptr); + xassert(nc_id); + if((rc = bridge_get_data(block_ptr, RM_PartitionFirstNodeCard, &ncard)) @@ -349,12 +358,12 @@ rc = SLURM_ERROR; goto cleanup; } - if(strcmp(my_card_name,card_name)) { + if(strcmp(my_card_name, card_name)) { free(card_name); continue; } free(card_name); - bg_record->nodecard = (i%4); + (*nc_id) = i; break; } cleanup: @@ -372,7 +381,7 @@ #endif _pre_allocate(bg_record); - if(bg_record->cpus_per_bp < procs_per_node) + if(bg_record->cpu_cnt < procs_per_node) configure_small_block(bg_record); else configure_block_switches(bg_record); @@ -389,7 +398,7 @@ { int rc = SLURM_SUCCESS; - int bp_cnt, i; + int bp_cnt, i, nc_cnt, io_cnt; rm_element_t *bp_ptr = NULL; rm_bp_id_t bpid; rm_partition_t *block_ptr = NULL; @@ -404,11 +413,8 @@ rm_partition_list_t *block_list = NULL; rm_partition_state_flag_t state = PARTITION_ALL_FLAG; rm_nodecard_t *ncard = NULL; -#ifdef HAVE_BGL - rm_quarter_t quarter; -#else int nc_id, io_start; -#endif + bool small = false; hostlist_t hostlist; /* expanded form of hosts */ @@ -465,6 +471,7 @@ free(tmp_char); continue; } + if(bg_recover) { if ((rc = bridge_get_block(tmp_char, &block_ptr)) != STATUS_OK) { @@ -484,51 +491,45 @@ free(tmp_char); bg_record->state = NO_VAL; -#ifdef HAVE_BGL - bg_record->quarter = (uint16_t) NO_VAL; - bg_record->nodecard = (uint16_t) NO_VAL; -#endif - bg_record->job_running = NO_JOB_RUNNING; - +#ifndef HAVE_BGL if ((rc = bridge_get_data(block_ptr, - RM_PartitionBPNum, + RM_PartitionSize, &bp_cnt)) != STATUS_OK) { - error("bridge_get_data(RM_BPNum): %s", + error("bridge_get_data(RM_PartitionSize): %s", bg_err_str(rc)); - bp_cnt = 0; + goto clean_up; } if(bp_cnt==0) goto clean_up; - bg_record->bp_count = bp_cnt; -#ifndef HAVE_BGL + bg_record->node_cnt = bp_cnt; + bg_record->cpu_cnt = bluegene_proc_ratio * bg_record->node_cnt; +#endif + bg_record->job_running = NO_JOB_RUNNING; + if ((rc = bridge_get_data(block_ptr, - RM_PartitionSize, + RM_PartitionBPNum, &bp_cnt)) != STATUS_OK) { - error("bridge_get_data(RM_PartitionSize): %s", + error("bridge_get_data(RM_BPNum): %s", bg_err_str(rc)); - bp_cnt = 0; + goto clean_up; } if(bp_cnt==0) goto clean_up; + bg_record->bp_count = bp_cnt; - bg_record->node_cnt = bp_cnt; - bg_record->cpus_per_bp = - bluegene_proc_ratio * bg_record->node_cnt; - -#endif - debug3("has %d BPs", - bg_record->bp_count); + debug3("has %d BPs", bg_record->bp_count); if ((rc = bridge_get_data(block_ptr, RM_PartitionSwitchNum, &bg_record->switch_count)) != STATUS_OK) { error("bridge_get_data(RM_PartitionSwitchNum): %s", bg_err_str(rc)); + goto clean_up; } if ((rc = bridge_get_data(block_ptr, RM_PartitionSmall, @@ -536,10 +537,40 @@ != STATUS_OK) { error("bridge_get_data(RM_PartitionSmall): %s", bg_err_str(rc)); - bp_cnt = 0; + goto clean_up; } - + if(small) { + if ((rc = bridge_get_data(block_ptr, + RM_PartitionOptions, + &tmp_char)) + != STATUS_OK) { + error("bridge_get_data(RM_PartitionOptions): " + "%s", bg_err_str(rc)); + goto clean_up; + } else if(tmp_char) { + switch(tmp_char[0]) { + case 's': + bg_record->conn_type = SELECT_HTC_S; + break; + case 'd': + bg_record->conn_type = SELECT_HTC_D; + break; + case 'v': + bg_record->conn_type = SELECT_HTC_V; + break; + case 'l': + bg_record->conn_type = SELECT_HTC_L; + break; + default: + bg_record->conn_type = SELECT_SMALL; + break; + } + + free(tmp_char); + } else + bg_record->conn_type = SELECT_SMALL; + if((rc = bridge_get_data(block_ptr, RM_PartitionFirstNodeCard, &ncard)) @@ -547,56 +578,53 @@ error("bridge_get_data(" "RM_PartitionFirstNodeCard): %s", bg_err_str(rc)); - bp_cnt = 0; + goto clean_up; } - bg_record->conn_type = SELECT_SMALL; if((rc = bridge_get_data(block_ptr, RM_PartitionNodeCardNum, - &i)) + &nc_cnt)) != STATUS_OK) { error("bridge_get_data(" "RM_PartitionNodeCardNum): %s", bg_err_str(rc)); - bp_cnt = 0; + goto clean_up; } #ifdef HAVE_BGL - if(i == 1) { - _find_nodecard(bg_record, block_ptr); - i = bluegene_bp_nodecard_cnt; - } + /* Translate nodecard count to ionode count */ + if((io_cnt = nc_cnt * bluegene_io_ratio)) + io_cnt--; + + nc_id = 0; + if(nc_cnt == 1) + _find_nodecard(block_ptr, &nc_id); + + bg_record->node_cnt = + nc_cnt * bluegene_nodecard_node_cnt; + bg_record->cpu_cnt = + bluegene_proc_ratio * bg_record->node_cnt; + if ((rc = bridge_get_data(ncard, RM_NodeCardQuarter, - &quarter)) != STATUS_OK) { + &io_start)) != STATUS_OK) { error("bridge_get_data(CardQuarter): %d",rc); - bp_cnt = 0; + goto clean_up; } - bg_record->quarter = quarter; - - - debug3("%s is in quarter %d nodecard %d", - bg_record->bg_block_id, - bg_record->quarter, - bg_record->nodecard); - bg_record->cpus_per_bp = procs_per_node/i; - bg_record->node_cnt = bluegene_bp_node_cnt/i; - if(set_ionodes(bg_record) == SLURM_ERROR) - error("couldn't create ionode_bitmap " - "for %d.%d", - bg_record->quarter, bg_record->nodecard); + io_start *= bluegene_quarter_ionode_cnt; + io_start += bluegene_nodecard_ionode_cnt * (nc_id%4); #else /* Translate nodecard count to ionode count */ - if((i *= bluegene_io_ratio)) - i--; + if((io_cnt = nc_cnt * bluegene_io_ratio)) + io_cnt--; if ((rc = bridge_get_data(ncard, RM_NodeCardID, &tmp_char)) != STATUS_OK) { error("bridge_get_data(RM_NodeCardID): %d",rc); - bp_cnt = 0; + goto clean_up; } - if(bp_cnt==0) + if(!tmp_char) goto clean_up; /* From the first nodecard id we can figure @@ -605,15 +633,54 @@ nc_id = atoi((char*)tmp_char+1); free(tmp_char); io_start = nc_id * bluegene_io_ratio; - bg_record->ionode_bitmap = bit_alloc(bluegene_numpsets); - /* Set the correct ionodes being used in this - block */ - bit_nset(bg_record->ionode_bitmap, - io_start, io_start+i); + if(bg_record->node_cnt < bluegene_nodecard_node_cnt) { + rm_ionode_t *ionode; + + /* figure out the ionode we are using */ + if ((rc = bridge_get_data( + ncard, + RM_NodeCardFirstIONode, + &ionode)) != STATUS_OK) { + error("bridge_get_data(" + "RM_NodeCardFirstIONode): %d", + rc); + goto clean_up; + } + if ((rc = bridge_get_data(ionode, + RM_IONodeID, + &tmp_char)) + != STATUS_OK) { + error("bridge_get_data(" + "RM_NodeCardIONodeNum): %s", + bg_err_str(rc)); + rc = SLURM_ERROR; + goto clean_up; + } + + if(!tmp_char) + goto clean_up; + /* just add the ionode num to the + * io_start */ + io_start += atoi((char*)tmp_char+1); + free(tmp_char); + /* make sure i is 0 since we are only using + * 1 ionode */ + io_cnt = 0; + } #endif + + if(set_ionodes(bg_record, io_start, io_cnt) + == SLURM_ERROR) + error("couldn't create ionode_bitmap " + "for ionodes %d to %d", + io_start, io_start+io_cnt); + debug3("%s uses ionodes %s", + bg_record->bg_block_id, + bg_record->ionodes); } else { #ifdef HAVE_BGL - bg_record->cpus_per_bp = procs_per_node; + bg_record->cpu_cnt = procs_per_node + * bg_record->bp_count; bg_record->node_cnt = bluegene_bp_node_cnt * bg_record->bp_count; #endif @@ -624,14 +691,14 @@ error("bridge_get_data" "(RM_PartitionConnection): %s", bg_err_str(rc)); + goto clean_up; } /* Set the bitmap blank here if it is a full node we don't want anything set we also don't want the bg_record->ionodes set. */ bg_record->ionode_bitmap = bit_alloc(bluegene_numpsets); - } - + } bg_record->bg_block_list = get_and_set_block_wiring(bg_record->bg_block_id); @@ -640,11 +707,6 @@ bg_record->bg_block_id); hostlist = hostlist_create(NULL); - /* this needs to be changed for small blocks, - we just don't know what they are suppose to look - like just yet. - */ - for (i=0; istate)) != STATUS_OK) { error("bridge_get_data(RM_PartitionState): %s", bg_err_str(rc)); + goto clean_up; } else if(bg_record->state == RM_PARTITION_CONFIGURING) bg_record->boot_state = 1; else @@ -736,7 +799,10 @@ bg_record->state); process_nodes(bg_record, false); - + + /* We can stop processing information now since we + don't need to rest of the information to decide if + this is the correct block. */ if(bluegene_layout_mode == LAYOUT_DYNAMIC) { bg_record_t *tmp_record = xmalloc(sizeof(bg_record_t)); copy_bg_record(bg_record, tmp_record); @@ -747,6 +813,7 @@ &bp_cnt)) != STATUS_OK) { error("bridge_get_data(RM_PartitionUsersNum): %s", bg_err_str(rc)); + goto clean_up; } else { if(bp_cnt==0) { @@ -765,6 +832,7 @@ error("bridge_get_data" "(RM_PartitionFirstUser): %s", bg_err_str(rc)); + goto clean_up; } if(!user_name) { error("No user name was " @@ -802,6 +870,7 @@ != STATUS_OK) { error("bridge_get_data(RM_PartitionBlrtsImg): %s", bg_err_str(rc)); + goto clean_up; } if(!user_name) { error("No BlrtsImg was returned from database"); @@ -815,6 +884,7 @@ != STATUS_OK) { error("bridge_get_data(RM_PartitionLinuxImg): %s", bg_err_str(rc)); + goto clean_up; } if(!user_name) { error("No LinuxImg was returned from database"); @@ -828,6 +898,7 @@ != STATUS_OK) { error("bridge_get_data(RM_PartitionRamdiskImg): %s", bg_err_str(rc)); + goto clean_up; } if(!user_name) { error("No RamdiskImg was returned from database"); @@ -842,6 +913,7 @@ != STATUS_OK) { error("bridge_get_data(RM_PartitionCnloadImg): %s", bg_err_str(rc)); + goto clean_up; } if(!user_name) { error("No CnloadImg was returned from database"); @@ -855,6 +927,7 @@ != STATUS_OK) { error("bridge_get_data(RM_PartitionIoloadImg): %s", bg_err_str(rc)); + goto clean_up; } if(!user_name) { error("No IoloadImg was returned from database"); @@ -869,6 +942,7 @@ != STATUS_OK) { error("bridge_get_data(RM_PartitionMloaderImg): %s", bg_err_str(rc)); + goto clean_up; } if(!user_name) { error("No MloaderImg was returned from database"); @@ -878,9 +952,8 @@ clean_up: - if (bg_recover - && ((rc = bridge_free_block(block_ptr)) - != STATUS_OK)) { + if (bg_recover + && ((rc = bridge_free_block(block_ptr)) != STATUS_OK)) { error("bridge_free_block(): %s", bg_err_str(rc)); } } @@ -889,7 +962,7 @@ return rc; } -#else +#endif extern int load_state_file(char *dir_name) { @@ -912,6 +985,10 @@ uid_t my_uid; int ionodes = 0; char *name = NULL; + struct part_record *part_ptr = NULL; + char *non_usable_nodes = NULL; + bitstr_t *bitmap = NULL; + ListIterator itr = NULL; if(!dir_name) { debug2("Starting bluegene with clean slate"); @@ -978,9 +1055,50 @@ error("select_p_state_restore: problem unpacking node_info"); goto unpack_error; } + +#ifdef HAVE_BG_FILES + for (i=0; irecord_count; i++) { + bg_info_record = &(node_select_ptr->bg_info_array[i]); + + /* we only care about the states we need here + * everthing else should have been set up already */ + if(bg_info_record->state == RM_PARTITION_ERROR) { + if((bg_record = find_bg_record_in_list( + bg_curr_block_list, + bg_info_record->bg_block_id))) + /* put_block_in_error_state should be + called after the bg_list has been + made. We can't call it here since + this record isn't the record kept + around in bg_list. + */ + bg_record->state = bg_info_record->state; + } + } + + select_g_free_node_info(&node_select_ptr); + free_buf(buffer); + return SLURM_SUCCESS; +#endif + slurm_mutex_lock(&block_state_mutex); reset_ba_system(false); + /* Locks are already in place to protect part_list here */ + bitmap = bit_alloc(node_record_count); + itr = list_iterator_create(part_list); + while ((part_ptr = list_next(itr))) { + /* we only want to use bps that are in partitions + */ + bit_or(bitmap, part_ptr->node_bitmap); + } + list_iterator_destroy(itr); + + bit_not(bitmap); + non_usable_nodes = bitmap2node_name(bitmap); + FREE_NULL_BITMAP(bitmap); + removable_set_bps(non_usable_nodes); + node_bitmap = bit_alloc(node_record_count); ionode_bitmap = bit_alloc(bluegene_numpsets); for (i=0; irecord_count; i++) { @@ -1027,24 +1145,24 @@ bg_record->ionodes = xstrdup(bg_info_record->ionodes); bg_record->ionode_bitmap = bit_copy(ionode_bitmap); + /* put_block_in_error_state should be + called after the bg_list has been + made. We can't call it here since + this record isn't the record kept + around in bg_list. + */ bg_record->state = bg_info_record->state; -#ifdef HAVE_BGL - bg_record->quarter = bg_info_record->quarter; - bg_record->nodecard = bg_info_record->nodecard; -#endif - if(bg_info_record->state == RM_PARTITION_ERROR) - bg_record->job_running = BLOCK_ERROR_STATE; - else - bg_record->job_running = NO_JOB_RUNNING; + bg_record->job_running = NO_JOB_RUNNING; + bg_record->bp_count = bit_size(node_bitmap); bg_record->node_cnt = bg_info_record->node_cnt; if(bluegene_bp_node_cnt > bg_record->node_cnt) { ionodes = bluegene_bp_node_cnt / bg_record->node_cnt; - bg_record->cpus_per_bp = - procs_per_node / ionodes; + bg_record->cpu_cnt = procs_per_node / ionodes; } else { - bg_record->cpus_per_bp = procs_per_node; + bg_record->cpu_cnt = procs_per_node + * bg_record->bp_count; } #ifdef HAVE_BGL bg_record->node_use = bg_info_record->node_use; @@ -1079,8 +1197,11 @@ for(j=0; jgeo[j]; - if(bluegene_layout_mode == LAYOUT_OVERLAP) + if(bluegene_layout_mode == LAYOUT_OVERLAP) { reset_ba_system(false); + removable_set_bps(non_usable_nodes); + } + results = list_create(NULL); name = set_bg_block(results, bg_record->start, @@ -1124,6 +1245,7 @@ } } + xfree(non_usable_nodes); FREE_NULL_BITMAP(ionode_bitmap); FREE_NULL_BITMAP(node_bitmap); @@ -1141,5 +1263,3 @@ free_buf(buffer); return SLURM_FAILURE; } - -#endif diff -Nru slurm-llnl-1.3.13/src/plugins/select/bluegene/plugin/bluegene.c slurm-llnl-1.3.15/src/plugins/select/bluegene/plugin/bluegene.c --- slurm-llnl-1.3.13/src/plugins/select/bluegene/plugin/bluegene.c 2009-01-06 19:20:48.000000000 +0100 +++ slurm-llnl-1.3.15/src/plugins/select/bluegene/plugin/bluegene.c 2009-04-09 18:56:23.000000000 +0200 @@ -1,7 +1,7 @@ /*****************************************************************************\ * bluegene.c - blue gene node configuration processing module. * - * $Id: bluegene.c 16146 2009-01-06 18:20:48Z da $ + * $Id: bluegene.c 17202 2009-04-09 16:56:23Z da $ ***************************************************************************** * Copyright (C) 2004 The Regents of the University of California. * Produced at Lawrence Livermore National Laboratory (cf, DISCLAIMER). @@ -40,7 +40,7 @@ #include "defined_block.h" #include -#define MMCS_POLL_TIME 120 /* poll MMCS for down switches and nodes +#define MMCS_POLL_TIME 30 /* poll MMCS for down switches and nodes * every 120 secs */ #define BG_POLL_TIME 0 /* poll bg blocks every 3 secs */ @@ -49,8 +49,6 @@ char* bg_conf = NULL; /* Global variables */ -my_bluegene_t *bg = NULL; - List bg_list = NULL; /* total list of bg_record entries */ List bg_curr_block_list = NULL; /* current bg blocks in bluegene.conf*/ List bg_job_block_list = NULL; /* jobs running in these blocks */ @@ -88,6 +86,7 @@ uint16_t bluegene_nodecard_node_cnt = 0; uint16_t bluegene_nodecard_ionode_cnt = 0; uint16_t bridge_api_verb = 0; + bool agent_fini = false; time_t last_bg_update; pthread_mutex_t block_state_mutex = PTHREAD_MUTEX_INITIALIZER; @@ -97,6 +96,8 @@ int num_unused_cpus = 0; pthread_mutex_t freed_cnt_mutex = PTHREAD_MUTEX_INITIALIZER; +static pthread_cond_t freed_cond = PTHREAD_COND_INITIALIZER; +static pthread_cond_t destroy_cond = PTHREAD_COND_INITIALIZER; List bg_free_block_list = NULL; /* blocks to be deleted */ List bg_destroy_block_list = NULL; /* blocks to be destroyed */ int free_cnt = 0; @@ -121,26 +122,6 @@ /* Initialize all plugin variables */ extern int init_bg(void) { -#ifdef HAVE_BG_FILES - int rc; - rm_size3D_t bp_size; - - info("Attempting to contact MMCS"); - if ((rc = bridge_get_bg(&bg)) != STATUS_OK) { - fatal("init_bg: rm_get_BG(): %s", bg_err_str(rc)); - return SLURM_ERROR; - } - - if ((rc = bridge_get_data(bg, RM_Msize, &bp_size)) != STATUS_OK) { - fatal("init_bg: rm_get_data(): %s", bg_err_str(rc)); - return SLURM_ERROR; - } - verbose("BlueGene configured with %d x %d x %d base blocks", - bp_size.X, bp_size.Y, bp_size.Z); - DIM_SIZE[X]=bp_size.X; - DIM_SIZE[Y]=bp_size.Y; - DIM_SIZE[Z]=bp_size.Z; -#endif ba_init(NULL); info("BlueGene plugin loaded successfully"); @@ -151,10 +132,18 @@ /* Purge all plugin variables */ extern void fini_bg(void) { -#ifdef HAVE_BG_FILES - int rc; -#endif - _set_bg_lists(); + if(!agent_fini) { + error("The agent hasn't been finied yet!"); + agent_fini = true; + } + /* wait for the agent threads to finish up */ + waitfor_block_agents(); + + /* wait for the destroy/free threads to finish up */ + if(free_cnt) + pthread_cond_wait(&freed_cond, &freed_cnt_mutex); + if(destroy_cnt) + pthread_cond_wait(&destroy_cond, &freed_cnt_mutex); if (bg_list) { list_destroy(bg_list); @@ -173,16 +162,7 @@ list_destroy(bg_booted_block_list); bg_booted_block_list = NULL; } - - /* wait for the free threads to finish up don't destroy the - * bg_free_block_list here */ - while(free_cnt > 0) - usleep(1000); - /* wait for the destroy threads to finish up don't destroy the - * bg_destroy_block_list here */ - while(destroy_cnt > 0) - usleep(1000); - + #ifdef HAVE_BGL if(bg_blrtsimage_list) { list_destroy(bg_blrtsimage_list); @@ -232,11 +212,6 @@ xfree(bg_slurm_user_name); xfree(bg_slurm_node_prefix); -#ifdef HAVE_BG_FILES - if(bg) - if ((rc = bridge_free_bg(bg)) != STATUS_OK) - error("bridge_free_BG(): %s", bg_err_str(rc)); -#endif ba_fini(); } @@ -258,29 +233,13 @@ if (!bit_overlap(rec_a->bitmap, rec_b->bitmap)) return false; -#ifdef HAVE_BGL - - if(rec_a->quarter != (uint16_t) NO_VAL) { - if(rec_b->quarter == (uint16_t) NO_VAL) - return true; - else if(rec_a->quarter != rec_b->quarter) - return false; - if(rec_a->nodecard != (uint16_t) NO_VAL) { - if(rec_b->nodecard == (uint16_t) NO_VAL) - return true; - else if(rec_a->nodecard - != rec_b->nodecard) - return false; - } - } -#else if((rec_a->node_cnt >= bluegene_bp_node_cnt) || (rec_b->node_cnt >= bluegene_bp_node_cnt)) return true; if (!bit_overlap(rec_a->ionode_bitmap, rec_b->ionode_bitmap)) return false; -#endif + return true; } @@ -408,6 +367,20 @@ return "SMALL"; case (SELECT_NAV): return "NAV"; +#ifndef HAVE_BGL + case SELECT_HTC_S: + return "HTC_S"; + break; + case SELECT_HTC_D: + return "HTC_D"; + break; + case SELECT_HTC_V: + return "HTC_V"; + break; + case SELECT_HTC_L: + return "HTC_L"; + break; +#endif default: break; } @@ -440,51 +413,69 @@ } /* - * bluegene_agent - detached thread periodically updates status of - * bluegene nodes. + * block_agent - thread periodically updates status of + * bluegene blocks. * - * NOTE: I don't grab any locks here because slurm_drain_nodes grabs - * the necessary locks. */ -extern void *bluegene_agent(void *args) +extern void *block_agent(void *args) { - static time_t last_mmcs_test; static time_t last_bg_test; int rc; + time_t now = time(NULL); - last_mmcs_test = time(NULL) + MMCS_POLL_TIME; - last_bg_test = time(NULL) + BG_POLL_TIME; + last_bg_test = now - BG_POLL_TIME; while (!agent_fini) { - time_t now = time(NULL); if (difftime(now, last_bg_test) >= BG_POLL_TIME) { if (agent_fini) /* don't bother */ - return NULL; /* quit now */ + break; /* quit now */ if(blocks_are_created) { last_bg_test = now; if((rc = update_block_list()) == 1) { - slurm_mutex_lock(&block_state_mutex); last_bg_update = now; - slurm_mutex_unlock(&block_state_mutex); } else if(rc == -1) error("Error with update_block_list"); if(bluegene_layout_mode == LAYOUT_DYNAMIC) { if((rc = update_freeing_block_list()) - == -1) + == 1) { + last_bg_update = now; + } else if(rc == -1) error("Error with " "update_block_list 2"); } } + now = time(NULL); } + + sleep(1); + } + return NULL; +} + +/* + * state_agent - thread periodically updates status of + * bluegene nodes. + * + */ +extern void *state_agent(void *args) +{ + static time_t last_mmcs_test; + time_t now = time(NULL); + last_mmcs_test = now - MMCS_POLL_TIME; + while (!agent_fini) { if (difftime(now, last_mmcs_test) >= MMCS_POLL_TIME) { if (agent_fini) /* don't bother */ - return NULL; /* quit now */ - last_mmcs_test = now; - test_mmcs_failures(); /* can run for a while */ - } + break; /* quit now */ + if(blocks_are_created) { + last_mmcs_test = now; + /* can run for a while */ + test_mmcs_failures(); + } + } sleep(1); + now = time(NULL); } return NULL; } @@ -502,7 +493,7 @@ //slurm_mutex_lock(&block_state_mutex); itr = list_iterator_create(my_bg_list); - while ((found_record = (bg_record_t *) list_next(itr)) != NULL) { + while ((found_record = list_next(itr))) { if(found_record) if(bg_record == found_record) { list_remove(itr); @@ -615,97 +606,26 @@ } if ((bg_record->state == RM_PARTITION_FREE) - || (bg_record->state == RM_PARTITION_ERROR)) { - break; - } - slurm_mutex_unlock(&block_state_mutex); - sleep(3); - } - remove_from_bg_list(bg_booted_block_list, bg_record); - slurm_mutex_unlock(&block_state_mutex); - - return SLURM_SUCCESS; -} - -#ifndef HAVE_BGL -/* This function not available in bgl land */ -extern int bg_reboot_block(bg_record_t *bg_record) -{ -#ifdef HAVE_BG_FILES - int rc; -#endif - if(!bg_record) { - error("bg_reboot_block: there was no bg_record"); - return SLURM_ERROR; - } - - while (1) { - if(!bg_record) { - error("bg_reboot_block: there was no bg_record"); - return SLURM_ERROR; - } - - slurm_mutex_lock(&block_state_mutex); - if (bg_record->state != NO_VAL - && bg_record->state != RM_PARTITION_REBOOTING) { -#ifdef HAVE_BG_FILES - debug2("bridge_reboot %s", bg_record->bg_block_id); - - rc = bridge_reboot_block(bg_record->bg_block_id); - if (rc != STATUS_OK) { - if(rc == PARTITION_NOT_FOUND) { - debug("block %s is not found", - bg_record->bg_block_id); - break; - } else if(rc == INCOMPATIBLE_STATE) { - debug2("bridge_reboot_partition" - "(%s): %s State = %d", - bg_record->bg_block_id, - bg_err_str(rc), - bg_record->state); - } else { - error("bridge_reboot_partition" - "(%s): %s State = %d", - bg_record->bg_block_id, - bg_err_str(rc), - bg_record->state); - } - } -#else - bg_record->state = RM_PARTITION_READY; - break; +#ifdef HAVE_BGL + || (bg_record->state == RM_PARTITION_ERROR) #endif - } - - if (bg_record->state == RM_PARTITION_CONFIGURING) { - if(!block_exist_in_list(bg_booted_block_list, - bg_record)) - list_push(bg_booted_block_list, bg_record); - break; - } else if (bg_record->state == RM_PARTITION_ERROR) { - remove_from_bg_list(bg_booted_block_list, bg_record); + ) { break; } slurm_mutex_unlock(&block_state_mutex); sleep(3); } + remove_from_bg_list(bg_booted_block_list, bg_record); slurm_mutex_unlock(&block_state_mutex); return SLURM_SUCCESS; } -#endif /* Free multiple blocks in parallel */ extern void *mult_free_block(void *args) { bg_record_t *bg_record = NULL; - - slurm_mutex_lock(&freed_cnt_mutex); - if ((bg_freeing_list == NULL) - && ((bg_freeing_list = list_create(destroy_bg_record)) == NULL)) - fatal("malloc failure in bg_freeing_list"); - slurm_mutex_unlock(&freed_cnt_mutex); - + /* * Don't just exit when there is no work left. Creating * pthreads from within a dynamically linked object (plugin) @@ -737,13 +657,10 @@ } slurm_mutex_lock(&freed_cnt_mutex); free_cnt--; - if(bg_freeing_list) { - list_destroy(bg_freeing_list); - bg_freeing_list = NULL; - } if(free_cnt == 0) { list_destroy(bg_free_block_list); bg_free_block_list = NULL; + pthread_cond_signal(&freed_cond); } slurm_mutex_unlock(&freed_cnt_mutex); return NULL; @@ -788,8 +705,7 @@ sort_bg_record_inc_size(bg_freeing_list); if(remove_from_bg_list(bg_job_block_list, bg_record) == SLURM_SUCCESS) { - num_unused_cpus += - bg_record->bp_count*bg_record->cpus_per_bp; + num_unused_cpus += bg_record->cpu_cnt; } slurm_mutex_unlock(&block_state_mutex); debug3("removing the jobs on block %s\n", @@ -824,7 +740,10 @@ debug2("done %s", (char *)bg_record->bg_block_id); #endif + slurm_mutex_lock(&block_state_mutex); destroy_bg_record(bg_record); + slurm_mutex_unlock(&block_state_mutex); + last_bg_update = time(NULL); debug2("destroyed"); already_here: @@ -835,13 +754,14 @@ } slurm_mutex_lock(&freed_cnt_mutex); destroy_cnt--; - if(bg_freeing_list) { - list_destroy(bg_freeing_list); - bg_freeing_list = NULL; - } if(destroy_cnt == 0) { + if(bg_freeing_list) { + list_destroy(bg_freeing_list); + bg_freeing_list = NULL; + } list_destroy(bg_destroy_block_list); bg_destroy_block_list = NULL; + pthread_cond_signal(&destroy_cond); } slurm_mutex_unlock(&freed_cnt_mutex); @@ -878,9 +798,15 @@ while ((found_record = (bg_record_t*)list_pop(delete_list)) != NULL) { /* push job onto queue in a FIFO */ debug3("adding %s to be freed", found_record->bg_block_id); - if (list_push(*block_list, found_record) == NULL) - fatal("malloc failure in _block_op/list_push"); - + if(!block_ptr_exist_in_list(*block_list, found_record)) { + if (list_push(*block_list, found_record) == NULL) + fatal("malloc failure in _block_op/list_push"); + } else { + error("we had block %s already on the freeing list", + found_record->bg_block_id); + num_block_to_free--; + continue; + } /* already running MAX_AGENTS we don't really need more since they don't end until we shut down the controller */ if (*count > MAX_AGENT_COUNT) @@ -1198,9 +1124,16 @@ bitstr_t *tmp_bitmap = NULL; int small_size = 1; - bluegene_quarter_ionode_cnt = bluegene_numpsets/4; - bluegene_nodecard_ionode_cnt = bluegene_quarter_ionode_cnt/4; - + /* THIS IS A HACK TO MAKE A 1 NODECARD SYSTEM WORK */ + if(bluegene_bp_node_cnt == bluegene_nodecard_node_cnt) { + bluegene_quarter_ionode_cnt = 2; + bluegene_nodecard_ionode_cnt = 2; + } else { + bluegene_quarter_ionode_cnt = bluegene_numpsets/4; + bluegene_nodecard_ionode_cnt = + bluegene_quarter_ionode_cnt/4; + } + /* How many nodecards per ionode */ bluegene_nc_ratio = ((double)bluegene_bp_node_cnt @@ -1255,6 +1188,11 @@ list_append(bg_valid_small32, tmp_bitmap); } } + /* If we only have 1 nodecard just jump to the end + since this will never need to happen below. + Pretty much a hack to avoid seg fault;). */ + if(bluegene_bp_node_cnt == bluegene_nodecard_node_cnt) + goto no_calc; bg_valid_small128 = list_create(_destroy_bitmap); if((small_size = bluegene_quarter_ionode_cnt)) @@ -1294,6 +1232,8 @@ fatal("your numpsets is 0"); } +no_calc: + if (!s_p_get_uint16(&bridge_api_verb, "BridgeAPIVerbose", tbl)) info("Warning: BridgeAPIVerbose not configured " "in bluegene.conf"); @@ -1301,6 +1241,20 @@ info("BridgeAPILogFile not configured in bluegene.conf"); else _reopen_bridge_log(); + + if (s_p_get_string(&layout, "DenyPassthrough", tbl)) { + if(strstr(layout, "X")) + ba_deny_pass |= PASS_DENY_X; + if(strstr(layout, "Y")) + ba_deny_pass |= PASS_DENY_Y; + if(strstr(layout, "Z")) + ba_deny_pass |= PASS_DENY_Z; + if(!strcasecmp(layout, "ALL")) + ba_deny_pass |= PASS_DENY_ALL; + + xfree(layout); + } + if (!s_p_get_string(&layout, "LayoutMode", tbl)) { info("Warning: LayoutMode was not specified in bluegene.conf " "defaulting to STATIC partitioning"); @@ -1329,7 +1283,7 @@ } for (i = 0; i < count; i++) { - add_bg_record(bg_list, NULL, blockreq_array[i]); + add_bg_record(bg_list, NULL, blockreq_array[i], 0, 0); } } s_p_hashtbl_destroy(tbl); @@ -1340,14 +1294,17 @@ extern int validate_current_blocks(char *dir) { /* found bg blocks already on system */ - List bg_found_block_list = list_create(NULL); + List bg_found_block_list = NULL; static time_t last_config_update = (time_t) 0; + ListIterator itr = NULL; + bg_record_t *bg_record = NULL; /* only run on startup */ if(last_config_update) return SLURM_SUCCESS; last_config_update = time(NULL); + bg_found_block_list = list_create(NULL); //#if 0 /* Check to see if the configs we have are correct */ if (_validate_config_nodes(&bg_found_block_list, dir) == SLURM_ERROR) { @@ -1371,6 +1328,17 @@ } } + /* ok now since bg_list has been made we now can put blocks in + an error state this needs to be done outside of a lock + it doesn't matter much in the first place though since + no threads are started before this function. */ + itr = list_iterator_create(bg_list); + while((bg_record = list_next(itr))) { + if(bg_record->state == RM_PARTITION_ERROR) + put_block_in_error_state(bg_record, BLOCK_ERROR_STATE); + } + list_iterator_destroy(itr); + slurm_mutex_lock(&block_state_mutex); list_destroy(bg_curr_block_list); bg_curr_block_list = NULL; @@ -1378,6 +1346,7 @@ list_destroy(bg_found_block_list); bg_found_block_list = NULL; } + last_bg_update = time(NULL); blocks_are_created = 1; sort_bg_record_inc_size(bg_list); @@ -1421,8 +1390,7 @@ bg_mloaderimage_list = list_create(destroy_image); if(bg_ramdiskimage_list) list_destroy(bg_ramdiskimage_list); - bg_ramdiskimage_list = list_create(destroy_image); - + bg_ramdiskimage_list = list_create(destroy_image); } /* @@ -1439,7 +1407,6 @@ int rc = SLURM_ERROR; bg_record_t* bg_record = NULL; bg_record_t* init_bg_record = NULL; - bg_record_t* full_system_bg_record = NULL; int full_created = 0; ListIterator itr_conf; ListIterator itr_curr; @@ -1450,8 +1417,12 @@ * happens in the state load before this in emulation mode */ if (read_bg_blocks() == SLURM_ERROR) return SLURM_ERROR; + /* since we only care about error states here we don't care + about the return code this must be done after the bg_list + is created */ + load_state_file(dir); #else - /* read in state from last run. Only for emulation mode */ + /* read in state from last run. */ if ((rc = load_state_file(dir)) != SLURM_SUCCESS) return rc; /* This needs to be reset to SLURM_ERROR or it will never we @@ -1463,44 +1434,32 @@ if(!bg_curr_block_list) return SLURM_ERROR; - - itr_curr = list_iterator_create(bg_curr_block_list); - while ((init_bg_record = list_next(itr_curr))) - if(init_bg_record->full_block) - full_system_bg_record = init_bg_record; if(!*bg_found_block_list) (*bg_found_block_list) = list_create(NULL); - + + itr_curr = list_iterator_create(bg_curr_block_list); itr_conf = list_iterator_create(bg_list); while ((bg_record = (bg_record_t*) list_next(itr_conf))) { - /* translate hostlist to ranged - string for consistent format - search here - */ list_iterator_reset(itr_curr); - while ((init_bg_record = list_next(itr_curr))) { - if (strcasecmp(bg_record->nodes, - init_bg_record->nodes)) + while ((init_bg_record = list_next(itr_curr))) { + if (strcasecmp(bg_record->nodes, init_bg_record->nodes)) continue; /* wrong nodes */ - if (bg_record->conn_type - != init_bg_record->conn_type) - continue; /* wrong conn_type */ + if(!bit_equal(bg_record->ionode_bitmap, + init_bg_record->ionode_bitmap)) + continue; #ifdef HAVE_BGL - if(bg_record->quarter != - init_bg_record->quarter) - continue; /* wrong quart */ - if(bg_record->nodecard != - init_bg_record->nodecard) - continue; /* wrong nodecard */ + if (bg_record->conn_type != init_bg_record->conn_type) + continue; /* wrong conn_type */ if(bg_record->blrtsimage && strcasecmp(bg_record->blrtsimage, init_bg_record->blrtsimage)) continue; #else - if(!bit_equal(bg_record->ionode_bitmap, - init_bg_record->ionode_bitmap)) - continue; + if ((bg_record->conn_type != init_bg_record->conn_type) + && ((bg_record->conn_type < SELECT_SMALL) + && (init_bg_record->conn_type < SELECT_SMALL))) + continue; /* wrong conn_type */ #endif if(bg_record->linuximage && strcasecmp(bg_record->linuximage, @@ -1516,6 +1475,10 @@ continue; copy_bg_record(init_bg_record, bg_record); + /* remove from the curr list since we just + matched it no reason to keep it around + anymore */ + list_delete_item(itr_curr); break; } @@ -1539,37 +1502,45 @@ convert_conn_type(bg_record->conn_type)); if(((bg_record->state == RM_PARTITION_READY) || (bg_record->state == RM_PARTITION_CONFIGURING)) - && !block_exist_in_list(bg_booted_block_list, - bg_record)) + && !block_ptr_exist_in_list(bg_booted_block_list, + bg_record)) list_push(bg_booted_block_list, bg_record); } } - list_iterator_destroy(itr_conf); - list_iterator_destroy(itr_curr); if(bluegene_layout_mode == LAYOUT_DYNAMIC) goto finished; - if(!full_created && full_system_bg_record) { - bg_record = xmalloc(sizeof(bg_record_t)); - copy_bg_record(full_system_bg_record, bg_record); - list_append(bg_list, bg_record); - list_push(*bg_found_block_list, bg_record); - format_node_name(bg_record, tmp_char, sizeof(tmp_char)); - info("Existing: BlockID:%s Nodes:%s Conn:%s", - bg_record->bg_block_id, - tmp_char, - convert_conn_type(bg_record->conn_type)); - if(((bg_record->state == RM_PARTITION_READY) - || (bg_record->state == RM_PARTITION_CONFIGURING)) - && !block_exist_in_list(bg_booted_block_list, - bg_record)) - list_push(bg_booted_block_list, bg_record); + if(!full_created) { + list_iterator_reset(itr_curr); + while ((init_bg_record = list_next(itr_curr))) { + if(init_bg_record->full_block) { + list_remove(itr_curr); + bg_record = init_bg_record; + list_append(bg_list, bg_record); + list_push(*bg_found_block_list, bg_record); + format_node_name(bg_record, tmp_char, + sizeof(tmp_char)); + info("Existing: BlockID:%s Nodes:%s Conn:%s", + bg_record->bg_block_id, + tmp_char, + convert_conn_type(bg_record->conn_type)); + if(((bg_record->state == RM_PARTITION_READY) + || (bg_record->state + == RM_PARTITION_CONFIGURING)) + && !block_ptr_exist_in_list( + bg_booted_block_list, bg_record)) + list_push(bg_booted_block_list, + bg_record); + break; + } + } } finished: - if(list_count(bg_list) == list_count(bg_curr_block_list)) + list_iterator_destroy(itr_conf); + list_iterator_destroy(itr_curr); + if(!list_count(bg_curr_block_list)) rc = SLURM_SUCCESS; - return rc; } @@ -1681,6 +1652,8 @@ retries=30; while(num_block_to_free > num_block_freed) { + /* no need to check for return code here, things + haven't started up yet. */ update_freeing_block_list(); if(retries==30) { info("Waiting for old blocks to be " diff -Nru slurm-llnl-1.3.13/src/plugins/select/bluegene/plugin/bluegene.h slurm-llnl-1.3.15/src/plugins/select/bluegene/plugin/bluegene.h --- slurm-llnl-1.3.13/src/plugins/select/bluegene/plugin/bluegene.h 2009-01-06 19:20:48.000000000 +0100 +++ slurm-llnl-1.3.15/src/plugins/select/bluegene/plugin/bluegene.h 2009-04-01 01:23:01.000000000 +0200 @@ -1,7 +1,7 @@ /*****************************************************************************\ * bluegene.h - header for blue gene configuration processing module. * - * $Id: bluegene.h 16146 2009-01-06 18:20:48Z da $ + * $Id: bluegene.h 17102 2009-03-31 23:23:01Z da $ ***************************************************************************** * Copyright (C) 2004 The Regents of the University of California. * Produced at Lawrence Livermore National Laboratory (cf, DISCLAIMER). @@ -49,10 +49,7 @@ LAYOUT_DYNAMIC /* slurm will make all blocks */ } bg_layout_t; - /* Global variables */ - -extern my_bluegene_t *bg; #ifdef HAVE_BGL extern char *default_blrtsimage; #endif @@ -74,6 +71,7 @@ extern uint16_t bluegene_nodecard_ionode_cnt; extern uint16_t bluegene_quarter_node_cnt; extern uint16_t bluegene_quarter_ionode_cnt; + extern ba_system_t *ba_system_ptr; extern time_t last_bg_update; @@ -104,6 +102,7 @@ #define MAX_PTHREAD_RETRIES 1 #define BLOCK_ERROR_STATE -3 +#define ADMIN_ERROR_STATE -4 #define NO_JOB_RUNNING -1 #define MAX_AGENT_COUNT 30 #define BUFSIZE 4096 @@ -145,16 +144,16 @@ /* sort a list of bg_records by size (node count) */ extern void sort_bg_record_inc_size(List records); -/* bluegene_agent - detached thread periodically tests status of bluegene - * nodes and switches */ -extern void *bluegene_agent(void *args); +/* block_agent - detached thread periodically tests status of bluegene + * blocks */ +extern void *block_agent(void *args); + +/* state_agent - thread periodically tests status of bluegene + * nodes, nodecards, and switches */ +extern void *state_agent(void *args); extern int bg_free_block(bg_record_t *bg_record); -#ifndef HAVE_BGL -extern int bg_reboot_block(bg_record_t *bg_record); -#endif - extern int remove_from_bg_list(List my_bg_list, bg_record_t *bg_record); extern bg_record_t *find_and_remove_org_from_bg_list(List my_list, bg_record_t *bg_record); diff -Nru slurm-llnl-1.3.13/src/plugins/select/bluegene/plugin/defined_block.c slurm-llnl-1.3.15/src/plugins/select/bluegene/plugin/defined_block.c --- slurm-llnl-1.3.13/src/plugins/select/bluegene/plugin/defined_block.c 2008-11-18 00:20:37.000000000 +0100 +++ slurm-llnl-1.3.15/src/plugins/select/bluegene/plugin/defined_block.c 2009-04-01 20:10:50.000000000 +0200 @@ -59,10 +59,29 @@ int geo[BA_SYSTEM_DIMENSIONS]; char temp[256]; List results = NULL; - + struct part_record *part_ptr = NULL; + char *non_usable_nodes = NULL; + bitstr_t *bitmap = bit_alloc(node_record_count); + #ifdef HAVE_BG_FILES init_wires(); #endif + + /* Locks are already in place to protect part_list here */ + itr = list_iterator_create(part_list); + while ((part_ptr = list_next(itr))) { + /* we only want to use bps that are in + * partitions + */ + bit_or(bitmap, part_ptr->node_bitmap); + } + list_iterator_destroy(itr); + + bit_not(bitmap); + non_usable_nodes = bitmap2node_name(bitmap); + removable_set_bps(non_usable_nodes); + FREE_NULL_BITMAP(bitmap); + slurm_mutex_lock(&block_state_mutex); reset_ba_system(false); if(bg_list) { @@ -73,27 +92,18 @@ bg_found_block_list); while ((found_record = (bg_record_t*) list_next(itr_found)) != NULL) { -/* info("%s.%d.%d ?= %s.%d.%d\n", */ +/* info("%s[%s[ ?= %s[%s]\n", */ /* bg_record->nodes, */ -/* bg_record->quarter, */ -/* bg_record->nodecard, */ +/* bg_record->ionodes, */ /* found_record->nodes, */ -/* found_record->quarter, */ -/* found_record->nodecard); */ +/* found_record->ionodes); */ if ((bit_equal(bg_record->bitmap, found_record->bitmap)) -#ifdef HAVE_BGL - && (bg_record->quarter == - found_record->quarter) - && (bg_record->nodecard == - found_record->nodecard) -#else && (bit_equal(bg_record-> ionode_bitmap, found_record-> ionode_bitmap)) -#endif ) { /* don't reboot this one */ break; @@ -104,14 +114,16 @@ error("create_defined_blocks: " "no bg_found_block_list 1"); } - if(bg_record->bp_count>0 + if(bg_record->bp_count > 0 && !bg_record->full_block - && bg_record->cpus_per_bp == procs_per_node) { + && bg_record->cpu_cnt >= procs_per_node) { char *name = NULL; - if(overlapped == LAYOUT_OVERLAP) + if(overlapped == LAYOUT_OVERLAP) { reset_ba_system(false); - + removable_set_bps(non_usable_nodes); + } + /* we want the bps that aren't * in this record to mark them as used */ @@ -149,6 +161,7 @@ reset_all_removed_bps(); slurm_mutex_unlock( &block_state_mutex); + xfree(non_usable_nodes); return SLURM_ERROR; } } else { @@ -167,6 +180,7 @@ list_iterator_destroy(itr); slurm_mutex_unlock( &block_state_mutex); + xfree(non_usable_nodes); return SLURM_ERROR; } @@ -215,6 +229,7 @@ == SLURM_ERROR) { list_iterator_destroy(itr); slurm_mutex_unlock(&block_state_mutex); + xfree(non_usable_nodes); return rc; } print_bg_record(bg_record); @@ -224,8 +239,11 @@ } else { error("create_defined_blocks: no bg_list 2"); slurm_mutex_unlock(&block_state_mutex); + xfree(non_usable_nodes); return SLURM_ERROR; } + xfree(non_usable_nodes); + slurm_mutex_unlock(&block_state_mutex); create_full_system_block(bg_found_block_list); @@ -263,7 +281,28 @@ int i; blockreq_t blockreq; List results = NULL; + struct part_record *part_ptr = NULL; + bitstr_t *bitmap = bit_alloc(node_record_count); + /* Locks are already in place to protect part_list here */ + itr = list_iterator_create(part_list); + while ((part_ptr = list_next(itr))) { + /* we only want to use bps that are in + * partitions + */ + bit_or(bitmap, part_ptr->node_bitmap); + } + list_iterator_destroy(itr); + + bit_not(bitmap); + if(bit_ffs(bitmap) != -1) { + error("We don't have the entire system covered by partitions, " + "can't create full system block"); + FREE_NULL_BITMAP(bitmap); + return SLURM_ERROR; + } + FREE_NULL_BITMAP(bitmap); + /* Here we are adding a block that in for the entire machine just in case it isn't in the bluegene.conf file. */ @@ -331,7 +370,7 @@ blockreq.block = name; blockreq.conn_type = SELECT_TORUS; - add_bg_record(records, NULL, &blockreq); + add_bg_record(records, NULL, &blockreq, 0 , 0); xfree(name); bg_record = (bg_record_t *) list_pop(records); @@ -357,8 +396,7 @@ geo, bg_record->conn_type); if(!name) { - error("I was unable to make the " - "requested block."); + error("I was unable to make the full system block."); list_destroy(results); list_iterator_destroy(itr); slurm_mutex_unlock(&block_state_mutex); diff -Nru slurm-llnl-1.3.13/src/plugins/select/bluegene/plugin/dynamic_block.c slurm-llnl-1.3.15/src/plugins/select/bluegene/plugin/dynamic_block.c --- slurm-llnl-1.3.13/src/plugins/select/bluegene/plugin/dynamic_block.c 2008-11-21 21:26:13.000000000 +0100 +++ slurm-llnl-1.3.15/src/plugins/select/bluegene/plugin/dynamic_block.c 2009-04-10 21:25:52.000000000 +0200 @@ -38,13 +38,9 @@ #include "dynamic_block.h" -#ifdef HAVE_BGL -static int _split_block(List block_list, List new_blocks, - bg_record_t *bg_record, int procs); -#else static int _split_block(List block_list, List new_blocks, bg_record_t *bg_record, int cnodes); -#endif + static int _breakup_blocks(List block_list, List new_blocks, ba_request_t *request, List my_block_list); @@ -54,7 +50,8 @@ * RET - a list of created block(s) or NULL on failure errno is set. */ extern List create_dynamic_block(List block_list, - ba_request_t *request, List my_block_list) + ba_request_t *request, List my_block_list, + bool track_down_nodes) { int rc = SLURM_SUCCESS; @@ -79,7 +76,7 @@ slurm_mutex_lock(&block_state_mutex); if(my_block_list) { - reset_ba_system(true); + reset_ba_system(track_down_nodes); itr = list_iterator_create(my_block_list); while ((bg_record = list_next(itr))) { if(!my_bitmap) { @@ -137,44 +134,18 @@ xfree(nodes); FREE_NULL_BITMAP(bitmap); } -#ifdef HAVE_BGL - if(request->size==1 && cnodes < bluegene_bp_node_cnt) { - request->conn_type = SELECT_SMALL; - if(request->procs == (procs_per_node/16)) { - if(!bluegene_nodecard_ionode_cnt) { - error("can't create this size %d " - "on this system numpsets is %d", - request->procs, - bluegene_numpsets); - goto finished; - } - blockreq.small32=4; - blockreq.small128=3; - } else { - if(!bluegene_quarter_ionode_cnt) { - error("can't create this size %d " - "on this system numpsets is %d", - request->procs, - bluegene_numpsets); - goto finished; - } - blockreq.small128=4; - } - new_blocks = list_create(destroy_bg_record); - if(_breakup_blocks(block_list, new_blocks, - request, my_block_list) - != SLURM_SUCCESS) { - list_destroy(new_blocks); - new_blocks = NULL; - debug2("small block not able to be placed"); - //rc = SLURM_ERROR; - } else - goto finished; - } -#else if(request->size==1 && cnodes < bluegene_bp_node_cnt) { switch(cnodes) { +#ifdef HAVE_BGL + case 32: + blockreq.small32 = 4; + blockreq.small128 = 3; + break; + case 128: + blockreq.small128 = 4; + break; +#else case 16: blockreq.small16 = 2; blockreq.small32 = 1; @@ -200,6 +171,7 @@ case 256: blockreq.small256 = 2; break; +#endif default: error("This size %d is unknown on this system", cnodes); goto finished; @@ -218,16 +190,23 @@ } else goto finished; } -#endif + if(request->conn_type == SELECT_NAV) request->conn_type = SELECT_TORUS; + //debug("going to create %d", request->size); if(!new_ba_request(request)) { - error("Problems with request for size %d geo %dx%dx%d", - request->size, - request->geometry[X], - request->geometry[Y], - request->geometry[Z]); + if(geo[X] == (uint16_t)NO_VAL) { + error("Problems with request for size %d geo %dx%dx%d", + request->size, + request->geometry[X], + request->geometry[Y], + request->geometry[Z]); + } else { + error("Problems with request for size %d. " + "No geo given.", + request->size); + } rc = ESLURM_INTERCONNECT_FAILURE; goto finished; } @@ -251,16 +230,8 @@ set in the ionode_bitmap. */ if(bg_record->job_running == NO_JOB_RUNNING -#ifdef HAVE_BGL - && (bg_record->quarter == (uint16_t) NO_VAL - || (bg_record->quarter == 0 - && (bg_record->nodecard == (uint16_t) NO_VAL - || bg_record->nodecard == 0))) -#else && ((bg_record->node_cnt >= bluegene_bp_node_cnt) - || (bit_ffs(bg_record->ionode_bitmap) == 0)) -#endif - ) { + || (bit_ffs(bg_record->ionode_bitmap) == 0))) { for(i=0; istart[i] = @@ -325,7 +296,7 @@ blockreq.ramdiskimage = request->ramdiskimage; blockreq.conn_type = request->conn_type; - add_bg_record(new_blocks, results, &blockreq); + add_bg_record(new_blocks, results, &blockreq, 0, 0); finished: reset_all_removed_bps(); @@ -344,91 +315,14 @@ return new_blocks; } -#ifdef HAVE_BGL -extern bg_record_t *create_small_record(bg_record_t *bg_record, - uint16_t quarter, uint16_t nodecard) -{ - bg_record_t *found_record = NULL; - int small_size = 4; - ba_node_t *new_ba_node = NULL; - ba_node_t *ba_node = NULL; - found_record = (bg_record_t*) xmalloc(sizeof(bg_record_t)); - - found_record->job_running = NO_JOB_RUNNING; - found_record->user_name = xstrdup(bg_record->user_name); - found_record->user_uid = bg_record->user_uid; - found_record->bg_block_list = list_create(destroy_ba_node); - ba_node = list_peek(bg_record->bg_block_list); - if(!ba_node) { - hostlist_t hl = hostlist_create(bg_record->nodes); - char *host = hostlist_shift(hl); - hostlist_destroy(hl); - found_record->nodes = xstrdup(host); - free(host); - error("you gave me a list with no ba_nodes using %s", - found_record->nodes); - } else { - int i=0,j=0; - new_ba_node = ba_copy_node(ba_node); - for (i=0; iaxis_switch[i].int_wire[j].used = 0; - if(i!=X) { - if(j==3 || j==4) - ba_node->axis_switch[i]. - int_wire[j]. - used = 1; - } - ba_node->axis_switch[i].int_wire[j]. - port_tar = j; - } - } - list_append(found_record->bg_block_list, new_ba_node); - found_record->bp_count = 1; - found_record->nodes = xstrdup_printf( - "%s%c%c%c", - bg_slurm_node_prefix, - alpha_num[ba_node->coord[X]], - alpha_num[ba_node->coord[Y]], - alpha_num[ba_node->coord[Z]]); - } - - found_record->blrtsimage = xstrdup(bg_record->blrtsimage); - found_record->linuximage = xstrdup(bg_record->linuximage); - found_record->mloaderimage = xstrdup(bg_record->mloaderimage); - found_record->ramdiskimage = xstrdup(bg_record->ramdiskimage); - - process_nodes(found_record, false); - - found_record->conn_type = SELECT_SMALL; - - found_record->node_use = SELECT_COPROCESSOR_MODE; - - if(nodecard != (uint16_t) NO_VAL) - small_size = bluegene_bp_nodecard_cnt; - found_record->cpus_per_bp = procs_per_node/small_size; - found_record->node_cnt = bluegene_bp_node_cnt/small_size; - found_record->quarter = quarter; - found_record->nodecard = nodecard; - - if(set_ionodes(found_record) == SLURM_ERROR) - error("couldn't create ionode_bitmap for %d.%d", - found_record->quarter, found_record->nodecard); - return found_record; -} - -#else extern bg_record_t *create_small_record(bg_record_t *bg_record, bitstr_t *ionodes, int size) { bg_record_t *found_record = NULL; ba_node_t *new_ba_node = NULL; ba_node_t *ba_node = NULL; -#ifdef HAVE_BGL - int small_size = 4; -#else char bitstring[BITSIZE]; -#endif + found_record = (bg_record_t*) xmalloc(sizeof(bg_record_t)); found_record->job_running = NO_JOB_RUNNING; @@ -449,7 +343,7 @@ new_ba_node = ba_copy_node(ba_node); for (i=0; iaxis_switch[i].int_wire[j].used = 0; + ba_node->axis_switch[i].int_wire[j].used = 0; if(i!=X) { if(j==3 || j==4) ba_node->axis_switch[i]. @@ -470,6 +364,7 @@ alpha_num[ba_node->coord[Z]]); } #ifdef HAVE_BGL + found_record->node_use = SELECT_COPROCESSOR_MODE; found_record->blrtsimage = xstrdup(bg_record->blrtsimage); #endif found_record->linuximage = xstrdup(bg_record->linuximage); @@ -480,348 +375,64 @@ found_record->conn_type = SELECT_SMALL; -#ifdef HAVE_BGL - found_record->node_use = SELECT_COPROCESSOR_MODE; - if(nodecard != (uint16_t) NO_VAL) - small_size = bluegene_bp_nodecard_cnt; - found_record->cpus_per_bp = procs_per_node/small_size; - found_record->node_cnt = bluegene_bp_node_cnt/small_size; - found_record->quarter = quarter; - found_record->nodecard = nodecard; - - if(set_ionodes(found_record) == SLURM_ERROR) - error("couldn't create ionode_bitmap for %d.%d", - found_record->quarter, found_record->nodecard); -#else xassert(bluegene_proc_ratio); - found_record->cpus_per_bp = bluegene_proc_ratio * size; + found_record->cpu_cnt = bluegene_proc_ratio * size; found_record->node_cnt = size; found_record->ionode_bitmap = bit_copy(ionodes); bit_fmt(bitstring, BITSIZE, found_record->ionode_bitmap); found_record->ionodes = xstrdup(bitstring); -#endif + return found_record; } -#endif /*********************** Local Functions *************************/ -#ifdef HAVE_BGL static int _split_block(List block_list, List new_blocks, - bg_record_t *bg_record, int procs) + bg_record_t *bg_record, int cnodes) { - bg_record_t *found_record = NULL; bool full_bp = false; - int small_count = 0; - int small_size = 0; - uint16_t num_nodecard = 0, num_quarter = 0; - int i; - int node_cnt = 0; - uint16_t quarter = 0; - uint16_t nodecard = 0; - - if(bg_record->quarter == (uint16_t) NO_VAL) - full_bp = true; - - if(procs == (procs_per_node/bluegene_bp_nodecard_cnt) - && bluegene_nodecard_ionode_cnt) { - num_nodecard=4; - if(full_bp) - num_quarter=3; - } else if(full_bp) { - num_quarter = 4; - } else { - error("you asked for something that was already this size"); - return SLURM_ERROR; - } - debug2("asking for %d 32s from a %d block", - num_nodecard, bg_record->node_cnt); - small_count = num_nodecard+num_quarter; - - /* break base partition up into 16 parts */ - small_size = bluegene_bp_node_cnt/bluegene_nodecard_node_cnt; - node_cnt = 0; - if(!full_bp) - quarter = bg_record->quarter; - else - quarter = 0; - nodecard = 0; - for(i=0; iprocs, request->size); + bitoff_t start = 0; + blockreq_t blockreq; - itr = list_iterator_create(block_list); - while ((bg_record = (bg_record_t *) list_next(itr)) != NULL) { - if(bg_record->job_running != NO_JOB_RUNNING) - continue; - if(bg_record->state != RM_PARTITION_FREE) - continue; - if (request->avail_node_bitmap && - !bit_super_set(bg_record->bitmap, - request->avail_node_bitmap)) { - debug2("bg block %s has nodes not usable by this job", - bg_record->bg_block_id); - continue; - } + memset(&blockreq, 0, sizeof(blockreq_t)); - if(request->start_req) { - if ((request->start[X] != bg_record->start[X]) - || (request->start[Y] != bg_record->start[Y]) - || (request->start[Z] != bg_record->start[Z])) { - debug4("small got %c%c%c looking for %c%c%c", - alpha_num[bg_record->start[X]], - alpha_num[bg_record->start[Y]], - alpha_num[bg_record->start[Z]], - alpha_num[request->start[X]], - alpha_num[request->start[Y]], - alpha_num[request->start[Z]]); - continue; - } - debug3("small found %c%c%c looking for %c%c%c", - alpha_num[bg_record->start[X]], - alpha_num[bg_record->start[Y]], - alpha_num[bg_record->start[Z]], - alpha_num[request->start[X]], - alpha_num[request->start[Y]], - alpha_num[request->start[Z]]); - } - proc_cnt = bg_record->bp_count * - bg_record->cpus_per_bp; - if(proc_cnt == request->procs) { - debug2("found it here %s, %s", - bg_record->bg_block_id, - bg_record->nodes); - request->save_name = xstrdup_printf( - "%c%c%c", - alpha_num[bg_record->start[X]], - alpha_num[bg_record->start[Y]], - alpha_num[bg_record->start[Z]]); - rc = SLURM_SUCCESS; - goto finished; - } - if(bg_record->node_cnt > bluegene_bp_node_cnt) - continue; - if(proc_cnt < request->procs) { - if(last_quarter != bg_record->quarter){ - last_quarter = bg_record->quarter; - total_proc_cnt = proc_cnt; - } else { - total_proc_cnt += proc_cnt; - } - debug2("1 got %d on quarter %d", - total_proc_cnt, last_quarter); - if(total_proc_cnt == request->procs) { - request->save_name = xstrdup_printf( - "%c%c%c", - alpha_num[bg_record->start[X]], - alpha_num[bg_record->start[Y]], - alpha_num[bg_record->start[Z]]); - if(!my_block_list) { - rc = SLURM_SUCCESS; - goto finished; - } - - bg_record = create_small_record( - bg_record, - last_quarter, - (uint16_t) NO_VAL); - list_append(new_blocks, bg_record); - - rc = SLURM_SUCCESS; - goto finished; - } - continue; - } + switch(bg_record->node_cnt) { +#ifdef HAVE_BGL + case 32: + error("We got a 32 we should never have this"); + goto finished; break; - } - if(bg_record) { - debug2("got one on the first pass"); - goto found_one; - } - list_iterator_reset(itr); - last_quarter = (uint16_t) NO_VAL; - while ((bg_record = (bg_record_t *) list_next(itr)) - != NULL) { - if(bg_record->job_running != NO_JOB_RUNNING) - continue; - if (request->avail_node_bitmap && - !bit_super_set(bg_record->bitmap, - request->avail_node_bitmap)) { - debug2("bg block %s has nodes not usable by this job", - bg_record->bg_block_id); - continue; - } - - if(request->start_req) { - if ((request->start[X] != bg_record->start[X]) - || (request->start[Y] != bg_record->start[Y]) - || (request->start[Z] != bg_record->start[Z])) { - debug4("small 2 got %c%c%c looking for %c%c%c", - alpha_num[bg_record->start[X]], - alpha_num[bg_record->start[Y]], - alpha_num[bg_record->start[Z]], - alpha_num[request->start[X]], - alpha_num[request->start[Y]], - alpha_num[request->start[Z]]); - continue; - } - debug3("small 2 found %c%c%c looking for %c%c%c", - alpha_num[bg_record->start[X]], - alpha_num[bg_record->start[Y]], - alpha_num[bg_record->start[Z]], - alpha_num[request->start[X]], - alpha_num[request->start[Y]], - alpha_num[request->start[Z]]); - } - - proc_cnt = bg_record->bp_count * bg_record->cpus_per_bp; - if(proc_cnt == request->procs) { - debug2("found it here %s, %s", - bg_record->bg_block_id, - bg_record->nodes); - request->save_name = xstrdup_printf( - "%c%c%c", - alpha_num[bg_record->start[X]], - alpha_num[bg_record->start[Y]], - alpha_num[bg_record->start[Z]]); - rc = SLURM_SUCCESS; + case 128: + switch(cnodes) { + case 32: + blockreq.small32 = 4; + break; + default: + error("We don't make a %d from size %d", + cnodes, bg_record->node_cnt); goto finished; - } - - if(bg_record->node_cnt > bluegene_bp_node_cnt) - continue; - if(proc_cnt < request->procs) { - if(last_quarter != bg_record->quarter){ - last_quarter = bg_record->quarter; - total_proc_cnt = proc_cnt; - } else { - total_proc_cnt += proc_cnt; - } - debug2("got %d on quarter %d", - total_proc_cnt, last_quarter); - if(total_proc_cnt == request->procs) { - request->save_name = xstrdup_printf( - "%c%c%c", - alpha_num[bg_record->start[X]], - alpha_num[bg_record->start[Y]], - alpha_num[bg_record->start[Z]]); - if(!my_block_list) { - rc = SLURM_SUCCESS; - goto finished; - } - bg_record = create_small_record( - bg_record, - last_quarter, - (uint16_t) NO_VAL); - list_append(new_blocks, bg_record); - - rc = SLURM_SUCCESS; - goto finished; - } - continue; - } - break; - } -found_one: - if(bg_record) { - List temp_list = NULL; - bg_record_t *found_record = NULL; - - if(bg_record->original) { - debug3("This was a copy"); - found_record = bg_record->original; - } else { - debug3("looking for original"); - found_record = find_org_in_bg_list( - bg_list, bg_record); + break; } - if(!found_record) { - error("this record wasn't found in the list!"); - rc = SLURM_ERROR; + break; + default: + switch(cnodes) { + case 32: + blockreq.small32 = 4; + blockreq.small128 = 3; + break; + case 128: + blockreq.small128 = 4; + break; + default: + error("We don't make a %d from size %d", + cnodes, bg_record->node_cnt); goto finished; + break; } - - format_node_name(found_record, tmp_char, sizeof(tmp_char)); - - debug2("going to split %s, %s", - found_record->bg_block_id, - tmp_char); - request->save_name = xstrdup_printf( - "%c%c%c", - alpha_num[found_record->start[X]], - alpha_num[found_record->start[Y]], - alpha_num[found_record->start[Z]]); - if(!my_block_list) { - rc = SLURM_SUCCESS; - goto finished; - } - _split_block(block_list, new_blocks, - found_record, request->procs); - remove_from_bg_list(block_list, bg_record); - destroy_bg_record(bg_record); - remove_from_bg_list(bg_list, found_record); - temp_list = list_create(NULL); - list_push(temp_list, found_record); - num_block_to_free++; - free_block_list(temp_list); - list_destroy(temp_list); - rc = SLURM_SUCCESS; - goto finished; - } - -finished: - list_iterator_destroy(itr); - - return rc; -} + full_bp = true; + break; #else - -static int _split_block(List block_list, List new_blocks, - bg_record_t *bg_record, int cnodes) -{ - bool full_bp = false; - bitoff_t start = 0; - blockreq_t blockreq; - - memset(&blockreq, 0, sizeof(blockreq_t)); - - switch(bg_record->node_cnt) { case 16: error("We got a 16 we should never have this"); goto finished; @@ -937,17 +548,27 @@ } full_bp = true; break; +#endif } - if(!full_bp && bg_record->ionode_bitmap) - start = bit_ffs(bg_record->ionode_bitmap); + if(!full_bp && bg_record->ionode_bitmap) { + if((start = bit_ffs(bg_record->ionode_bitmap)) == -1) + start = 0; + } +#ifdef HAVE_BGL + debug2("Asking for %u 32CNBlocks, and %u 128CNBlocks " + "from a %u block, starting at ionode %d.", + blockreq.small32, blockreq.small128, + bg_record->node_cnt, start); +#else debug2("Asking for %u 16CNBlocks, %u 32CNBlocks, " "%u 64CNBlocks, %u 128CNBlocks, and %u 256CNBlocks" "from a %u block, starting at ionode %d.", blockreq.small16, blockreq.small32, blockreq.small64, blockreq.small128, blockreq.small256, bg_record->node_cnt, start); +#endif handle_small_record_request(new_blocks, &blockreq, bg_record, start); finished: @@ -997,7 +618,7 @@ * smallest blocks. */ again: - while ((bg_record = (bg_record_t *) list_next(itr)) != NULL) { + while ((bg_record = list_next(itr))) { if(bg_record->job_running != NO_JOB_RUNNING) continue; /* on the third time through look for just a block @@ -1056,7 +677,7 @@ } /* lets see if we can combine some small ones */ if(bg_record->node_cnt < cnodes) { - //char bitstring[BITSIZE]; + char bitstring[BITSIZE]; bitstr_t *bitstr = NULL; bit_or(ionodes, bg_record->ionode_bitmap); @@ -1076,9 +697,12 @@ } else total_cnode_cnt += bg_record->node_cnt; - //bit_fmt(bitstring, BITSIZE, ionodes); - debug2("1 adding %d got %d set", - bg_record->node_cnt, total_cnode_cnt); + bit_fmt(bitstring, BITSIZE, ionodes); + debug2("1 adding %s %d got %d set " + "ionodes %s total is %s", + bg_record->bg_block_id, + bg_record->node_cnt, total_cnode_cnt, + bg_record->ionodes, bitstring); if(total_cnode_cnt == cnodes) { request->save_name = xstrdup_printf( "%c%c%c", @@ -1168,5 +792,3 @@ return rc; } - -#endif diff -Nru slurm-llnl-1.3.13/src/plugins/select/bluegene/plugin/dynamic_block.h slurm-llnl-1.3.15/src/plugins/select/bluegene/plugin/dynamic_block.h --- slurm-llnl-1.3.13/src/plugins/select/bluegene/plugin/dynamic_block.h 2008-11-12 23:37:27.000000000 +0100 +++ slurm-llnl-1.3.15/src/plugins/select/bluegene/plugin/dynamic_block.h 2009-04-01 01:23:01.000000000 +0200 @@ -42,9 +42,10 @@ #include "bluegene.h" extern List create_dynamic_block(List block_list, - ba_request_t *request, List my_block_list); + ba_request_t *request, List my_block_list, + bool track_down_nodes); -#ifdef HAVE_BGL +#ifdef HAVE_BGQ extern bg_record_t *create_small_record(bg_record_t *bg_record, uint16_t quarter, uint16_t nodecard); #else diff -Nru slurm-llnl-1.3.13/src/plugins/select/bluegene/plugin/select_bluegene.c slurm-llnl-1.3.15/src/plugins/select/bluegene/plugin/select_bluegene.c --- slurm-llnl-1.3.13/src/plugins/select/bluegene/plugin/select_bluegene.c 2009-01-06 19:20:48.000000000 +0100 +++ slurm-llnl-1.3.15/src/plugins/select/bluegene/plugin/select_bluegene.c 2009-04-07 19:24:20.000000000 +0200 @@ -1,7 +1,7 @@ /*****************************************************************************\ * select_bluegene.c - node selection plugin for Blue Gene system. * - * $Id: select_bluegene.c 16146 2009-01-06 18:20:48Z da $ + * $Id: select_bluegene.c 17175 2009-04-07 17:24:20Z da $ ***************************************************************************** * Copyright (C) 2004-2006 The Regents of the University of California. * Produced at Lawrence Livermore National Laboratory (cf, DISCLAIMER). @@ -85,12 +85,12 @@ const uint32_t plugin_version = 100; /* pthread stuff for updating BG node status */ -static pthread_t bluegene_thread = 0; +static pthread_t block_thread = 0; +static pthread_t state_thread = 0; static pthread_mutex_t thread_flag_mutex = PTHREAD_MUTEX_INITIALIZER; /** initialize the status pthread */ static int _init_status_pthread(void); -static int _wait_for_thread (pthread_t thread_id); static char *_block_state_str(int state); extern int select_p_alter_node_cnt(enum select_node_cnt type, void *data); @@ -108,22 +108,21 @@ fatal("SYSTEM_DIMENSIONS value (%d) invalid for Blue Gene", SYSTEM_DIMENSIONS); #endif -#ifdef HAVE_BG_FILES - if (!getenv("CLASSPATH") || !getenv("DB2INSTANCE") - || !getenv("VWSPATH")) - fatal("db2profile has not been run to setup DB2 environment"); - - if ((SELECT_MESH != RM_MESH) - || (SELECT_TORUS != RM_TORUS) - || (SELECT_NAV != RM_NAV)) - fatal("enum conn_type out of sync with rm_api.h"); +#ifdef HAVE_BG_FILES #ifdef HAVE_BGL + if (!getenv("CLASSPATH") || !getenv("DB2INSTANCE") + || !getenv("VWSPATH")) + fatal("db2profile has not been run to setup DB2 environment"); + if ((SELECT_COPROCESSOR_MODE != RM_PARTITION_COPROCESSOR_MODE) - || (SELECT_VIRTUAL_NODE_MODE != RM_PARTITION_VIRTUAL_NODE_MODE)) + || (SELECT_VIRTUAL_NODE_MODE != RM_PARTITION_VIRTUAL_NODE_MODE)) fatal("enum node_use_type out of sync with rm_api.h"); #endif - + if ((SELECT_MESH != RM_MESH) + || (SELECT_TORUS != RM_TORUS) + || (SELECT_NAV != RM_NAV)) + fatal("enum conn_type out of sync with rm_api.h"); #endif verbose("%s loading...", plugin_name); @@ -138,37 +137,29 @@ pthread_attr_t attr; pthread_mutex_lock( &thread_flag_mutex ); - if ( bluegene_thread ) { - debug2("Bluegene thread already running, not starting " - "another"); + if ( block_thread ) { + debug2("Bluegene threads already running, not starting " + "another"); pthread_mutex_unlock( &thread_flag_mutex ); return SLURM_ERROR; } slurm_attr_init( &attr ); - pthread_attr_setdetachstate( &attr, PTHREAD_CREATE_DETACHED ); - if (pthread_create( &bluegene_thread, &attr, bluegene_agent, NULL) + /* since we do a join on this later we don't make it detached */ + if (pthread_create( &block_thread, &attr, block_agent, NULL) != 0) - error("Failed to create bluegene_agent thread"); + error("Failed to create block_agent thread"); + slurm_attr_init( &attr ); + /* since we do a join on this later we don't make it detached */ + if (pthread_create( &state_thread, &attr, state_agent, NULL) + != 0) + error("Failed to create state_agent thread"); pthread_mutex_unlock( &thread_flag_mutex ); slurm_attr_destroy( &attr ); return SLURM_SUCCESS; } -static int _wait_for_thread (pthread_t thread_id) -{ - int i; - - for (i=0; i<4; i++) { - if (pthread_kill(thread_id, 0)) - return SLURM_SUCCESS; - sleep(1); - } - error("Could not kill select script pthread"); - return SLURM_ERROR; -} - static char *_block_state_str(int state) { static char tmp[16]; @@ -190,15 +181,18 @@ { int rc = SLURM_SUCCESS; - pthread_mutex_lock( &thread_flag_mutex ); agent_fini = true; - if ( bluegene_thread ) { + pthread_mutex_lock( &thread_flag_mutex ); + if ( block_thread ) { verbose("Bluegene select plugin shutting down"); - rc = _wait_for_thread(bluegene_thread); - bluegene_thread = 0; + pthread_join(block_thread, NULL); + block_thread = 0; + } + if ( state_thread ) { + pthread_join(state_thread, NULL); + state_thread = 0; } pthread_mutex_unlock( &thread_flag_mutex ); - fini_bg(); return rc; @@ -370,7 +364,7 @@ extern int select_p_node_init(struct node_record *node_ptr, int node_cnt) { if(node_cnt>0) - if(node_ptr->cpus > 512) + if(node_ptr->cpus >= bluegene_bp_node_cnt) procs_per_node = node_ptr->cpus; return SLURM_SUCCESS; } @@ -476,7 +470,7 @@ debug2("Node select info hasn't changed since %d", last_bg_update); return SLURM_NO_CHANGE_IN_DATA; - } else { + } else if(blocks_are_created) { *buffer_ptr = NULL; buffer = init_buf(HUGE_BUF_SIZE); pack32(blocks_packed, buffer); @@ -485,10 +479,7 @@ if(bg_list) { slurm_mutex_lock(&block_state_mutex); itr = list_iterator_create(bg_list); - while ((bg_record = (bg_record_t *) list_next(itr)) - != NULL) { - xassert(bg_record->bg_block_id != NULL); - + while ((bg_record = list_next(itr))) { pack_block(bg_record, buffer); blocks_packed++; } @@ -521,6 +512,9 @@ set_buf_offset(buffer, tmp_offset); *buffer_ptr = buffer; + } else { + error("select_p_pack_node_info: bg_list not ready yet"); + return SLURM_ERROR; } return SLURM_SUCCESS; @@ -623,23 +617,9 @@ } if(!part_desc_ptr->state_up) { - /* Since we are putting this block in an error state we need - to wait for the job to be removed. We don't really - need to free the block though since we may just - want it to be in an error state for some reason. */ - while(bg_record->job_running > NO_JOB_RUNNING) - sleep(1); - - slurm_mutex_lock(&block_state_mutex); - bg_record->job_running = BLOCK_ERROR_STATE; - bg_record->state = RM_PARTITION_ERROR; - slurm_mutex_unlock(&block_state_mutex); - trigger_block_error(); + put_block_in_error_state(bg_record, BLOCK_ERROR_STATE); } else if(part_desc_ptr->state_up){ - slurm_mutex_lock(&block_state_mutex); - bg_record->job_running = NO_JOB_RUNNING; - bg_record->state = RM_PARTITION_FREE; - slurm_mutex_unlock(&block_state_mutex); + resume_block(bg_record); } else { return rc; } @@ -652,19 +632,12 @@ extern int select_p_update_sub_node (update_part_msg_t *part_desc_ptr) { int rc = SLURM_SUCCESS; - bg_record_t *bg_record = NULL, *found_record = NULL; - time_t now; - char reason[128], tmp[64], time_str[32]; - blockreq_t blockreq; int i = 0, j = 0; - char coord[BA_SYSTEM_DIMENSIONS]; + char coord[BA_SYSTEM_DIMENSIONS+1], *node_name = NULL; char ionodes[128]; int set = 0; - int set_error = 0; + double nc_pos = 0, last_pos = -1; bitstr_t *ionode_bitmap = NULL; - List requests = NULL; - List delete_list = NULL; - ListIterator itr; if(bluegene_layout_mode != LAYOUT_DYNAMIC) { info("You can't use this call unless you are on a Dynamically " @@ -673,19 +646,14 @@ goto end_it; } - memset(coord, -1, BA_SYSTEM_DIMENSIONS); + memset(coord, 0, sizeof(coord)); memset(ionodes, 0, 128); if(!part_desc_ptr->name) { error("update_sub_node: No name specified"); rc = SLURM_ERROR; goto end_it; - } - now = time(NULL); - slurm_make_time_str(&now, time_str, sizeof(time_str)); - snprintf(tmp, sizeof(tmp), "[SLURM@%s]", time_str); - while (part_desc_ptr->name[j] != '\0') { if (part_desc_ptr->name[j] == '[') { if(set<1) { @@ -736,9 +704,9 @@ goto end_it; } } + strncpy(coord, part_desc_ptr->name+j, BA_SYSTEM_DIMENSIONS); - j += BA_SYSTEM_DIMENSIONS-1; set++; } @@ -752,165 +720,42 @@ goto end_it; } ionode_bitmap = bit_alloc(bluegene_numpsets); - bit_unfmt(ionode_bitmap, ionodes); - - requests = list_create(destroy_bg_record); - memset(&blockreq, 0, sizeof(blockreq_t)); - - blockreq.block = coord; - blockreq.conn_type = SELECT_SMALL; - blockreq.small32 = bluegene_bp_nodecard_cnt; - - add_bg_record(requests, NULL, &blockreq); - - delete_list = list_create(NULL); - while((bg_record = list_pop(requests))) { - set_error = 0; - if(bit_overlap(bg_record->ionode_bitmap, ionode_bitmap)) - set_error = 1; - - slurm_mutex_lock(&block_state_mutex); - itr = list_iterator_create(bg_list); - while((found_record = list_next(itr))) { - if(!found_record || (bg_record == found_record)) - continue; - if(bit_equal(bg_record->bitmap, found_record->bitmap) - && bit_equal(bg_record->ionode_bitmap, - found_record->ionode_bitmap)) { - debug2("block %s[%s] already there", - found_record->nodes, - found_record->ionodes); - /* we don't need to set this error, it - doesn't overlap - */ - if(!set_error) - break; - - snprintf(reason, sizeof(reason), - "update_sub_node: " - "Admin set block %s state to %s %s", - found_record->bg_block_id, - _block_state_str( - part_desc_ptr->state_up), - tmp); - info("%s",reason); - if(found_record->job_running - > NO_JOB_RUNNING) { - slurm_fail_job( - found_record->job_running); + bit_unfmt(ionode_bitmap, ionodes); + if(bit_ffs(ionode_bitmap) == -1) { + error("update_sub_node: Invalid ionode '%s' given.", ionodes); + rc = SLURM_ERROR; + FREE_NULL_BITMAP(ionode_bitmap); + goto end_it; + } + node_name = xstrdup_printf("%s%s", bg_slurm_node_prefix, coord); + /* find out how many nodecards to get for each ionode */ + if(!part_desc_ptr->state_up) { + info("Admin setting %s[%s] in an error state", + node_name, ionodes); + for(i = 0; istate_up) { - found_record->job_running = - BLOCK_ERROR_STATE; - found_record->state = - RM_PARTITION_ERROR; - trigger_block_error(); - } else if(part_desc_ptr->state_up){ - found_record->job_running = - NO_JOB_RUNNING; - found_record->state = - RM_PARTITION_FREE; - } else { - error("update_sub_node: " - "Unknown state %d given", - part_desc_ptr->state_up); - rc = SLURM_ERROR; - break; - } - break; - } else if(!set_error - && bit_equal(bg_record->bitmap, - found_record->bitmap) - && bit_overlap( - bg_record->ionode_bitmap, - found_record->ionode_bitmap)) { - break; } - - } - list_iterator_destroy(itr); - slurm_mutex_unlock(&block_state_mutex); - /* we already found an existing record */ - if(found_record) { - destroy_bg_record(bg_record); - continue; - } - /* we need to add this record since it doesn't exist */ - if(configure_block(bg_record) == SLURM_ERROR) { - destroy_bg_record(bg_record); - error("update_sub_node: " - "unable to configure block in api"); - } - debug2("adding block %s to fill in small blocks " - "around bad blocks", - bg_record->bg_block_id); - print_bg_record(bg_record); - slurm_mutex_lock(&block_state_mutex); - list_append(bg_list, bg_record); - slurm_mutex_unlock(&block_state_mutex); - - /* We are just adding the block not deleting any or - setting this one to an error state. - */ - if(!set_error) - continue; - - if(!part_desc_ptr->state_up) { - bg_record->job_running = BLOCK_ERROR_STATE; - bg_record->state = RM_PARTITION_ERROR; - trigger_block_error(); - } else if(part_desc_ptr->state_up){ - bg_record->job_running = NO_JOB_RUNNING; - bg_record->state = RM_PARTITION_FREE; - } else { - error("update_sub_node: Unknown state %d given", - part_desc_ptr->state_up); - rc = SLURM_ERROR; - continue; + nc_pos += bluegene_nc_ratio; } - snprintf(reason, sizeof(reason), - "update_sub_node: " - "Admin set block %s state to %s %s", - bg_record->bg_block_id, - _block_state_str(part_desc_ptr->state_up), - tmp); - info("%s",reason); - - /* remove overlapping blocks */ - slurm_mutex_lock(&block_state_mutex); - itr = list_iterator_create(bg_list); - while((found_record = list_next(itr))) { - if ((!found_record) || (bg_record == found_record)) - continue; - if(!blocks_overlap(bg_record, found_record)) { - debug2("block %s isn't part of %s", - found_record->bg_block_id, - bg_record->bg_block_id); - continue; - } - debug2("removing block %s because there is something " - "wrong with part of the base partition", - found_record->bg_block_id); - if(found_record->job_running > NO_JOB_RUNNING) { - slurm_fail_job(found_record->job_running); - } - list_push(delete_list, found_record); - list_remove(itr); - num_block_to_free++; - } - list_iterator_destroy(itr); - free_block_list(delete_list); - slurm_mutex_unlock(&block_state_mutex); + } else if(part_desc_ptr->state_up){ + info("Admin setting %s[%s] in an free state", + node_name, ionodes); + up_nodecard(node_name, ionode_bitmap); + } else { + error("update_sub_node: Unknown state %d", + part_desc_ptr->state_up); + rc = SLURM_ERROR; } - list_destroy(delete_list); + FREE_NULL_BITMAP(ionode_bitmap); - - /* This only works for the error state, not free */ + xfree(node_name); last_bg_update = time(NULL); - -end_it: +end_it: return rc; } diff -Nru slurm-llnl-1.3.13/src/plugins/select/bluegene/plugin/sfree.c slurm-llnl-1.3.15/src/plugins/select/bluegene/plugin/sfree.c --- slurm-llnl-1.3.13/src/plugins/select/bluegene/plugin/sfree.c 2008-11-04 23:05:21.000000000 +0100 +++ slurm-llnl-1.3.15/src/plugins/select/bluegene/plugin/sfree.c 2009-01-30 19:05:07.000000000 +0100 @@ -1,6 +1,6 @@ /*****************************************************************************\ * sfree.c - free specified block or all blocks. - * $Id: sfree.c 15597 2008-11-04 22:05:21Z da $ + * $Id: sfree.c 16357 2009-01-30 18:05:07Z da $ ***************************************************************************** * Copyright (C) 2004 The Regents of the University of California. * Produced at Lawrence Livermore National Laboratory (cf, DISCLAIMER). @@ -284,7 +284,8 @@ if (delete_record->state != (rm_partition_state_t)NO_VAL && delete_record->state != RM_PARTITION_FREE && delete_record->state != RM_PARTITION_DEALLOCATING) { - info("bridge_destroy %s",delete_record->bg_block_id); + info("bridge_destroy %s", delete_record->bg_block_id); +#ifdef HAVE_BG_FILES if ((rc = bridge_destroy_block( delete_record->bg_block_id)) != STATUS_OK) { @@ -296,6 +297,9 @@ delete_record->bg_block_id, _bg_err_str(rc)); } +#else + bg_record->state = RM_PARTITION_FREE; +#endif } if(!wait_full) { @@ -306,8 +310,12 @@ } if ((delete_record->state == RM_PARTITION_FREE) - || (delete_record->state == RM_PARTITION_ERROR)) +#ifdef HAVE_BGL + || (delete_record->state == RM_PARTITION_ERROR) +#endif + ) { break; + } sleep(3); } info("bgblock %s is freed", delete_record->bg_block_id); diff -Nru slurm-llnl-1.3.13/src/plugins/select/bluegene/plugin/slurm_prolog.c slurm-llnl-1.3.15/src/plugins/select/bluegene/plugin/slurm_prolog.c --- slurm-llnl-1.3.13/src/plugins/select/bluegene/plugin/slurm_prolog.c 2008-09-03 19:38:39.000000000 +0200 +++ slurm-llnl-1.3.15/src/plugins/select/bluegene/plugin/slurm_prolog.c 2009-03-10 21:52:29.000000000 +0100 @@ -125,8 +125,11 @@ break; /* fatal error */ if (rc == READY_JOB_ERROR) /* error */ continue; /* retry */ - if ((rc & READY_JOB_STATE) == 0) /* job killed */ + if ((rc & READY_JOB_STATE) == 0) { /* job killed */ + /* return 1 so we don't get a prolog error */ + is_ready = 1; break; + } if (rc & READY_NODE_STATE) { /* job and node ready */ is_ready = 1; break; diff -Nru slurm-llnl-1.3.13/src/plugins/select/bluegene/plugin/state_test.c slurm-llnl-1.3.15/src/plugins/select/bluegene/plugin/state_test.c --- slurm-llnl-1.3.13/src/plugins/select/bluegene/plugin/state_test.c 2008-11-06 00:28:45.000000000 +0100 +++ slurm-llnl-1.3.15/src/plugins/select/bluegene/plugin/state_test.c 2009-04-09 18:56:23.000000000 +0200 @@ -2,7 +2,7 @@ * state_test.c - Test state of Bluegene base partitions and switches. * DRAIN nodes in SLURM that are not usable. * - * $Id: state_test.c 15611 2008-11-05 23:28:45Z da $ + * $Id: state_test.c 17202 2009-04-09 16:56:23Z da $ ***************************************************************************** * Copyright (C) 2004-2006 The Regents of the University of California. * Produced at Lawrence Livermore National Laboratory (cf, DISCLAIMER). @@ -55,7 +55,7 @@ #ifdef HAVE_BG_FILES /* Find the specified BlueGene node ID and drain it from SLURM */ -static void _configure_node_down(rm_bp_id_t bp_id, my_bluegene_t *bg) +static void _configure_node_down(rm_bp_id_t bp_id, my_bluegene_t *my_bg) { int bp_num, i, rc; rm_bp_id_t bpid; @@ -65,21 +65,21 @@ char bg_down_node[128], reason[128], time_str[32]; time_t now = time(NULL); - if ((rc = bridge_get_data(bg, RM_BPNum, &bp_num)) != STATUS_OK) { + if ((rc = bridge_get_data(my_bg, RM_BPNum, &bp_num)) != STATUS_OK) { error("bridge_get_data(RM_BPNum): %s", bg_err_str(rc)); bp_num = 0; } for (i=0; ijob_running != BLOCK_ERROR_STATE) */ +/* continue; */ + +/* if(!bit_test(bg_record->bitmap, bp_bit)) */ +/* continue; */ +/* info("bringing %s back to service", */ +/* bg_record->bg_block_id); */ +/* bg_record->job_running = NO_JOB_RUNNING; */ +/* bg_record->state = RM_PARTITION_FREE; */ +/* last_bg_update = time(NULL); */ +/* } */ +/* list_iterator_destroy(itr); */ +/* slurm_mutex_unlock(&block_state_mutex); */ + +/* /\* FIX ME: This needs to call the opposite of */ +/* slurm_drain_nodes which does not yet exist. */ +/* *\/ */ +/* if((ret = node_already_down(node_name))) { */ +/* /\* means it was drained *\/ */ +/* if(ret == 2) { */ +/* /\* debug("node %s put back into service after " *\/ */ +/* /\* "being in an error state", *\/ */ +/* /\* node_name); *\/ */ +/* } */ +/* } */ +/* } */ + +clean_up: + xfree(node_name); +/* if(ionode_bitmap) */ +/* FREE_NULL_BITMAP(ionode_bitmap); */ + free(bp_id); + + return rc; } /* Test for nodes that are not UP in MMCS and DRAIN them in SLURM */ -static void _test_down_nodes(my_bluegene_t *bg) +static void _test_down_nodes(my_bluegene_t *my_bg) { int bp_num, i, rc; rm_BP_t *my_bp; - rm_BP_state_t bp_state; - rm_location_t bp_loc; - char down_node_list[BUFSIZE]; - char bg_down_node[128]; - char reason[128], time_str[32]; - time_t now = time(NULL); debug2("Running _test_down_nodes"); - down_node_list[0] = '\0'; - if ((rc = bridge_get_data(bg, RM_BPNum, &bp_num)) != STATUS_OK) { + if ((rc = bridge_get_data(my_bg, RM_BPNum, &bp_num)) != STATUS_OK) { error("bridge_get_data(RM_BPNum): %s", bg_err_str(rc)); bp_num = 0; } for (i=0; inode_state & (~NODE_STATE_NO_RESPOND); - if ((base_state == NODE_STATE_DOWN) - || (base_state == NODE_STATE_DRAIN)) - return true; + + if(base_state & NODE_STATE_DRAIN) + return 2; + else if (base_state == NODE_STATE_DOWN) + return 1; else - return false; + return 0; } - return false; + return 0; } /* @@ -331,19 +453,19 @@ extern void test_mmcs_failures(void) { #ifdef HAVE_BG_FILES - my_bluegene_t *bg; + my_bluegene_t *local_bg; int rc; - if ((rc = bridge_get_bg(&bg)) != STATUS_OK) { + if ((rc = bridge_get_bg(&local_bg)) != STATUS_OK) { error("bridge_get_BG(): %s", bg_err_str(rc)); return; } - _test_down_switches(bg); - _test_down_nodes(bg); - if ((rc = bridge_free_bg(bg)) != STATUS_OK) + _test_down_switches(local_bg); + _test_down_nodes(local_bg); + if ((rc = bridge_free_bg(local_bg)) != STATUS_OK) error("bridge_free_BG(): %s", bg_err_str(rc)); #endif } @@ -354,16 +476,8 @@ #ifdef HAVE_BG_FILES rm_partition_t *block_ptr = NULL; rm_BP_t *bp_ptr = NULL; - char *bpid = NULL; int bp_cnt = 0; int i = 0; - int *coord = NULL; - rm_BP_state_t bp_state; - char bg_down_node[128], reason[128], time_str[32]; - char down_node_list[BUFSIZE]; - time_t now = time(NULL); - - down_node_list[0] = '\0'; if ((rc = bridge_get_block(bg_block_id, &block_ptr)) != STATUS_OK) { error("Block %s doesn't exist.", bg_block_id); @@ -402,60 +516,13 @@ break; } } - if ((rc = bridge_get_data(bp_ptr, RM_BPState, &bp_state)) - != STATUS_OK) { - error("bridge_get_data(RM_BPLoc): %s", - bg_err_str(rc)); - rc = SLURM_ERROR; - break; - } - if(bp_state == RM_BP_UP) - continue; - rc = SLURM_ERROR; - if ((rc = bridge_get_data(bp_ptr, RM_BPID, &bpid)) - != STATUS_OK) { - error("bridge_get_data(RM_BPID): %s", - bg_err_str(rc)); - break; - } - coord = find_bp_loc(bpid); - - if(!coord) { - fatal("Could not find coordinates for " - "BP ID %s", (char *) bpid); - } - free(bpid); - - snprintf(bg_down_node, sizeof(bg_down_node), "%s%c%c%c", - bg_slurm_node_prefix, - alpha_num[coord[X]], alpha_num[coord[Y]], - alpha_num[coord[Z]]); - - - if (node_already_down(bg_down_node)) - continue; - debug("check_block_bp_states: %s in state %s", - bg_down_node, _convert_bp_state(bp_state)); - if ((strlen(down_node_list) + strlen(bg_down_node) + 2) - < BUFSIZE) { - if (down_node_list[0] != '\0') - strcat(down_node_list,","); - strcat(down_node_list, bg_down_node); - } else - error("down_node_list overflow"); + _test_down_nodecards(bp_ptr); } cleanup: bridge_free_block(block_ptr); done: - if (down_node_list[0]) { - slurm_make_time_str(&now, time_str, sizeof(time_str)); - snprintf(reason, sizeof(reason), - "select_bluegene: MMCS state not UP [SLURM@%s]", - time_str); - slurm_drain_nodes(down_node_list, reason); - } #endif return rc; diff -Nru slurm-llnl-1.3.13/src/plugins/select/bluegene/plugin/state_test.h slurm-llnl-1.3.15/src/plugins/select/bluegene/plugin/state_test.h --- slurm-llnl-1.3.13/src/plugins/select/bluegene/plugin/state_test.h 2006-09-05 20:00:19.000000000 +0200 +++ slurm-llnl-1.3.15/src/plugins/select/bluegene/plugin/state_test.h 2009-04-01 01:23:01.000000000 +0200 @@ -1,6 +1,6 @@ /*****************************************************************************\ * state_test.h - header for Blue Gene node and switch state test. - * $Id: state_test.h 9170 2006-09-05 18:00:19Z jette $ + * $Id: state_test.h 17102 2009-03-31 23:23:01Z da $ ***************************************************************************** * Copyright (C) 2004 The Regents of the University of California. * Produced at Lawrence Livermore National Laboratory (cf, DISCLAIMER). @@ -38,8 +38,9 @@ #ifndef _STATE_TEST_H_ #define _STATE_TEST_H_ -/* Determine if specific slurm node is already in DOWN or DRAIN state */ -extern bool node_already_down(char *node_name); +/* Determine if specific slurm node is already in DOWN or DRAIN ret (1) or + * FAIL ret (2) state idle ret (0) */ +extern int node_already_down(char *node_name); /* * Search MMCS for failed switches and nodes. Failed resources are DRAINED in diff -Nru slurm-llnl-1.3.13/src/plugins/select/cons_res/select_cons_res.c slurm-llnl-1.3.15/src/plugins/select/cons_res/select_cons_res.c --- slurm-llnl-1.3.13/src/plugins/select/cons_res/select_cons_res.c 2008-12-05 01:45:29.000000000 +0100 +++ slurm-llnl-1.3.15/src/plugins/select/cons_res/select_cons_res.c 2009-03-25 19:42:18.000000000 +0100 @@ -2,7 +2,7 @@ * select_cons_res.c - node selection plugin supporting consumable * resources policies. * - * $Id: select_cons_res.c 15841 2008-12-05 00:45:29Z jette $ + * $Id: select_cons_res.c 17022 2009-03-25 18:42:18Z jette $ *****************************************************************************\ * * The following example below illustrates how four jobs are allocated @@ -1910,11 +1910,9 @@ } } - /* allocated node count should never exceed num_procs, right? - * if so, then this should be done earlier and max_nodes - * could be used to make this process more efficient (truncate - * # of available nodes when (# of idle nodes == max_nodes)*/ - if (max_nodes > job_ptr->num_procs) + /* NOTE: num_procs is 1 by default. + * Only reset max_nodes if user explicitly sets a process count */ + if ((job_ptr->num_procs > 1) && (max_nodes > job_ptr->num_procs)) max_nodes = job_ptr->num_procs; origmap = bit_copy(bitmap); diff -Nru slurm-llnl-1.3.13/src/plugins/select/linear/select_linear.c slurm-llnl-1.3.15/src/plugins/select/linear/select_linear.c --- slurm-llnl-1.3.13/src/plugins/select/linear/select_linear.c 2008-09-19 02:35:14.000000000 +0200 +++ slurm-llnl-1.3.15/src/plugins/select/linear/select_linear.c 2009-02-19 01:46:20.000000000 +0100 @@ -661,20 +661,34 @@ { ListIterator job_iterator; struct job_record *job_scan_ptr; + int rc = EINVAL; job_iterator = list_iterator_create(job_list); while ((job_scan_ptr = (struct job_record *) list_next(job_iterator))) { - if ((job_scan_ptr->part_ptr == job_ptr->part_ptr) && - (job_scan_ptr->job_state == JOB_RUNNING) && - (job_scan_ptr->node_cnt == req_nodes) && - (job_scan_ptr->total_procs >= job_ptr->num_procs) && - bit_super_set(job_scan_ptr->node_bitmap, bitmap)) { - bit_and(bitmap, job_scan_ptr->node_bitmap); - return SLURM_SUCCESS; - } + if ((job_scan_ptr->part_ptr != job_ptr->part_ptr) || + (job_scan_ptr->job_state != JOB_RUNNING) || + (job_scan_ptr->node_cnt != req_nodes) || + (job_scan_ptr->total_procs < job_ptr->num_procs) || + (!bit_super_set(job_scan_ptr->node_bitmap, bitmap))) + continue; + + if (job_ptr->details->req_node_bitmap && + (!bit_super_set(job_ptr->details->req_node_bitmap, + job_scan_ptr->node_bitmap))) + continue; /* Required nodes missing from job */ + + if (job_ptr->details->exc_node_bitmap && + (bit_overlap(job_ptr->details->exc_node_bitmap, + job_scan_ptr->node_bitmap) != 0)) + continue; /* Excluded nodes in this job */ + + bit_and(bitmap, job_scan_ptr->node_bitmap); + job_ptr->total_procs = job_scan_ptr->total_procs; + rc = SLURM_SUCCESS; + break; } list_iterator_destroy(job_iterator); - return EINVAL; + return rc; } /* _job_test - does most of the real work for select_p_job_test(), which diff -Nru slurm-llnl-1.3.13/src/sacct/options.c slurm-llnl-1.3.15/src/sacct/options.c --- slurm-llnl-1.3.13/src/sacct/options.c 2008-12-22 22:46:26.000000000 +0100 +++ slurm-llnl-1.3.15/src/sacct/options.c 2009-03-19 21:54:03.000000000 +0100 @@ -408,13 +408,15 @@ " Display job accounting data for all users. By default, only\n" " data for the current user is displayed for users other than\n" " root.\n" + "-A, --accounts\n" + " Only send data about these accounts. Default is all.\n" "-b, --brief\n" " Equivalent to \"--fields=jobstep,state,error\". This option\n" " has no effect if --dump is specified.\n" "-c, --completion\n" " Use job completion instead of accounting data.\n" - "-C, --cluster\n" - " Only send data about this cluster -1 for all clusters.\n" + "-C, --clusters\n" + " Only send data about these clusters. -1 for all clusters.\n" "-d, --dump\n" " Dump the raw data records\n" "--duplicates\n" @@ -444,10 +446,12 @@ " followed by \"d\", it is interpreted as number of days. For\n" " example, \"--expire=14d\" means that you wish to purge the job\n" " accounting log of all jobs that completed more than 14 days ago.\n" - "-F , --fields=\n" - " Display the specified data (use \"--help-fields\" for a\n" - " list of available fields). If no field option is specified,\n" - " we use \"--fields=jobstep,jobname,partition,alloc_cpus,state,error\".\n" + "--endtime: \n" + " Select jobs eligible before this time. \n" + "-F , --format=\n" + " Display the specified data (use \"--helpformat\" for a\n" + " list of available fields). If no format option is specified,\n" + " we use \"--format=jobstep,jobname,partition,alloc_cpus,state,error\".\n" "-f, --file=\n" " Read data from the specified file, rather than SLURM's current\n" " accounting log file.\n" @@ -486,6 +490,8 @@ "-S, --stat\n" " Get real time state of a jobstep supplied by the -j\n" " option\n" + "--starttime: \n" + " Select jobs eligible after this time. \n" "-t, --total\n" " Only show cumulative statistics for each job, not the\n" " intermediate steps\n" @@ -496,7 +502,15 @@ " Pointer to this message.\n" "-v, --verbose\n" " Primarily for debugging purposes, report the state of various\n" - " variables during processing.\n", conf->slurm_conf); + " variables during processing.\n" + "-W, --wckeys\n" + " Only send data about these wckeys. Default is all.\n" + "\n" + "Note, valid start/end time formats are...\n" + " HH:MM[:SS] [AM|PM]\n" + " MMDD[YY] or MM/DD[/YY] or MM.DD[.YY]\n" + " MM/DD[/YY]-HH:MM[:SS]\n" + , conf->slurm_conf); slurm_conf_unlock(); @@ -624,19 +638,22 @@ {"accounts", 1, 0, 'A'}, {"begin", 1, 0, 'B'}, {"brief", 0, 0, 'b'}, - {"cluster", 1, 0, 'C'}, + {"clusters", 1, 0, 'C'}, {"completion", 0, ¶ms.opt_completion, 'c'}, {"duplicates", 0, ¶ms.opt_dup, 1}, {"dump", 0, 0, 'd'}, {"end", 1, 0, 'E'}, + {"endtime", 1, 0, 'E'}, {"expire", 1, 0, 'e'}, {"fields", 1, 0, 'F'}, + {"format", 1, 0, 'F'}, {"file", 1, 0, 'f'}, {"formatted_dump", 0, 0, 'O'}, {"gid", 1, 0, 'g'}, {"group", 1, 0, 'g'}, {"help", 0, ¶ms.opt_help, 1}, {"help-fields", 0, ¶ms.opt_help, 2}, + {"helpformat", 0, ¶ms.opt_help, 2}, {"jobs", 1, 0, 'j'}, {"long", 0, 0, 'l'}, {"big_logfile", 0, ¶ms.opt_lowmem, 1}, @@ -646,15 +663,16 @@ {"purge", 0, 0, 'P'}, {"state", 1, 0, 's'}, {"stat", 0, 0, 'S'}, + {"starttime", 1, 0, 'B'}, {"total", 0, 0, 't'}, {"uid", 1, 0, 'u'}, {"usage", 0, ¶ms.opt_help, 3}, {"user", 1, 0, 'u'}, {"verbose", 0, 0, 'v'}, {"version", 0, 0, 'V'}, + {"wckeys", 1, 0, 'W'}, {0, 0, 0, 0}}; - job_cond->duplicates = params.opt_dup; params.opt_uid = getuid(); params.opt_gid = getgid(); @@ -662,7 +680,7 @@ while (1) { /* now cycle through the command line */ c = getopt_long(argc, argv, - "aA:bB:cC:deE:F:f:g:hj:lOP:p:s:StUu:Vv", + "aA:bB:cC:deE:F:f:g:hj:lOP:p:s:StUu:VvW:", long_options, &optionIndex); if (c == -1) break; @@ -811,6 +829,12 @@ params.opt_verbose++; break; + case 'W': + if(!job_cond->wckey_list) + job_cond->wckey_list = + list_create(slurm_destroy_char); + slurm_addto_char_list(job_cond->wckey_list, optarg); + break; case 'V': printf("%s %s\n", PACKAGE, SLURM_VERSION); exit(0); @@ -828,6 +852,8 @@ if (params.opt_fdump) params.opt_dup |= FDUMP_FLAG; + job_cond->duplicates = params.opt_dup; + debug("Options selected:\n" "\topt_archive_jobs=%d\n" "\topt_archve_steps=%d\n" @@ -920,6 +946,12 @@ } } + /* if any jobs are specified set to look for all users if none + are set */ + if((job_cond->step_list && list_count(job_cond->step_list)) + && (!job_cond->userid_list || !list_count(job_cond->userid_list))) + all_users=1; + if(all_users) { if(job_cond->userid_list && list_count(job_cond->userid_list)) { @@ -996,6 +1028,15 @@ list_iterator_destroy(itr); } + if (params.opt_verbose && job_cond->wckey_list + && list_count(job_cond->wckey_list)) { + fprintf(stderr, "Wckeys requested:\n"); + itr = list_iterator_create(job_cond->wckey_list); + while((start = list_next(itr))) + fprintf(stderr, "\t: %s\n", start); + list_iterator_destroy(itr); + } + /* select the output fields */ if(brief_output) { if(params.opt_completion) @@ -1318,6 +1359,9 @@ jobacct_job_rec_t *job = NULL; jobacct_step_rec_t *step = NULL; + if(!jobs) + return; + if (params.opt_total) do_jobsteps = 0; itr = list_iterator_create(jobs); @@ -1364,7 +1408,10 @@ { ListIterator itr = NULL; jobcomp_job_rec_t *job = NULL; - + + if(!jobs) + return; + itr = list_iterator_create(jobs); while((job = list_next(itr))) { print_fields(JOBCOMP, job); diff -Nru slurm-llnl-1.3.13/src/sacct/print.c slurm-llnl-1.3.15/src/sacct/print.c --- slurm-llnl-1.3.13/src/sacct/print.c 2008-12-09 17:45:47.000000000 +0100 +++ slurm-llnl-1.3.15/src/sacct/print.c 2009-03-17 23:07:11.000000000 +0100 @@ -39,6 +39,7 @@ #include "sacct.h" #include "src/common/parse_time.h" +#include "src/common/hostlist.h" #include "slurm.h" #define FORMAT_STRING_SIZE 34 @@ -69,16 +70,20 @@ if (days) snprintf(str, FORMAT_STRING_SIZE, - "%ld-%2.2ld:%2.2ld:%2.2ld", + "%2.2ld-%2.2ld:%2.2ld:%2.2ld", days, hours, minutes, seconds); else if (hours) snprintf(str, FORMAT_STRING_SIZE, - "%ld:%2.2ld:%2.2ld", + "%2.2ld:%2.2ld:%2.2ld", hours, minutes, seconds); - else + else if(subsec) snprintf(str, FORMAT_STRING_SIZE, - "%ld:%2.2ld.%3.3ld", + "%2.2ld:%2.2ld.%3.3ld", minutes, seconds, subsec); + else + snprintf(str, FORMAT_STRING_SIZE, + "00:%2.2ld:%2.2ld", + minutes, seconds); } void print_fields(type_t type, void *object) @@ -406,25 +411,50 @@ void print_nnodes(type_t type, void *object) { + jobacct_job_rec_t *job = (jobacct_job_rec_t *)object; jobcomp_job_rec_t *jobcomp = (jobcomp_job_rec_t *)object; - char temp[FORMAT_STRING_SIZE]; + jobacct_step_rec_t *step = (jobacct_step_rec_t *)object; + char *tmp_char = NULL; + int tmp_int = NO_VAL; + hostlist_t hl = NULL; switch(type) { case HEADLINE: printf("%-8s", "Node Cnt"); + tmp_int = INFINITE; break; case UNDERSCORE: printf("%-8s", "--------"); + tmp_int = INFINITE; + break; + case JOB: + tmp_char = job->nodes; + break; + case JOBSTEP: + tmp_char = step->nodes; break; case JOBCOMP: - convert_num_unit((float)jobcomp->node_cnt, temp, - sizeof(temp), UNIT_NONE); - printf("%-8s", temp); + tmp_int = jobcomp->node_cnt; break; default: - printf("%-8s", "n/a"); break; - } + } + if(tmp_char) { + hl = hostlist_create(tmp_char); + tmp_int = hostlist_count(hl); + hostlist_destroy(hl); + } + + if(tmp_int == INFINITE) + return; + else if(tmp_int == NO_VAL) + printf("%-8s", "n/a"); + else { + char outbuf[FORMAT_STRING_SIZE]; + convert_num_unit((float)tmp_int, + outbuf, sizeof(outbuf), UNIT_NONE); + printf("%-8s", outbuf); + } } void print_ntasks(type_t type, void *object) @@ -724,6 +754,37 @@ } } +void print_eligible(type_t type, void *object) +{ + jobacct_job_rec_t *job = (jobacct_job_rec_t *)object; + jobacct_step_rec_t *step = (jobacct_step_rec_t *)object; + char time_str[32]; + + switch(type) { + case HEADLINE: + printf("%-14s", "Eligible Time"); + break; + case UNDERSCORE: + printf("%-14.14s", "--------------"); + break; + case JOB: + slurm_make_time_str(&job->eligible, + time_str, + sizeof(time_str)); + printf("%-14s", time_str); + break; + case JOBSTEP: + slurm_make_time_str(&step->start, + time_str, + sizeof(time_str)); + printf("%-14s", time_str); + break; + default: + printf("%-14s", "n/a"); + break; + } +} + void print_start(type_t type, void *object) { jobacct_job_rec_t *job = (jobacct_job_rec_t *)object; diff -Nru slurm-llnl-1.3.13/src/sacct/sacct.c slurm-llnl-1.3.15/src/sacct/sacct.c --- slurm-llnl-1.3.13/src/sacct/sacct.c 2008-12-22 22:46:26.000000000 +0100 +++ slurm-llnl-1.3.15/src/sacct/sacct.c 2009-03-13 20:28:05.000000000 +0100 @@ -158,6 +158,7 @@ {"cpu", print_cpu}, {"cputime", print_cputime}, {"elapsed", print_elapsed}, + {"eligible", print_eligible}, {"end", print_end}, {"exitcode", print_exitcode}, {"finished", print_end}, /* Defunct name */ diff -Nru slurm-llnl-1.3.13/src/sacct/sacct.h slurm-llnl-1.3.15/src/sacct/sacct.h --- slurm-llnl-1.3.13/src/sacct/sacct.h 2008-12-22 22:46:26.000000000 +0100 +++ slurm-llnl-1.3.15/src/sacct/sacct.h 2009-03-13 20:28:05.000000000 +0100 @@ -151,6 +151,7 @@ void print_rss(type_t type, void *object); void print_state(type_t type, void *object); void print_submit(type_t type, void *object); +void print_eligible(type_t type, void *object); void print_start(type_t type, void *object); void print_end(type_t type, void *object); void print_systemcpu(type_t type, void *object); diff -Nru slurm-llnl-1.3.13/src/sacctmgr/account_functions.c slurm-llnl-1.3.15/src/sacctmgr/account_functions.c --- slurm-llnl-1.3.13/src/sacctmgr/account_functions.c 2008-12-23 00:56:27.000000000 +0100 +++ slurm-llnl-1.3.15/src/sacctmgr/account_functions.c 2009-03-10 00:07:16.000000000 +0100 @@ -101,7 +101,9 @@ || !strncasecmp (argv[i], "Names", MAX(command_len, 1)) || !strncasecmp (argv[i], "Accounts", - MAX(command_len, 1))) { + MAX(command_len, 1)) + || !strncasecmp (argv[i], "Acct", + MAX(command_len, 4))) { if(!assoc_cond->acct_list) { assoc_cond->acct_list = list_create(slurm_destroy_char); @@ -134,6 +136,8 @@ if(format_list) slurm_addto_char_list(format_list, argv[i]+end); } else if (!strncasecmp (argv[i], "FairShare", + MAX(command_len, 1)) + || !strncasecmp (argv[i], "Shares", MAX(command_len, 1))) { if(!assoc_cond->fairshare_list) assoc_cond->fairshare_list = @@ -339,7 +343,9 @@ || !strncasecmp (argv[i], "Account", MAX(command_len, 1)) || !strncasecmp (argv[i], "Names", - MAX(command_len, 1))) { + MAX(command_len, 1)) + || !strncasecmp (argv[i], "Acct", + MAX(command_len, 4))) { if(acct_list) slurm_addto_char_list(acct_list, argv[i]+end); @@ -353,6 +359,8 @@ acct->description = strip_quotes(argv[i]+end, NULL, 1); u_set = 1; } else if (!strncasecmp (argv[i], "FairShare", + MAX(command_len, 1)) + || !strncasecmp (argv[i], "Shares", MAX(command_len, 1))) { if(!assoc) continue; @@ -1123,6 +1131,12 @@ field->name = xstrdup("Par Name"); field->len = 10; field->print_routine = print_fields_str; + } else if(!strncasecmp("Shares", object, + MAX(command_len, 1))) { + field->type = PRINT_FAIRSHARE; + field->name = xstrdup("Shares"); + field->len = 9; + field->print_routine = print_fields_uint; } else if(!strncasecmp("User", object, MAX(command_len, 1))) { field->type = PRINT_USER; field->name = xstrdup("User"); diff -Nru slurm-llnl-1.3.13/src/sacctmgr/association_functions.c slurm-llnl-1.3.15/src/sacctmgr/association_functions.c --- slurm-llnl-1.3.13/src/sacctmgr/association_functions.c 2008-12-20 00:18:59.000000000 +0100 +++ slurm-llnl-1.3.15/src/sacctmgr/association_functions.c 2009-03-10 00:07:16.000000000 +0100 @@ -108,7 +108,9 @@ list_iterator_destroy(itr); set = 1; } else if (!strncasecmp (argv[i], "Accounts", - MAX(command_len, 2))) { + MAX(command_len, 2)) + || !strncasecmp (argv[i], "Acct", + MAX(command_len, 4))) { if(!assoc_cond->acct_list) assoc_cond->acct_list = list_create(slurm_destroy_char); @@ -129,6 +131,8 @@ slurm_addto_char_list(format_list, argv[i]+end); } else if (!strncasecmp (argv[i], "FairShare", + MAX(command_len, 1)) + || !strncasecmp (argv[i], "Shares", MAX(command_len, 1))) { if(!assoc_cond->fairshare_list) assoc_cond->fairshare_list = @@ -373,7 +377,8 @@ field = xmalloc(sizeof(print_field_t)); - if(!strncasecmp("Account", object, MAX(command_len, 1))) { + if(!strncasecmp("Account", object, MAX(command_len, 1)) + || !strncasecmp("Acct", object, MAX(command_len, 4))) { field->type = PRINT_ACCOUNT; field->name = xstrdup("Account"); if(tree_display) @@ -522,6 +527,12 @@ field->name = xstrdup("RGT"); field->len = 6; field->print_routine = print_fields_uint; + } else if(!strncasecmp("Shares", object, + MAX(command_len, 1))) { + field->type = PRINT_FAIRSHARE; + field->name = xstrdup("Shares"); + field->len = 9; + field->print_routine = print_fields_uint; } else if(!strncasecmp("User", object, MAX(command_len, 1))) { field->type = PRINT_USER; field->name = xstrdup("User"); diff -Nru slurm-llnl-1.3.13/src/sacctmgr/cluster_functions.c slurm-llnl-1.3.15/src/sacctmgr/cluster_functions.c --- slurm-llnl-1.3.13/src/sacctmgr/cluster_functions.c 2008-12-12 22:04:09.000000000 +0100 +++ slurm-llnl-1.3.15/src/sacctmgr/cluster_functions.c 2009-03-10 00:07:16.000000000 +0100 @@ -130,6 +130,8 @@ if(name_list) slurm_addto_char_list(name_list, argv[i]+end); } else if (!strncasecmp (argv[i], "FairShare", + MAX(command_len, 1)) + || !strncasecmp (argv[i], "Shares", MAX(command_len, 1))) { if (get_uint(argv[i]+end, &assoc->fairshare, "FairShare") == SLURM_SUCCESS) @@ -211,7 +213,7 @@ " Bad MaxWall time format: %s\n", argv[i]); } - } else if (!strncasecmp (argv[i], "QosLevel", + } else if (!strncasecmp (argv[i], "QOSLevel", MAX(command_len, 1))) { if(!assoc->qos_list) assoc->qos_list = @@ -567,6 +569,12 @@ field->name = xstrdup("RPC"); field->len = 3; field->print_routine = print_fields_uint; + } else if(!strncasecmp("Shares", object, + MAX(command_len, 1))) { + field->type = PRINT_FAIRSHARE; + field->name = xstrdup("Shares"); + field->len = 9; + field->print_routine = print_fields_uint; } else { exit_code=1; fprintf(stderr, "Unknown field '%s'\n", object); diff -Nru slurm-llnl-1.3.13/src/sacctmgr/file_functions.c slurm-llnl-1.3.15/src/sacctmgr/file_functions.c --- slurm-llnl-1.3.13/src/sacctmgr/file_functions.c 2008-12-24 21:31:19.000000000 +0100 +++ slurm-llnl-1.3.15/src/sacctmgr/file_functions.c 2009-03-10 00:07:16.000000000 +0100 @@ -336,6 +336,8 @@ MAX(command_len, 3))) { file_opts->desc = xstrdup(option); } else if (!strncasecmp (sub, "FairShare", + MAX(command_len, 1)) + || !strncasecmp (sub, "Shares", MAX(command_len, 1))) { if (get_uint(option, &file_opts->fairshare, "FairShare") != SLURM_SUCCESS) { @@ -584,7 +586,9 @@ field->len = 20; field->print_routine = print_fields_str; } else if(!strncasecmp("FairShare", object, - MAX(command_len, 1))) { + MAX(command_len, 1)) + || !strncasecmp("Shares", object, + MAX(command_len, 1))) { field->type = PRINT_FAIRSHARE; field->name = xstrdup("FairShare"); field->len = 9; @@ -1119,7 +1123,7 @@ user->wckey_list = list_create(destroy_acct_wckey_rec); wckey_itr = list_iterator_create(file_opts->wckey_list); - printf(" Adding WCKey(s) '"); + printf(" Adding WCKey(s) "); while((temp_char = list_next(wckey_itr))) { wckey = xmalloc(sizeof(acct_wckey_rec_t)); wckey->name = xstrdup(temp_char); @@ -1128,13 +1132,13 @@ list_push(user->wckey_list, wckey); if(first) { - printf(" %s", temp_char); + printf("'%s'", temp_char); first = 0; } else - printf(", %s", temp_char); + printf(", '%s'", temp_char); } list_iterator_destroy(wckey_itr); - printf("' for user '%s'\n", user->name); + printf(" for user '%s'\n", user->name); set = 1; notice_thread_init(); rc = acct_storage_g_add_wckeys(db_conn, my_uid, @@ -1587,7 +1591,7 @@ acct->organization = xstrdup(parent); else acct->organization = xstrdup(file_opts->name); - /* info("adding acct %s (%s) (%s)", */ + /* info("adding account %s (%s) (%s)", */ /* acct->name, acct->description, */ /* acct->organization); */ @@ -1686,7 +1690,9 @@ if(user_rec) { xstrfmtcat(line, ":DefaultAccount='%s'", user_rec->default_acct); - if(track_wckey) + if(track_wckey + && user_rec->default_wckey + && user_rec->default_wckey[0]) xstrfmtcat(line, ":DefaultWCKey='%s'", user_rec->default_wckey); diff -Nru slurm-llnl-1.3.13/src/sacctmgr/sacctmgr.c slurm-llnl-1.3.15/src/sacctmgr/sacctmgr.c --- slurm-llnl-1.3.13/src/sacctmgr/sacctmgr.c 2008-12-23 00:56:27.000000000 +0100 +++ slurm-llnl-1.3.15/src/sacctmgr/sacctmgr.c 2009-03-02 22:37:40.000000000 +0100 @@ -489,7 +489,8 @@ acct_storage_g_commit(db_conn, 0); /* First identify the entity to add */ - if (strncasecmp (argv[0], "Account", MAX(command_len, 1)) == 0) { + if (strncasecmp (argv[0], "Account", MAX(command_len, 1)) == 0 + || !strncasecmp (argv[0], "Acct", MAX(command_len, 4))) { error_code = sacctmgr_add_account((argc - 1), &argv[1]); } else if (strncasecmp (argv[0], "Cluster", MAX(command_len, 2)) == 0) { error_code = sacctmgr_add_cluster((argc - 1), &argv[1]); @@ -576,7 +577,8 @@ acct_storage_g_commit(db_conn, 0); /* First identify the entity to list */ - if (strncasecmp (argv[0], "Accounts", MAX(command_len, 2)) == 0) { + if (strncasecmp (argv[0], "Accounts", MAX(command_len, 2)) == 0 + || !strncasecmp (argv[0], "Acct", MAX(command_len, 4))) { error_code = sacctmgr_list_account((argc - 1), &argv[1]); } else if (strncasecmp (argv[0], "Associations", MAX(command_len, 2)) == 0) { @@ -633,7 +635,8 @@ acct_storage_g_commit(db_conn, 0); /* First identify the entity to modify */ - if (strncasecmp (argv[0], "Accounts", MAX(command_len, 1)) == 0) { + if (strncasecmp (argv[0], "Accounts", MAX(command_len, 1)) == 0 + || !strncasecmp (argv[0], "Acct", MAX(command_len, 4))) { error_code = sacctmgr_modify_account((argc - 1), &argv[1]); } else if (strncasecmp (argv[0], "Clusters", MAX(command_len, 1)) == 0) { @@ -677,7 +680,8 @@ acct_storage_g_commit(db_conn, 0); /* First identify the entity to delete */ - if (strncasecmp (argv[0], "Accounts", MAX(command_len, 1)) == 0) { + if (strncasecmp (argv[0], "Accounts", MAX(command_len, 1)) == 0 + || !strncasecmp (argv[0], "Acct", MAX(command_len, 4))) { error_code = sacctmgr_delete_account((argc - 1), &argv[1]); } else if (strncasecmp (argv[0], "Clusters", MAX(command_len, 2)) == 0) { diff -Nru slurm-llnl-1.3.13/src/sacctmgr/user_functions.c slurm-llnl-1.3.15/src/sacctmgr/user_functions.c --- slurm-llnl-1.3.13/src/sacctmgr/user_functions.c 2008-12-23 00:56:27.000000000 +0100 +++ slurm-llnl-1.3.15/src/sacctmgr/user_functions.c 2009-03-10 00:07:16.000000000 +0100 @@ -112,7 +112,9 @@ argv[i]+end)) u_set = 1; } else if (!strncasecmp (argv[i], "Account", - MAX(command_len, 2))) { + MAX(command_len, 2)) + || !strncasecmp (argv[i], "Acct", + MAX(command_len, 4))) { if(!assoc_cond->acct_list) { assoc_cond->acct_list = list_create(slurm_destroy_char); @@ -157,6 +159,8 @@ if(format_list) slurm_addto_char_list(format_list, argv[i]+end); } else if (!strncasecmp (argv[i], "FairShare", + MAX(command_len, 1)) + || !strncasecmp (argv[i], "Shares", MAX(command_len, 1))) { if(!assoc_cond->fairshare_list) assoc_cond->fairshare_list = @@ -370,6 +374,8 @@ strip_quotes(argv[i]+end, NULL, 1); u_set = 1; } else if (!strncasecmp (argv[i], "FairShare", + MAX(command_len, 1)) + || !strncasecmp (argv[i], "Shares", MAX(command_len, 1))) { if(!assoc) continue; @@ -518,7 +524,7 @@ /* * IN: user_cond - used for the assoc_cond pointing to the user and - * acct list + * account list * IN: check - whether or not to check if the existance of the above lists */ static int _check_coord_request(acct_user_cond_t *user_cond, bool check) @@ -704,7 +710,9 @@ slurm_addto_char_list(assoc_cond->user_list, argv[i]+end); } else if (!strncasecmp (argv[i], "Accounts", - MAX(command_len, 2))) { + MAX(command_len, 2)) + || !strncasecmp (argv[i], "Acct", + MAX(command_len, 4))) { slurm_addto_char_list(assoc_cond->acct_list, argv[i]+end); } else if (!strncasecmp (argv[i], "AdminLevel", @@ -739,6 +747,8 @@ slurm_addto_char_list(wckey_cond->name_list, default_wckey); } else if (!strncasecmp (argv[i], "FairShare", + MAX(command_len, 1)) + || !strncasecmp (argv[i], "Shares", MAX(command_len, 1))) { if (get_uint(argv[i]+end, &start_assoc.fairshare, "FairShare") == SLURM_SUCCESS) @@ -967,7 +977,7 @@ destroy_acct_wckey_cond(wckey_cond); destroy_acct_association_cond(assoc_cond); exit_code=1; - fprintf(stderr, " Need name of acct to " + fprintf(stderr, " Need name of account to " "add user to.\n"); return SLURM_ERROR; } @@ -1538,7 +1548,8 @@ command_len = strlen(object); field = xmalloc(sizeof(print_field_t)); - if(!strncasecmp("Account", object, MAX(command_len, 2))) { + if(!strncasecmp("Account", object, MAX(command_len, 2)) + || !strncasecmp ("Acct", object, MAX(command_len, 4))) { field->type = PRINT_ACCOUNT; field->name = xstrdup("Account"); field->len = 10; @@ -1680,6 +1691,12 @@ field->name = xstrdup("Partition"); field->len = 10; field->print_routine = print_fields_str; + } else if(!strncasecmp("Shares", object, + MAX(command_len, 1))) { + field->type = PRINT_FAIRSHARE; + field->name = xstrdup("Shares"); + field->len = 9; + field->print_routine = print_fields_uint; } else if(!strncasecmp("User", object, MAX(command_len, 1)) || !strncasecmp("Name", object, MAX(command_len, 2))) { diff -Nru slurm-llnl-1.3.13/src/salloc/opt.c slurm-llnl-1.3.15/src/salloc/opt.c --- slurm-llnl-1.3.13/src/salloc/opt.c 2008-12-04 22:26:47.000000000 +0100 +++ slurm-llnl-1.3.15/src/salloc/opt.c 2009-03-17 19:50:45.000000000 +0100 @@ -546,8 +546,10 @@ {"reboot", no_argument, 0, LONG_OPT_REBOOT}, {"blrts-image", required_argument, 0, LONG_OPT_BLRTS_IMAGE}, {"linux-image", required_argument, 0, LONG_OPT_LINUX_IMAGE}, + {"cnload-image", required_argument, 0, LONG_OPT_LINUX_IMAGE}, {"mloader-image", required_argument, 0, LONG_OPT_MLOADER_IMAGE}, {"ramdisk-image", required_argument, 0, LONG_OPT_RAMDISK_IMAGE}, + {"ioload-image", required_argument, 0, LONG_OPT_RAMDISK_IMAGE}, {"acctg-freq", required_argument, 0, LONG_OPT_ACCTG_FREQ}, {"no-shell", no_argument, 0, LONG_OPT_NOSHELL}, {"get-user-env", optional_argument, 0, LONG_OPT_GET_USER_ENV}, @@ -1352,14 +1354,25 @@ xfree(str); info("reboot : %s", opt.reboot ? "no" : "yes"); info("rotate : %s", opt.no_rotate ? "yes" : "no"); +#ifdef HAVE_BGL if (opt.blrtsimage) info("BlrtsImage : %s", opt.blrtsimage); +#endif if (opt.linuximage) +#ifdef HAVE_BGL info("LinuxImage : %s", opt.linuximage); +#else + info("CnloadImage : %s", opt.linuximage); +#endif if (opt.mloaderimage) info("MloaderImage : %s", opt.mloaderimage); if (opt.ramdiskimage) +#ifdef HAVE_BGL info("RamDiskImage : %s", opt.ramdiskimage); +#else + info("IoloadImage : %s", opt.ramdiskimage); +#endif + if (opt.begin) { char time_str[32]; slurm_make_time_str(&opt.begin, time_str, sizeof(time_str)); @@ -1396,8 +1409,13 @@ " [--account=name] [--dependency=type:jobid] [--comment=name]\n" #ifdef HAVE_BG /* Blue gene specific options */ " [--geometry=XxYxZ] [--conn-type=type] [--no-rotate] [ --reboot]\n" +#ifdef HAVE_BGL " [--blrts-image=path] [--linux-image=path]\n" " [--mloader-image=path] [--ramdisk-image=path]\n" +#else +" [--cnload-image=path]\n" +" [--mloader-image=path] [--ioload-image=path]\n" +#endif #endif " [--mail-type=type] [--mail-user=user][--nice[=value]]\n" " [--bell] [--no-bell] [--kill-command[=signal]]\n" @@ -1490,24 +1508,34 @@ printf("\n" #ifdef HAVE_AIX /* AIX/Federation specific options */ - "AIX related options:\n" - " --network=type communication protocol to be used\n" - "\n" +"AIX related options:\n" +" --network=type communication protocol to be used\n" +"\n" #endif #ifdef HAVE_BG /* Blue gene specific options */ "\n" - "Blue Gene related options:\n" - " -g, --geometry=XxYxZ geometry constraints of the job\n" - " -R, --no-rotate disable geometry rotation\n" - " --reboot reboot nodes before starting job\n" - " --conn-type=type constraint on type of connection, MESH or TORUS\n" - " if not set, then tries to fit TORUS else MESH\n" - " --blrts-image=path path to blrts image for bluegene block. Default if not set\n" - " --linux-image=path path to linux image for bluegene block. Default if not set\n" - " --mloader-image=path path to mloader image for bluegene block. Default if not set\n" - " --ramdisk-image=path path to ramdisk image for bluegene block. Default if not set\n" - "\n" +"Blue Gene related options:\n" +" -g, --geometry=XxYxZ geometry constraints of the job\n" +" -R, --no-rotate disable geometry rotation\n" +" --reboot reboot nodes before starting job\n" +" --conn-type=type constraint on type of connection, MESH or TORUS\n" +" if not set, then tries to fit TORUS else MESH\n" +#ifndef HAVE_BGL +" If wanting to run in HTC mode (only for 1\n" +" midplane and below). You can use HTC_S for\n" +" SMP, HTC_D for Dual, HTC_V for\n" +" virtual node mode, and HTC_L for Linux mode.\n" +" --cnload-image=path path to compute node image for bluegene block. Default if not set\n" +" --mloader-image=path path to mloader image for bluegene block. Default if not set\n" +" --ioload-image=path path to ioload image for bluegene block. Default if not set\n" +#else +" --blrts-image=path path to blrts image for bluegene block. Default if not set\n" +" --linux-image=path path to linux image for bluegene block. Default if not set\n" +" --mloader-image=path path to mloader image for bluegene block. Default if not set\n" +" --ramdisk-image=path path to ramdisk image for bluegene block. Default if not set\n" #endif +#endif +"\n" "Help options:\n" " -h, --help show this help message\n" " -u, --usage display brief usage message\n" diff -Nru slurm-llnl-1.3.13/src/salloc/salloc.c slurm-llnl-1.3.15/src/salloc/salloc.c --- slurm-llnl-1.3.13/src/salloc/salloc.c 2008-12-03 00:38:47.000000000 +0100 +++ slurm-llnl-1.3.15/src/salloc/salloc.c 2009-03-10 22:55:42.000000000 +0100 @@ -209,7 +209,8 @@ info("Granted job allocation %d", alloc->job_id); #ifdef HAVE_BG if (!_wait_bluegene_block_ready(alloc)) { - error("Something is wrong with the boot of the block."); + if(!allocation_interrupted) + error("Something is wrong with the boot of the block."); goto relinquish; } @@ -614,6 +615,7 @@ int max_delay = BG_FREE_PREVIOUS_BLOCK + BG_MIN_BLOCK_BOOT + (BG_INCR_BLOCK_BOOT * alloc->node_cnt); + pending_job_id = alloc->job_id; select_g_get_jobinfo(alloc->select_jobinfo, SELECT_DATA_BLOCK_ID, &block_id); @@ -642,12 +644,16 @@ break; } } - if (is_ready) info("Block %s is ready for job", block_id); - else + else if(!allocation_interrupted) error("Block %s still not ready", block_id); + else /* this should never happen, but if allocation_intrrupted + send back not ready */ + is_ready = 0; + xfree(block_id); + pending_job_id = 0; return is_ready; } diff -Nru slurm-llnl-1.3.13/src/sbatch/opt.c slurm-llnl-1.3.15/src/sbatch/opt.c --- slurm-llnl-1.3.13/src/sbatch/opt.c 2008-12-03 00:38:47.000000000 +0100 +++ slurm-llnl-1.3.15/src/sbatch/opt.c 2009-03-17 19:50:45.000000000 +0100 @@ -331,6 +331,7 @@ {"SBATCH_JOBID", OPT_INT, &opt.jobid, NULL }, {"SBATCH_JOB_NAME", OPT_STRING, &opt.job_name, NULL }, {"SBATCH_LINUX_IMAGE", OPT_STRING, &opt.linuximage, NULL }, + {"SBATCH_CNLOAD_IMAGE", OPT_STRING, &opt.linuximage, NULL }, {"SBATCH_MLOADER_IMAGE", OPT_STRING, &opt.mloaderimage, NULL }, {"SBATCH_NO_REQUEUE", OPT_NO_REQUEUE, NULL, NULL }, {"SBATCH_REQUEUE", OPT_REQUEUE, NULL, NULL }, @@ -338,6 +339,7 @@ {"SBATCH_OVERCOMMIT", OPT_OVERCOMMIT, NULL, NULL }, {"SBATCH_PARTITION", OPT_STRING, &opt.partition, NULL }, {"SBATCH_RAMDISK_IMAGE", OPT_STRING, &opt.ramdiskimage, NULL }, + {"SBATCH_IOLOAD_IMAGE", OPT_STRING, &opt.ramdiskimage, NULL }, {"SBATCH_TIMELIMIT", OPT_STRING, &opt.time_limit_str,NULL }, {"SBATCH_EXCLUSIVE", OPT_EXCLUSIVE, NULL, NULL }, {"SBATCH_OPEN_MODE", OPT_OPEN_MODE, NULL, NULL }, @@ -553,8 +555,10 @@ {"ntasks-per-core", required_argument, 0, LONG_OPT_NTASKSPERCORE}, {"blrts-image", required_argument, 0, LONG_OPT_BLRTS_IMAGE}, {"linux-image", required_argument, 0, LONG_OPT_LINUX_IMAGE}, + {"cnload-image", required_argument, 0, LONG_OPT_LINUX_IMAGE}, {"mloader-image", required_argument, 0, LONG_OPT_MLOADER_IMAGE}, {"ramdisk-image", required_argument, 0, LONG_OPT_RAMDISK_IMAGE}, + {"ioload-image", required_argument, 0, LONG_OPT_RAMDISK_IMAGE}, {"reboot", no_argument, 0, LONG_OPT_REBOOT}, {"tasks-per-node",required_argument, 0, LONG_OPT_NTASKSPERNODE}, {"wrap", required_argument, 0, LONG_OPT_WRAP}, @@ -2095,15 +2099,24 @@ info("rotate : %s", opt.no_rotate ? "yes" : "no"); info("network : %s", opt.network); +#ifdef HAVE_BGL if (opt.blrtsimage) info("BlrtsImage : %s", opt.blrtsimage); +#endif if (opt.linuximage) +#ifdef HAVE_BGL info("LinuxImage : %s", opt.linuximage); +#else + info("CnloadImage : %s", opt.linuximage); +#endif if (opt.mloaderimage) info("MloaderImage : %s", opt.mloaderimage); if (opt.ramdiskimage) +#ifdef HAVE_BGL info("RamDiskImage : %s", opt.ramdiskimage); - +#else + info("IoloadImage : %s", opt.ramdiskimage); +#endif if (opt.begin) { char time_str[32]; slurm_make_time_str(&opt.begin, time_str, sizeof(time_str)); @@ -2143,8 +2156,13 @@ " [--account=name] [--dependency=type:jobid] [--comment=name]\n" #ifdef HAVE_BG /* Blue gene specific options */ " [--geometry=XxYxZ] [--conn-type=type] [--no-rotate] [ --reboot]\n" +#ifdef HAVE_BGL " [--blrts-image=path] [--linux-image=path]\n" " [--mloader-image=path] [--ramdisk-image=path]\n" +#else +" [--cnload-image=path]\n" +" [--mloader-image=path] [--ioload-image=path]\n" +#endif #endif " [--mail-type=type] [--mail-user=user][--nice[=value]]\n" " [--requeue] [--no-requeue] [--ntasks-per-node=n] [--propagate]\n" @@ -2249,11 +2267,21 @@ " --reboot reboot block before starting job\n" " --conn-type=type constraint on type of connection, MESH or TORUS\n" " if not set, then tries to fit TORUS else MESH\n" +#ifndef HAVE_BGL +" If wanting to run in HTC mode (only for 1\n" +" midplane and below). You can use HTC_S for\n" +" SMP, HTC_D for Dual, HTC_V for\n" +" virtual node mode, and HTC_L for Linux mode.\n" +" --cnload-image=path path to compute node image for bluegene block. Default if not set\n" +" --mloader-image=path path to mloader image for bluegene block. Default if not set\n" +" --ioload-image=path path to ioload image for bluegene block. Default if not set\n" +#else " --blrts-image=path path to blrts image for bluegene block. Default if not set\n" " --linux-image=path path to linux image for bluegene block. Default if not set\n" " --mloader-image=path path to mloader image for bluegene block. Default if not set\n" " --ramdisk-image=path path to ramdisk image for bluegene block. Default if not set\n" #endif +#endif "\n" "Help options:\n" " -h, --help show this help message\n" diff -Nru slurm-llnl-1.3.13/src/sinfo/opts.c slurm-llnl-1.3.15/src/sinfo/opts.c --- slurm-llnl-1.3.13/src/sinfo/opts.c 2008-08-28 01:57:13.000000000 +0200 +++ slurm-llnl-1.3.15/src/sinfo/opts.c 2009-02-23 18:46:41.000000000 +0100 @@ -246,7 +246,7 @@ } else { params.format = params.long_output ? "%9P %.5a %.10l %.10s %.4r %.5h %.10g %.6D %.11T %N" : - "%9P %.5a %.10l %.5D %.6t %N"; + "%9P %.5a %.10l %.6D %.6t %N"; } } _parse_format( params.format ); diff -Nru slurm-llnl-1.3.13/src/slurmctld/acct_policy.c slurm-llnl-1.3.15/src/slurmctld/acct_policy.c --- slurm-llnl-1.3.13/src/slurmctld/acct_policy.c 2008-11-19 00:02:04.000000000 +0100 +++ slurm-llnl-1.3.15/src/slurmctld/acct_policy.c 2009-03-02 22:37:40.000000000 +0100 @@ -101,7 +101,7 @@ { acct_association_rec_t *assoc_ptr = NULL; - if (accounting_enforce != ACCOUNTING_ENFORCE_WITH_LIMITS + if (!(accounting_enforce & ACCOUNTING_ENFORCE_LIMITS) || !_valid_job_assoc(job_ptr)) return; @@ -125,7 +125,7 @@ acct_association_rec_t *assoc_ptr = NULL; if (!job_ptr->assoc_ptr || - accounting_enforce != ACCOUNTING_ENFORCE_WITH_LIMITS) + !(accounting_enforce & ACCOUNTING_ENFORCE_LIMITS)) return; slurm_mutex_lock(&assoc_mgr_association_lock); @@ -150,7 +150,7 @@ { acct_association_rec_t *assoc_ptr = NULL; - if (accounting_enforce != ACCOUNTING_ENFORCE_WITH_LIMITS + if (!(accounting_enforce & ACCOUNTING_ENFORCE_LIMITS) || !_valid_job_assoc(job_ptr)) return; @@ -174,7 +174,7 @@ { acct_association_rec_t *assoc_ptr = NULL; - if (accounting_enforce != ACCOUNTING_ENFORCE_WITH_LIMITS + if (!(accounting_enforce & ACCOUNTING_ENFORCE_LIMITS) || !job_ptr->assoc_ptr) return; @@ -231,7 +231,7 @@ } /* now see if we are enforcing limits */ - if (accounting_enforce != ACCOUNTING_ENFORCE_WITH_LIMITS) + if (!(accounting_enforce & ACCOUNTING_ENFORCE_LIMITS)) return true; /* clear old state reason */ @@ -301,7 +301,7 @@ /* we don't need to check submit_jobs here */ /* FIX ME: Once we start tracking time of running jobs - * we will need toupdate the amount of time we have + * we will need to update the amount of time we have * used and check against that here. When we start * keeping track of time we will also need to come up * with a way to refresh the time. diff -Nru slurm-llnl-1.3.13/src/slurmctld/controller.c slurm-llnl-1.3.15/src/slurmctld/controller.c --- slurm-llnl-1.3.13/src/slurmctld/controller.c 2009-01-06 17:21:17.000000000 +0100 +++ slurm-llnl-1.3.15/src/slurmctld/controller.c 2009-04-07 20:09:43.000000000 +0200 @@ -201,7 +201,8 @@ slurmctld_lock_t config_write_lock = { WRITE_LOCK, WRITE_LOCK, WRITE_LOCK, WRITE_LOCK }; assoc_init_args_t assoc_init_arg; - pthread_t assoc_cache_thread; + pthread_t assoc_cache_thread = 0; + gid_t slurm_user_gid; /* * Establish initial configuration @@ -224,19 +225,43 @@ */ _init_pidfile(); - /* Initialize supplementary group ID list for SlurmUser */ - if ((getuid() == 0) - && (slurmctld_conf.slurm_user_id != getuid()) - && initgroups(slurmctld_conf.slurm_user_name, - gid_from_string(slurmctld_conf.slurm_user_name))) { - error("initgroups: %m"); + /* Determine SlurmUser gid */ + slurm_user_gid = gid_from_uid(slurmctld_conf.slurm_user_id); + if (slurm_user_gid == (gid_t) -1) { + fatal("Failed to determine gid of SlurmUser(%d)", + slurm_user_gid); + } + + /* Initialize supplementary groups ID list for SlurmUser */ + if (getuid() == 0) { + /* root does not need supplementary groups */ + if ((slurmctld_conf.slurm_user_id == 0) && + (setgroups(0, NULL) != 0)) { + fatal("Failed to drop supplementary groups, " + "setgroups: %m"); + } else if ((slurmctld_conf.slurm_user_id != getuid()) && + initgroups(slurmctld_conf.slurm_user_name, + slurm_user_gid)) { + fatal("Failed to set supplementary groups, " + "initgroups: %m"); + } + } else { + info("Not running as root. Can't drop supplementary groups"); } - if ((slurmctld_conf.slurm_user_id != getuid()) - && (setuid(slurmctld_conf.slurm_user_id))) { + /* Set GID to GID of SlurmUser */ + if ((slurm_user_gid != getegid()) && + (setgid(slurm_user_gid))) { + fatal("Failed to set GID to %d", slurm_user_gid); + } + + /* Set UID to UID of SlurmUser */ + if ((slurmctld_conf.slurm_user_id != getuid()) && + (setuid(slurmctld_conf.slurm_user_id))) { fatal("Can not set uid to SlurmUser(%d): %m", - slurmctld_conf.slurm_user_id); + slurmctld_conf.slurm_user_id); } + if (stat(slurmctld_conf.mail_prog, &stat_buf) != 0) error("Configured MailProg is invalid"); @@ -308,6 +333,21 @@ association_based_accounting = slurm_get_is_association_based_accounting(); accounting_enforce = slurmctld_conf.accounting_storage_enforce; + + if(accounting_enforce && !association_based_accounting) { + slurm_ctl_conf_t *conf = slurm_conf_lock(); + conf->track_wckey = false; + conf->accounting_storage_enforce = 0; + accounting_enforce = 0; + slurmctld_conf.track_wckey = false; + slurmctld_conf.accounting_storage_enforce = 0; + slurm_conf_unlock(); + + error("You can not have AccountingStorageEnforce " + "set for AccountingStorageType='%s'", + slurmctld_conf.accounting_storage_type); + } + acct_db_conn = acct_storage_g_get_connection(true, 0, false); memset(&assoc_init_arg, 0, sizeof(assoc_init_args_t)); @@ -317,7 +357,7 @@ ASSOC_MGR_CACHE_USER | ASSOC_MGR_CACHE_QOS; if (assoc_mgr_init(acct_db_conn, &assoc_init_arg)) { - if(accounting_enforce) + if(accounting_enforce & ACCOUNTING_ENFORCE_ASSOCS) error("Association database appears down, " "reading from state file."); else @@ -325,7 +365,8 @@ "reading from state file."); if ((load_assoc_mgr_state(slurmctld_conf.state_save_location) - != SLURM_SUCCESS) && accounting_enforce) { + != SLURM_SUCCESS) + && (accounting_enforce & ACCOUNTING_ENFORCE_ASSOCS)) { error("Unable to get any information from " "the state file"); fatal("slurmdbd and/or database must be up at " @@ -405,16 +446,9 @@ } unlock_slurmctld(config_write_lock); - if ((recover == 0) || - (!stat("/tmp/slurm_accounting_first", &stat_buf))) { - /* When first starting to write node state - * information to Gold or SlurmDBD, create - * a file called "/tmp/slurm_accounting_first" - * to capture node initialization information */ - + if (recover == 0) _accounting_mark_all_nodes_down("cold-start"); - unlink("/tmp/slurm_accounting_first"); - } + } else { error("this host (%s) not valid controller (%s or %s)", node_name, slurmctld_conf.control_machine, @@ -431,7 +465,8 @@ NULL will just use those set before. */ if (assoc_mgr_init(acct_db_conn, NULL) && - accounting_enforce && !running_cache) { + (accounting_enforce & ACCOUNTING_ENFORCE_ASSOCS) + && !running_cache) { error("assoc_mgr_init failure"); fatal("slurmdbd and/or database must be up at " "slurmctld start time"); @@ -546,6 +581,8 @@ if (i >= 10) error("Left %d agent threads active", cnt); + slurm_sched_fini(); + /* Purge our local data structures */ job_fini(); part_fini(); /* part_fini() must preceed node_fini() */ @@ -558,7 +595,6 @@ g_slurm_jobcomp_fini(); slurm_acct_storage_fini(); slurm_jobacct_gather_fini(); - slurm_sched_fini(); slurm_select_fini(); checkpoint_fini(); slurm_auth_fini(); @@ -964,8 +1000,7 @@ /* see if we are running directly to a database * instead of a slurmdbd. */ - send_jobs_to_accounting(event_time); - send_nodes_to_accounting(event_time); + send_all_to_accounting(event_time); rc = SLURM_SUCCESS; } @@ -1016,7 +1051,7 @@ { int cnt = 0; - if (accounting_enforce) + if (accounting_enforce & ACCOUNTING_ENFORCE_ASSOCS) cnt = job_cancel_by_assoc_id(rec->id); if (cnt) { @@ -1269,6 +1304,18 @@ dump_assoc_mgr_state(slurmctld_conf.state_save_location); } +/* send all info for the controller to accounting */ +extern void send_all_to_accounting(time_t event_time) +{ + /* ignore the rcs here because if there was an error we will + push the requests on the queue and process them when the + database server comes back up. + */ + debug2("send_all_to_accounting: called"); + send_jobs_to_accounting(); + send_nodes_to_accounting(event_time); +} + /* * _report_locks_set - report any slurmctld locks left set * RET count of locks currently set @@ -1423,7 +1470,8 @@ /* * Tell the backup_controller to relinquish control, primary control_machine * has resumed operation - * wait_time - How long to wait for backup controller to write state, seconds + * wait_time - How long to wait for backup controller to write state, seconds. + * Must be zero when called from _slurmctld_background() loop. * RET 0 or an error code * NOTE: READ lock_slurmctld config before entry (or be single-threaded) */ @@ -1452,9 +1500,17 @@ } if (rc == ESLURM_DISABLED) debug("backup controller responding"); - else if (rc == 0) + else if (rc == 0) { debug("backup controller has relinquished control"); - else { + if (wait_time == 0) { + /* In case primary controller really did not terminate, + * but just temporarily became non-responsive */ + clusteracct_storage_g_register_ctld( + acct_db_conn, + slurmctld_cluster_name, + slurmctld_conf.slurmctld_port); + } + } else { error("_shutdown_backup_controller: %s", slurm_strerror(rc)); return SLURM_ERROR; } @@ -1587,8 +1643,9 @@ ListIterator itr = NULL; struct job_record *job_ptr = NULL; acct_association_rec_t assoc_rec; + /* Write lock on jobs, read lock on nodes and partitions */ slurmctld_lock_t job_write_lock = - { READ_LOCK, WRITE_LOCK, WRITE_LOCK, READ_LOCK }; + { NO_LOCK, WRITE_LOCK, READ_LOCK, READ_LOCK }; while(running_cache == 1) { slurm_mutex_lock(&assoc_cache_mutex); @@ -1636,5 +1693,8 @@ } list_iterator_destroy(itr); unlock_slurmctld(job_write_lock); + /* This needs to be after the lock and after we update the + jobs so if we need to send them we are set. */ + _accounting_cluster_ready(); return NULL; } diff -Nru slurm-llnl-1.3.13/src/slurmctld/job_mgr.c slurm-llnl-1.3.15/src/slurmctld/job_mgr.c --- slurm-llnl-1.3.13/src/slurmctld/job_mgr.c 2009-01-12 21:33:23.000000000 +0100 +++ slurm-llnl-1.3.15/src/slurmctld/job_mgr.c 2009-03-25 00:21:14.000000000 +0100 @@ -757,7 +757,8 @@ accounting_enforce, (acct_association_rec_t **) &job_ptr->assoc_ptr) && - accounting_enforce && (!IS_JOB_FINISHED(job_ptr))) { + (accounting_enforce & ACCOUNTING_ENFORCE_ASSOCS) + && (!IS_JOB_FINISHED(job_ptr))) { info("Cancelling job %u with invalid association", job_id); job_ptr->job_state = JOB_CANCELLED; @@ -2045,7 +2046,8 @@ error_code = ESLURM_INVALID_ACCOUNT; return error_code; } else if(association_based_accounting - && !assoc_ptr && !accounting_enforce) { + && !assoc_ptr + && !(accounting_enforce & ACCOUNTING_ENFORCE_ASSOCS)) { /* if not enforcing associations we want to look for the default account and use it to avoid getting trash in the accounting records. @@ -2063,7 +2065,7 @@ } if (job_desc->account == NULL) job_desc->account = xstrdup(assoc_rec.acct); - if ((accounting_enforce == ACCOUNTING_ENFORCE_WITH_LIMITS) && + if ((accounting_enforce & ACCOUNTING_ENFORCE_LIMITS) && (!_validate_acct_policy(job_desc, part_ptr, &assoc_rec))) { info("_job_create: exceeded association's node or time limit " "for user %u", job_desc->user_id); @@ -2229,7 +2231,8 @@ part_ptr, &req_bitmap, &exc_bitmap))) { - error_code = ESLURM_ERROR_ON_DESC_TO_RECORD_COPY; + if(error_code == SLURM_ERROR) + error_code = ESLURM_ERROR_ON_DESC_TO_RECORD_COPY; goto cleanup_fail; } @@ -2795,6 +2798,52 @@ struct job_details *detail_ptr; struct job_record *job_ptr; + if(slurm_get_track_wckey()) { + char *wckey = NULL; + if(!job_desc->name || !strchr(job_desc->name, '\"')) { + /* get the default wckey for this user since none was + * given */ + acct_user_rec_t user_rec; + memset(&user_rec, 0, sizeof(acct_user_rec_t)); + user_rec.uid = job_desc->user_id; + assoc_mgr_fill_in_user(acct_db_conn, &user_rec, + accounting_enforce); + if(user_rec.default_wckey) + xstrfmtcat(job_desc->name, "\"*%s", + user_rec.default_wckey); + else if(!(accounting_enforce + & ACCOUNTING_ENFORCE_WCKEYS)) + xstrcat(job_desc->name, "\"*"); + else { + error("Job didn't specify wckey and user " + "%d has no default.", job_desc->user_id); + return ESLURM_INVALID_WCKEY; + } + } else if(job_desc->name + && (wckey = strchr(job_desc->name, '\"')) + && (accounting_enforce & ACCOUNTING_ENFORCE_WCKEYS)) { + acct_wckey_rec_t wckey_rec, *wckey_ptr = NULL; + wckey++; + + memset(&wckey_rec, 0, sizeof(acct_wckey_rec_t)); + wckey_rec.uid = job_desc->user_id; + wckey_rec.name = wckey; + + if (assoc_mgr_fill_in_wckey(acct_db_conn, &wckey_rec, + accounting_enforce, + &wckey_ptr)) { + info("_job_create: invalid wckey '%s' " + "for user %u.", + wckey_rec.name, job_desc->user_id); + return ESLURM_INVALID_WCKEY; + } + } else if (accounting_enforce & ACCOUNTING_ENFORCE_WCKEYS) { + /* This should never happen */ + info("_job_create: no wckey was given."); + return ESLURM_INVALID_WCKEY; + } + } + job_ptr = create_job_record(&error_code); if (error_code) return error_code; @@ -2805,26 +2854,11 @@ job_ptr->job_id = job_desc->job_id; else _set_job_id(job_ptr); - _add_job_hash(job_ptr); if (job_desc->name) - job_ptr->name = xstrdup(job_desc->name); - - if(slurm_get_track_wckey() - && (!job_ptr->name || !strchr(job_ptr->name, '\"'))) { - /* get the default wckey for this user since none was - * given */ - acct_user_rec_t user_rec; - memset(&user_rec, 0, sizeof(acct_user_rec_t)); - user_rec.uid = job_desc->user_id; - assoc_mgr_fill_in_user(acct_db_conn, &user_rec, - accounting_enforce); - if(user_rec.default_wckey) - xstrfmtcat(job_ptr->name, "\"*%s", - user_rec.default_wckey); - else - xstrcat(job_ptr->name, "\"*"); - } + job_ptr->name = xstrdup(job_desc->name); + + _add_job_hash(job_ptr); job_ptr->user_id = (uid_t) job_desc->user_id; job_ptr->group_id = (gid_t) job_desc->group_id; @@ -3925,6 +3959,7 @@ bitstr_t *exc_bitmap = NULL, *req_bitmap = NULL; time_t now = time(NULL); multi_core_data_t *mc_ptr = NULL; + bool update_accounting = false; job_ptr = find_job_record(job_specs->job_id); if (job_ptr == NULL) { @@ -3995,13 +4030,53 @@ } } + if (job_specs->comment && wiki_sched && (!super_user)) { + /* User must use Moab command to change job comment */ + error("Attempt to change comment for job %u", + job_specs->job_id); + error_code = ESLURM_ACCESS_DENIED; +#if 0 + if (wiki_sched && strstr(job_ptr->comment, "QOS:")) { + if (strstr(job_ptr->comment, "FLAGS:PREEMPTOR")) + job_ptr->qos = QOS_EXPEDITE; + else if (strstr(job_ptr->comment, "FLAGS:PREEMPTEE")) + job_ptr->qos = QOS_STANDBY; + else + job_ptr->qos = QOS_NORMAL; +#endif + } else if (job_specs->comment) { + xfree(job_ptr->comment); + job_ptr->comment = job_specs->comment; + job_specs->comment = NULL; /* Nothing left to free */ + info("update_job: setting comment to %s for job_id %u", + job_ptr->comment, job_specs->job_id); + + if (wiki_sched && strstr(job_ptr->comment, "QOS:")) { + if (strstr(job_ptr->comment, "FLAGS:PREEMPTOR")) + job_ptr->qos = QOS_EXPEDITE; + else if (strstr(job_ptr->comment, "FLAGS:PREEMPTEE")) + job_ptr->qos = QOS_STANDBY; + else + job_ptr->qos = QOS_NORMAL; + } + } + + if (job_specs->requeue != (uint16_t) NO_VAL) { + detail_ptr->requeue = job_specs->requeue; + info("update_job: setting requeue to %u for job_id %u", + job_specs->requeue, job_specs->job_id); + } + if (job_specs->priority != NO_VAL) { - if (super_user - || (job_ptr->priority > job_specs->priority)) { + if (!IS_JOB_PENDING(job_ptr) || (detail_ptr == NULL)) + error_code = ESLURM_DISABLED; + else if (super_user + || (job_ptr->priority > job_specs->priority)) { job_ptr->priority = job_specs->priority; info("update_job: setting priority to %u for " "job_id %u", job_ptr->priority, job_specs->job_id); + update_accounting = true; } else { error("Attempt to increase priority for job %u", job_specs->job_id); @@ -4018,6 +4093,7 @@ info("update_job: setting priority to %u for " "job_id %u", job_ptr->priority, job_specs->job_id); + update_accounting = true; } else { error("Attempt to increase priority for job %u", job_specs->job_id); @@ -4124,6 +4200,7 @@ info("update_job: setting num_procs to %u for " "job_id %u", job_specs->num_procs, job_specs->job_id); + update_accounting = true; } else { error("Attempt to increase num_procs for job %u", job_specs->job_id); @@ -4140,6 +4217,7 @@ info("update_job: setting min_nodes to %u for " "job_id %u", job_specs->min_nodes, job_specs->job_id); + update_accounting = true; } else { error("Attempt to increase min_nodes for job %u", job_specs->job_id); @@ -4274,37 +4352,6 @@ } } - if (job_specs->comment && wiki_sched && (!super_user)) { - /* User must use Moab command to change job comment */ - error("Attempt to change comment for job %u", - job_specs->job_id); - error_code = ESLURM_ACCESS_DENIED; -#if 0 - if (wiki_sched && strstr(job_ptr->comment, "QOS:")) { - if (strstr(job_ptr->comment, "FLAGS:PREEMPTOR")) - job_ptr->qos = QOS_EXPEDITE; - else if (strstr(job_ptr->comment, "FLAGS:PREEMPTEE")) - job_ptr->qos = QOS_STANDBY; - else - job_ptr->qos = QOS_NORMAL; -#endif - } else if (job_specs->comment) { - xfree(job_ptr->comment); - job_ptr->comment = job_specs->comment; - job_specs->comment = NULL; /* Nothing left to free */ - info("update_job: setting comment to %s for job_id %u", - job_ptr->comment, job_specs->job_id); - - if (wiki_sched && strstr(job_ptr->comment, "QOS:")) { - if (strstr(job_ptr->comment, "FLAGS:PREEMPTOR")) - job_ptr->qos = QOS_EXPEDITE; - else if (strstr(job_ptr->comment, "FLAGS:PREEMPTEE")) - job_ptr->qos = QOS_STANDBY; - else - job_ptr->qos = QOS_NORMAL; - } - } - if (job_specs->name) { if (!IS_JOB_PENDING(job_ptr)) error_code = ESLURM_DISABLED; @@ -4352,21 +4399,34 @@ xstrfmtcat(job_ptr->name, "%s", jname); xfree(jname); } - + if(wckey) { - xstrfmtcat(job_ptr->name, "\"%s", wckey); + int rc = update_job_wckey("update_job", + job_ptr, + wckey); + if (rc != SLURM_SUCCESS) + error_code = rc; xfree(wckey); } info("update_job: setting name to %s for job_id %u", job_ptr->name, job_specs->job_id); + update_accounting = true; } } - if (job_specs->requeue != (uint16_t) NO_VAL) { - detail_ptr->requeue = job_specs->requeue; - info("update_job: setting requeue to %u for job_id %u", - job_specs->requeue, job_specs->job_id); + if (job_specs->account) { + if (!IS_JOB_PENDING(job_ptr)) + error_code = ESLURM_DISABLED; + else { + int rc = update_job_account("update_job", job_ptr, + job_specs->account); + if (rc != SLURM_SUCCESS) + error_code = rc; + else + update_accounting = true; + + } } if (job_specs->partition) { @@ -4400,6 +4460,7 @@ info("update_job: setting partition to %s for " "job_id %u", job_specs->partition, job_specs->job_id); + update_accounting = true; } else { error("Attempt to change partition for job %u", job_specs->job_id); @@ -4465,13 +4526,6 @@ } } - if (job_specs->account) { - int rc = update_job_account("update_job", job_ptr, - job_specs->account); - if (rc != SLURM_SUCCESS) - error_code = rc; - } - if (job_specs->ntasks_per_node != (uint16_t) NO_VAL) { if ((!IS_JOB_PENDING(job_ptr)) || (detail_ptr == NULL)) error_code = ESLURM_DISABLED; @@ -4502,9 +4556,10 @@ } if (job_specs->begin_time) { - if (IS_JOB_PENDING(job_ptr) && detail_ptr) + if (IS_JOB_PENDING(job_ptr) && detail_ptr) { detail_ptr->begin_time = job_specs->begin_time; - else + update_accounting = true; + } else error_code = ESLURM_DISABLED; } @@ -4682,7 +4737,13 @@ } } #endif - + if(update_accounting) { + if (job_ptr->details && job_ptr->details->begin_time) { + /* Update job record in accounting to reflect changes */ + jobacct_storage_g_job_start( + acct_db_conn, slurmctld_cluster_name, job_ptr); + } + } return error_code; } @@ -6002,6 +6063,8 @@ info("Association deleted, cancelling job %u", job_ptr->job_id); + /* make sure the assoc_mgr_association_lock isn't + locked before this. */ job_signal(job_ptr->job_id, SIGKILL, 0, 0); job_ptr->state_reason = FAIL_BANK_ACCOUNT; cnt++; @@ -6039,7 +6102,8 @@ module, new_account, job_ptr->job_id); return ESLURM_INVALID_ACCOUNT; } else if(association_based_accounting - && !assoc_ptr && !accounting_enforce) { + && !assoc_ptr + && !(accounting_enforce & ACCOUNTING_ENFORCE_ASSOCS)) { /* if not enforcing associations we want to look for the default account and use it to avoid getting trash in the accounting records. @@ -6049,11 +6113,12 @@ accounting_enforce, &assoc_ptr); if(!assoc_ptr) { debug("%s: we didn't have an association for account " - "'%s' and user '%s', and we can't seem to find " + "'%s' and user '%u', and we can't seem to find " "a default one either. Keeping new account " "'%s'. This will produce trash in accounting. " "If this is not what you desire please put " - "AccountStorageEnforce=1 in your slurm.conf " + "AccountStorageEnforce=associations " + "in your slurm.conf " "file.", module, new_account, job_ptr->user_id, new_account); assoc_rec.acct = new_account; @@ -6072,17 +6137,76 @@ job_ptr->assoc_id = assoc_rec.id; job_ptr->assoc_ptr = (void *) assoc_ptr; - if (job_ptr->details && job_ptr->details->begin_time) { - /* Update account associated with the eligible time */ - jobacct_storage_g_job_start( - acct_db_conn, slurmctld_cluster_name, job_ptr); + last_job_update = time(NULL); + + return SLURM_SUCCESS; +} + +/* + * Modify the account associated with a pending job + * IN module - where this is called from + * IN job_ptr - pointer to job which should be modified + * IN new_wckey - desired wckey name + * RET SLURM_SUCCESS or error code + */ +extern int update_job_wckey(char *module, struct job_record *job_ptr, + char *new_wckey) +{ + acct_wckey_rec_t wckey_rec, *wckey_ptr; + + if ((!IS_JOB_PENDING(job_ptr)) || (job_ptr->details == NULL)) { + info("%s: attempt to modify account for non-pending " + "job_id %u", module, job_ptr->job_id); + return ESLURM_DISABLED; } + + memset(&wckey_rec, 0, sizeof(acct_wckey_rec_t)); + wckey_rec.uid = job_ptr->user_id; + wckey_rec.name = new_wckey; + if (assoc_mgr_fill_in_wckey(acct_db_conn, &wckey_rec, + accounting_enforce, &wckey_ptr)) { + info("%s: invalid wckey %s for job_id %u", + module, new_wckey, job_ptr->job_id); + return ESLURM_INVALID_WCKEY; + } else if(association_based_accounting + && !wckey_ptr + && !(accounting_enforce & ACCOUNTING_ENFORCE_WCKEYS)) { + /* if not enforcing associations we want to look for + the default account and use it to avoid getting + trash in the accounting records. + */ + wckey_rec.name = NULL; + assoc_mgr_fill_in_wckey(acct_db_conn, &wckey_rec, + accounting_enforce, &wckey_ptr); + if(!wckey_ptr) { + debug("%s: we didn't have a wckey record for wckey " + "'%s' and user '%u', and we can't seem to find " + "a default one either. Setting it anyway. " + "This will produce trash in accounting. " + "If this is not what you desire please put " + "AccountStorageEnforce=wckeys in your slurm.conf " + "file.", module, new_wckey, + job_ptr->user_id, new_wckey); + wckey_rec.name = new_wckey; + } + } + + if (wckey_rec.name && wckey_rec.name[0] != '\0') { + xstrfmtcat(job_ptr->name, "\"%s", wckey_rec.name); + job_ptr->account = xstrdup(wckey_rec.name); + info("%s: setting wckey to %s for job_id %u", + module, wckey_rec.name, job_ptr->job_id); + } else { + info("%s: cleared wckey for job_id %u", + module, job_ptr->job_id); + } + last_job_update = time(NULL); return SLURM_SUCCESS; } -extern int send_jobs_to_accounting(time_t event_time) +extern int send_jobs_to_accounting() { ListIterator itr = NULL; struct job_record *job_ptr; @@ -6105,7 +6229,7 @@ accounting_enforce, (acct_association_rec_t **) &job_ptr->assoc_ptr) && - accounting_enforce + (accounting_enforce & ACCOUNTING_ENFORCE_ASSOCS) && (!IS_JOB_FINISHED(job_ptr))) { info("Cancelling job %u with " "invalid association", diff -Nru slurm-llnl-1.3.13/src/slurmctld/job_scheduler.c slurm-llnl-1.3.15/src/slurmctld/job_scheduler.c --- slurm-llnl-1.3.13/src/slurmctld/job_scheduler.c 2009-01-12 17:30:40.000000000 +0100 +++ slurm-llnl-1.3.15/src/slurmctld/job_scheduler.c 2009-02-13 01:57:02.000000000 +0100 @@ -763,8 +763,10 @@ /* Clear dependencies on NULL or empty dependency input */ if ((new_depend == NULL) || (new_depend[0] == '\0')) { xfree(job_ptr->details->dependency); - if (job_ptr->details->depend_list) + if (job_ptr->details->depend_list) { list_destroy(job_ptr->details->depend_list); + job_ptr->details->depend_list = NULL; + } return rc; } diff -Nru slurm-llnl-1.3.13/src/slurmctld/node_scheduler.c slurm-llnl-1.3.15/src/slurmctld/node_scheduler.c --- slurm-llnl-1.3.13/src/slurmctld/node_scheduler.c 2009-01-12 19:42:28.000000000 +0100 +++ slurm-llnl-1.3.15/src/slurmctld/node_scheduler.c 2009-01-20 21:49:38.000000000 +0100 @@ -996,8 +996,9 @@ /* Leave job queued, something is hosed */ error("select_g_job_begin(%u): %m", job_ptr->job_id); error_code = ESLURM_NODES_BUSY; - job_ptr->start_time = job_ptr->time_last_active - = job_ptr->end_time = 0; + job_ptr->start_time = 0; + job_ptr->time_last_active = 0; + job_ptr->end_time = 0; goto cleanup; } diff -Nru slurm-llnl-1.3.13/src/slurmctld/proc_req.c slurm-llnl-1.3.15/src/slurmctld/proc_req.c --- slurm-llnl-1.3.13/src/slurmctld/proc_req.c 2008-12-05 20:34:54.000000000 +0100 +++ slurm-llnl-1.3.15/src/slurmctld/proc_req.c 2009-03-23 21:37:54.000000000 +0100 @@ -3005,8 +3005,7 @@ return; } - send_jobs_to_accounting(event_time); - send_nodes_to_accounting(event_time); + send_all_to_accounting(event_time); END_TIMER2("_slurm_rpc_accounting_first_reg"); } diff -Nru slurm-llnl-1.3.13/src/slurmctld/read_config.c slurm-llnl-1.3.15/src/slurmctld/read_config.c --- slurm-llnl-1.3.13/src/slurmctld/read_config.c 2008-12-31 20:55:28.000000000 +0100 +++ slurm-llnl-1.3.15/src/slurmctld/read_config.c 2009-02-03 00:02:44.000000000 +0100 @@ -1203,7 +1203,7 @@ assoc_mgr_clear_used_info(); job_iterator = list_iterator_create(job_list); while ((job_ptr = (struct job_record *) list_next(job_iterator))) { - if (accounting_enforce == ACCOUNTING_ENFORCE_WITH_LIMITS) { + if (accounting_enforce & ACCOUNTING_ENFORCE_LIMITS) { if((job_ptr->job_state == JOB_RUNNING) || (job_ptr->job_state == JOB_SUSPENDED)) acct_policy_job_begin(job_ptr); diff -Nru slurm-llnl-1.3.13/src/slurmctld/sched_plugin.c slurm-llnl-1.3.15/src/slurmctld/sched_plugin.c --- slurm-llnl-1.3.13/src/slurmctld/sched_plugin.c 2008-06-06 21:15:24.000000000 +0200 +++ slurm-llnl-1.3.15/src/slurmctld/sched_plugin.c 2009-03-24 22:57:43.000000000 +0100 @@ -181,13 +181,14 @@ static int slurm_sched_context_destroy( slurm_sched_context_t *c ) { + int rc = SLURM_SUCCESS; /* * Must check return code here because plugins might still * be loaded and active. */ if ( c->plugin_list ) { if ( plugrack_destroy( c->plugin_list ) != SLURM_SUCCESS ) { - return SLURM_ERROR; + rc = SLURM_ERROR; } } else { plugin_unload(c->cur_plugin); @@ -196,7 +197,7 @@ xfree( c->sched_type ); xfree( c ); - return SLURM_SUCCESS; + return rc; } diff -Nru slurm-llnl-1.3.13/src/slurmctld/slurmctld.h slurm-llnl-1.3.15/src/slurmctld/slurmctld.h --- slurm-llnl-1.3.13/src/slurmctld/slurmctld.h 2008-12-31 23:25:19.000000000 +0100 +++ slurm-llnl-1.3.15/src/slurmctld/slurmctld.h 2009-03-23 21:37:54.000000000 +0100 @@ -1335,10 +1335,13 @@ /* save_all_state - save entire slurmctld state for later recovery */ extern void save_all_state(void); +/* send all info for the controller to accounting */ +extern void send_all_to_accounting(time_t event_time); + /* sends all jobs in eligible state to accounting. Only needed at * first registration */ -extern int send_jobs_to_accounting(time_t event_time); +extern int send_jobs_to_accounting(); /* send all nodes in a down like state to accounting. Only needed at * first registration @@ -1481,6 +1484,16 @@ extern int update_job_account(char *module, struct job_record *job_ptr, char *new_account); +/* + * Modify the wckey associated with a pending job + * IN module - where this is called from + * IN job_ptr - pointer to job which should be modified + * IN new_wckey - desired wckey name + * RET SLURM_SUCCESS or error code + */ +extern int update_job_wckey(char *module, struct job_record *job_ptr, + char *new_wckey); + /* Reset nodes_completing field for all jobs */ extern void update_job_nodes_completing(void); diff -Nru slurm-llnl-1.3.13/src/slurmctld/step_mgr.c slurm-llnl-1.3.15/src/slurmctld/step_mgr.c --- slurm-llnl-1.3.13/src/slurmctld/step_mgr.c 2008-12-04 21:17:23.000000000 +0100 +++ slurm-llnl-1.3.15/src/slurmctld/step_mgr.c 2009-02-18 20:03:40.000000000 +0100 @@ -1,6 +1,6 @@ /*****************************************************************************\ * step_mgr.c - manage the job step information of slurm - * $Id: step_mgr.c 15827 2008-12-04 20:17:23Z jette $ + * $Id: step_mgr.c 16584 2009-02-18 19:03:40Z jette $ ***************************************************************************** * Copyright (C) 2002-2007 The Regents of the University of California. * Produced at Lawrence Livermore National Laboratory (cf, DISCLAIMER). @@ -1100,7 +1100,7 @@ error("step_layout_create exclusive"); return NULL; } - usable_cpus = MAX(usable_cpus, + usable_cpus = MIN(usable_cpus, (num_tasks - set_cpus)); } else usable_cpus = job_ptr->alloc_lps[pos]; diff -Nru slurm-llnl-1.3.13/src/slurmctld/trigger_mgr.c slurm-llnl-1.3.15/src/slurmctld/trigger_mgr.c --- slurm-llnl-1.3.13/src/slurmctld/trigger_mgr.c 2008-07-30 19:39:47.000000000 +0200 +++ slurm-llnl-1.3.15/src/slurmctld/trigger_mgr.c 2009-04-02 23:22:36.000000000 +0200 @@ -977,6 +977,8 @@ trig_in->group_id = child; } else if (child == 0) { int i; + bool run_as_self = (uid == getuid()); + for (i=0; i<128; i++) close(i); #ifdef SETPGRP_TWO_ARGS @@ -985,9 +987,18 @@ setpgrp(); #endif setsid(); - setuid(uid); - setgid(gid); - initgroups(user_name, -1); + if ((initgroups(user_name, gid) == -1) && !run_as_self) { + error("trigger: initgroups: %m"); + exit(1); + } + if ((setgid(gid) == -1) && !run_as_self){ + error("trigger: setgid: %m"); + exit(1); + } + if ((setuid(uid) == -1) && !run_as_self) { + error("trigger: setuid: %m"); + exit(1); + } execl(program, arg0, arg1, NULL); exit(1); } else diff -Nru slurm-llnl-1.3.13/src/slurmd/common/proctrack.c slurm-llnl-1.3.15/src/slurmd/common/proctrack.c --- slurm-llnl-1.3.13/src/slurmd/common/proctrack.c 2008-06-06 21:15:24.000000000 +0200 +++ slurm-llnl-1.3.15/src/slurmd/common/proctrack.c 2009-03-24 22:57:43.000000000 +0100 @@ -173,13 +173,15 @@ static int _proctrack_context_destroy( slurm_proctrack_context_t *c ) { + int rc = SLURM_SUCCESS; + /* * Must check return code here because plugins might still * be loaded and active. */ if ( c->plugin_list ) { if ( plugrack_destroy( c->plugin_list ) != SLURM_SUCCESS ) { - return SLURM_ERROR; + rc = SLURM_ERROR; } } else { plugin_unload(c->cur_plugin); @@ -188,7 +190,7 @@ xfree( c->proctrack_type ); xfree( c ); - return SLURM_SUCCESS; + return rc; } diff -Nru slurm-llnl-1.3.13/src/slurmd/common/task_plugin.c slurm-llnl-1.3.15/src/slurmd/common/task_plugin.c --- slurm-llnl-1.3.13/src/slurmd/common/task_plugin.c 2008-06-06 21:15:24.000000000 +0200 +++ slurm-llnl-1.3.15/src/slurmd/common/task_plugin.c 2009-03-24 22:57:43.000000000 +0100 @@ -158,13 +158,14 @@ static int _slurmd_task_context_destroy(slurmd_task_context_t *c) { + int rc = SLURM_SUCCESS; /* * Must check return code here because plugins might still * be loaded and active. */ if ( c->plugin_list ) { if ( plugrack_destroy( c->plugin_list ) != SLURM_SUCCESS ) { - return SLURM_ERROR; + rc = SLURM_ERROR; } } else { plugin_unload(c->cur_plugin); @@ -173,7 +174,7 @@ xfree( c->task_type ); xfree( c ); - return SLURM_SUCCESS; + return rc; } diff -Nru slurm-llnl-1.3.13/src/slurmd/slurmd/req.c slurm-llnl-1.3.15/src/slurmd/slurmd/req.c --- slurm-llnl-1.3.13/src/slurmd/slurmd/req.c 2008-11-24 21:22:25.000000000 +0100 +++ slurm-llnl-1.3.15/src/slurmd/slurmd/req.c 2009-03-27 18:38:38.000000000 +0100 @@ -108,6 +108,7 @@ static void _job_limits_free(void *x); static int _job_limits_match(void *x, void *key); static bool _job_still_running(uint32_t job_id); +static int _init_groups(uid_t my_uid, gid_t my_gid); static int _kill_all_active_steps(uint32_t jobid, int sig, bool batch); static int _terminate_all_steps(uint32_t jobid, bool batch); static void _rpc_launch_tasks(slurm_msg_t *); @@ -1836,12 +1837,34 @@ } static int +_init_groups(uid_t my_uid, gid_t my_gid) +{ + char *user_name = uid_to_string(my_uid); + int rc; + + if (user_name == NULL) { + error("sbcast: Could not find uid %ld", (long)my_uid); + return -1; + } + + rc = initgroups(user_name, my_gid); + xfree(user_name); + if (rc) { + error("sbcast: Error in initgroups(%s, %ld): %m", + user_name, (long)my_gid); + return -1; + } + return 0; + +} + +static int _rpc_file_bcast(slurm_msg_t *msg) { file_bcast_msg_t *req = msg->data; int fd, flags, offset, inx, rc; uid_t req_uid = g_slurm_auth_get_uid(msg->auth_cred, NULL); - uid_t req_gid = g_slurm_auth_get_gid(msg->auth_cred, NULL); + gid_t req_gid = g_slurm_auth_get_gid(msg->auth_cred, NULL); pid_t child; #if 0 @@ -1867,6 +1890,10 @@ /* The child actually performs the I/O and exits with * a return code, do not return! */ + if (_init_groups(req_uid, req_gid) < 0) { + error("sbcast: initgroups(%u): %m", req_uid); + exit(errno); + } if (setgid(req_gid) < 0) { error("sbcast: uid:%u setgid(%u): %s", req_uid, req_gid, strerror(errno)); diff -Nru slurm-llnl-1.3.13/src/slurmd/slurmd/slurmd.c slurm-llnl-1.3.15/src/slurmd/slurmd/slurmd.c --- slurm-llnl-1.3.13/src/slurmd/slurmd/slurmd.c 2008-11-04 00:14:27.000000000 +0100 +++ slurm-llnl-1.3.15/src/slurmd/slurmd/slurmd.c 2009-04-07 20:09:43.000000000 +0200 @@ -1,6 +1,6 @@ /*****************************************************************************\ * src/slurmd/slurmd/slurmd.c - main slurm node server daemon - * $Id: slurmd.c 15572 2008-11-03 23:14:27Z jette $ + * $Id: slurmd.c 17177 2009-04-07 18:09:43Z jette $ ***************************************************************************** * Copyright (C) 2002-2007 The Regents of the University of California. * Copyright (C) 2008 Lawrence Livermore National Security. @@ -43,6 +43,7 @@ #endif #include +#include #include #include #include @@ -160,6 +161,18 @@ (void) close(i); /* + * Drop supplementary groups. + */ + if (geteuid() == 0) { + if (setgroups(0, NULL) != 0) { + fatal("Failed to drop supplementary groups, " + "setgroups: %m"); + } + } else { + info("Not running as root. Can't drop supplementary groups"); + } + + /* * Create and set default values for the slurmd global * config variable "conf" */ diff -Nru slurm-llnl-1.3.13/src/slurmd/slurmstepd/mgr.c slurm-llnl-1.3.15/src/slurmd/slurmstepd/mgr.c --- slurm-llnl-1.3.13/src/slurmd/slurmstepd/mgr.c 2008-12-05 00:59:29.000000000 +0100 +++ slurm-llnl-1.3.15/src/slurmd/slurmstepd/mgr.c 2009-03-26 16:03:18.000000000 +0100 @@ -1,6 +1,6 @@ /*****************************************************************************\ * src/slurmd/slurmstepd/mgr.c - job manager functions for slurmstepd - * $Id: mgr.c 15835 2008-12-04 23:59:29Z jette $ + * $Id: mgr.c 17040 2009-03-26 15:03:18Z jette $ ***************************************************************************** * Copyright (C) 2002-2007 The Regents of the University of California. * Copyright (C) 2008 Lawrence Livermore National Security. @@ -949,6 +949,7 @@ goto fail2; } else if (pid == 0) { /* child */ int j; + #ifdef HAVE_AIX (void) mkcrid(0); #endif diff -Nru slurm-llnl-1.3.13/src/slurmd/slurmstepd/slurmstepd.c slurm-llnl-1.3.15/src/slurmd/slurmstepd/slurmstepd.c --- slurm-llnl-1.3.13/src/slurmd/slurmstepd/slurmstepd.c 2008-12-04 00:32:14.000000000 +0100 +++ slurm-llnl-1.3.15/src/slurmd/slurmstepd/slurmstepd.c 2009-03-26 16:03:18.000000000 +0100 @@ -1,6 +1,6 @@ /*****************************************************************************\ * src/slurmd/slurmstepd/slurmstepd.c - SLURM job-step manager. - * $Id: slurmstepd.c 15819 2008-12-03 23:32:14Z jette $ + * $Id: slurmstepd.c 17040 2009-03-26 15:03:18Z jette $ ***************************************************************************** * Copyright (C) 2002-2006 The Regents of the University of California. * Produced at Lawrence Livermore National Laboratory (cf, DISCLAIMER). @@ -361,7 +361,7 @@ return job; } -#ifdef MEMORY_LEAK_TEST +#ifdef MEMORY_LEAK_DEBUG static void _step_cleanup(slurmd_job_t *job, slurm_msg_t *msg, int rc) { diff -Nru slurm-llnl-1.3.13/src/slurmd/slurmstepd/task.c slurm-llnl-1.3.15/src/slurmd/slurmstepd/task.c --- slurm-llnl-1.3.13/src/slurmd/slurmstepd/task.c 2008-10-15 23:47:57.000000000 +0200 +++ slurm-llnl-1.3.15/src/slurmd/slurmstepd/task.c 2009-01-22 19:08:43.000000000 +0100 @@ -98,6 +98,7 @@ * Static prototype definitions. */ static void _make_tmpdir(slurmd_job_t *job); +static void _print_stdout(char *buf); static int _run_script_and_set_env(const char *name, const char *path, slurmd_job_t *job); static void _update_env(char *buf, char ***env); @@ -105,8 +106,7 @@ /* Search for "export NAME=value" records in buf and * use them to add environment variables to env */ -static void -_update_env(char *buf, char ***env) +static void _update_env(char *buf, char ***env) { char *tmp_ptr, *name_ptr, *val_ptr, *buf_ptr = buf; @@ -132,7 +132,32 @@ } debug("name:%s:val:%s:",name_ptr,val_ptr); if (setenvf(env, name_ptr, "%s", val_ptr)) - error("Unable to set %s environment variable", name_ptr); + error("Unable to set %s environment variable", + name_ptr); + } +} + +/* Search for "print " records in buf and + * write that to the job's stdout */ +static void _print_stdout(char *buf) +{ + char *tmp_ptr, *buf_ptr = buf; + + while ((tmp_ptr = strstr(buf_ptr, "print "))) { + if ((tmp_ptr != buf_ptr) && (tmp_ptr[-1] != '\n')) { + /* Skip "print " if not at start of a line */ + buf_ptr +=6; + continue; + } + buf_ptr = tmp_ptr + 6; + tmp_ptr = strchr(buf_ptr, '\n'); + if (tmp_ptr) { + write(1, buf_ptr, (tmp_ptr - buf_ptr + 1)); + buf_ptr = tmp_ptr + 1; + } else { + write(1, buf_ptr, strlen(buf_ptr)); + break; + } } } @@ -196,6 +221,7 @@ buf[nread] = 0; //debug("read %d:%s:", nread, buf); _update_env(buf, &job->env); + _print_stdout(buf); } close(pfd[0]); diff -Nru slurm-llnl-1.3.13/src/slurmdbd/proc_req.c slurm-llnl-1.3.15/src/slurmdbd/proc_req.c --- slurm-llnl-1.3.13/src/slurmdbd/proc_req.c 2008-12-23 00:56:27.000000000 +0100 +++ slurm-llnl-1.3.15/src/slurmdbd/proc_req.c 2009-02-11 23:02:02.000000000 +0100 @@ -1,7 +1,7 @@ /*****************************************************************************\ * proc_req.c - functions for processing incoming RPCs. ***************************************************************************** - * Copyright (C) 2008 Lawrence Livermore National Security. + * Copyright (C) 2008-2009 Lawrence Livermore National Security. * Produced at Lawrence Livermore National Laboratory (cf, DISCLAIMER). * Written by Morris Jette * LLNL-CODE-402394. @@ -803,7 +803,12 @@ arch_cond->step_purge = slurmdbd_conf->step_purge; rc = jobacct_storage_g_archive(slurmdbd_conn->db_conn, arch_cond); - + if(rc != SLURM_SUCCESS) { + if(errno == EACCES) + comment = "Problem accessing file."; + else + comment = "Error with request."; + } end_it: slurmdbd_free_cond_msg(slurmdbd_conn->rpc_version, DBD_ARCHIVE_DUMP, get_msg); @@ -839,7 +844,10 @@ } rc = jobacct_storage_g_archive_load(slurmdbd_conn->db_conn, arch_rec); - + if(rc == ENOENT) + comment = "No archive file given to recover."; + else if(rc != SLURM_SUCCESS) + comment = "Error with request."; end_it: destroy_acct_archive_rec(arch_rec); *out_buffer = make_dbd_rc_msg(slurmdbd_conn->rpc_version, diff -Nru slurm-llnl-1.3.13/src/slurmdbd/read_config.c slurm-llnl-1.3.15/src/slurmdbd/read_config.c --- slurm-llnl-1.3.13/src/slurmdbd/read_config.c 2008-12-24 19:22:49.000000000 +0100 +++ slurm-llnl-1.3.15/src/slurmdbd/read_config.c 2009-02-10 18:32:01.000000000 +0100 @@ -114,8 +114,8 @@ extern int read_slurmdbd_conf(void) { s_p_options_t options[] = { - {"ArchiveJobs", S_P_BOOLEAN}, {"ArchiveDir", S_P_STRING}, + {"ArchiveJobs", S_P_BOOLEAN}, {"ArchiveScript", S_P_STRING}, {"ArchiveSteps", S_P_BOOLEAN}, {"AuthInfo", S_P_STRING}, diff -Nru slurm-llnl-1.3.13/src/slurmdbd/rpc_mgr.c slurm-llnl-1.3.15/src/slurmdbd/rpc_mgr.c --- slurm-llnl-1.3.13/src/slurmdbd/rpc_mgr.c 2008-12-23 23:19:51.000000000 +0100 +++ slurm-llnl-1.3.15/src/slurmdbd/rpc_mgr.c 2009-03-23 21:37:54.000000000 +0100 @@ -222,7 +222,7 @@ rc = proc_req( conn, msg, msg_size, first, &buffer, &uid); first = false; - if (rc != SLURM_SUCCESS) { + if (rc != SLURM_SUCCESS && rc != ACCOUNTING_FIRST_REG) { error("Processing last message from " "connection %d(%s) uid(%d)", conn->newsockfd, conn->ip, uid); diff -Nru slurm-llnl-1.3.13/src/smap/configure_functions.c slurm-llnl-1.3.15/src/smap/configure_functions.c --- slurm-llnl-1.3.13/src/smap/configure_functions.c 2008-12-29 22:56:17.000000000 +0100 +++ slurm-llnl-1.3.15/src/smap/configure_functions.c 2009-02-13 22:54:09.000000000 +0100 @@ -98,9 +98,12 @@ request->geometry[2]); return NULL; } else { - if(request->passthrough) + char *pass = ba_passthroughs_string(request->deny_pass); + if(pass) { sprintf(error_string,"THERE ARE PASSTHROUGHS IN " - "THIS ALLOCATION!!!!!!!"); + "THIS ALLOCATION DIM %s!!!!!!!", pass); + xfree(pass); + } allocated_block = (allocated_block_t *)xmalloc( sizeof(allocated_block_t)); @@ -204,7 +207,9 @@ allocated_block_t *allocated_block = NULL; ba_request_t *request = (ba_request_t*) xmalloc(sizeof(ba_request_t)); int diff=0; - +#ifndef HAVE_BGL + int small16=-1, small64=-1, small256=-1; +#endif request->geometry[0] = (uint16_t)NO_VAL; request->conn_type=SELECT_TORUS; request->rotate = false; @@ -213,7 +218,7 @@ request->size = 0; request->small32 = 0; request->small128 = 0; - request->passthrough = false; + request->deny_pass = 0; request->avail_node_bitmap = NULL; while(iconn_type = SELECT_SMALL; i+=5; + } else if(!strncasecmp(com+i, "deny", 4)) { + i+=4; + if(strstr(com+i, "X")) + request->deny_pass |= PASS_DENY_X; + if(strstr(com+i, "Y")) + request->deny_pass |= PASS_DENY_Y; + if(strstr(com+i, "Z")) + request->deny_pass |= PASS_DENY_Z; + if(!strcasecmp(com+i, "ALL")) + request->deny_pass |= PASS_DENY_ALL; } else if(!strncasecmp(com+i, "nodecard", 8)) { small32=0; - i+=5; + i+=8; } else if(!strncasecmp(com+i, "quarter", 7)) { small128=0; - i+=6; + i+=7; + } else if(!strncasecmp(com+i, "32CN", 4)) { + small32=0; + i+=4; + } else if(!strncasecmp(com+i, "128CN", 5)) { + small128=0; + i+=5; } else if(!strncasecmp(com+i, "rotate", 6)) { request->rotate=true; i+=6; @@ -244,13 +265,35 @@ || (com[i] >= 'A' && com[i] <= 'Z'))) { starti=i; i++; - } else if(small32 == 0 && (com[i] < 58 && com[i] > 47)) { + } else if(small32 == 0 && (com[i] >= '0' && com[i] <= '9')) { small32=i; i++; - } else if(small128 == 0 && (com[i] < 58 && com[i] > 47)) { + } else if(small128 == 0 && (com[i] >= '0' && com[i] <= '9')) { small128=i; i++; - } else if(geoi<0 && ((com[i] >= '0' && com[i] <= '9') + } +#ifndef HAVE_BGL + else if(!strncasecmp(com+i, "16CN", 4)) { + small16=0; + i+=4; + } else if(!strncasecmp(com+i, "64CN", 4)) { + small64=0; + i+=4; + } else if(!strncasecmp(com+i, "256CN", 5)) { + small256=0; + i+=5; + } else if(small16 == 0 && (com[i] >= '0' && com[i] <= '9')) { + small16=i; + i++; + } else if(small64 == 0 && (com[i] >= '0' && com[i] <= '9')) { + small64=i; + i++; + } else if(small256 == 0 && (com[i] >= '0' && com[i] <= '9')) { + small256=i; + i++; + } +#endif + else if(geoi<0 && ((com[i] >= '0' && com[i] <= '9') || (com[i] >= 'A' && com[i] <= 'Z'))) { geoi=i; i++; @@ -261,36 +304,79 @@ } if(request->conn_type == SELECT_SMALL) { + int total = 512; +#ifndef HAVE_BGL + if(small16 > 0) { + request->small16 = atoi(&com[small16]); + total -= request->small16 * 16; + } + + if(small64 > 0) { + request->small64 = atoi(&com[small64]); + total -= request->small64 * 64; + } + + if(small256 > 0) { + request->small256 = atoi(&com[small256]); + total -= request->small256 * 256; + } +#endif + if(small32 > 0) { request->small32 = atoi(&com[small32]); - small32 = request->small32/4; - request->small32 = small32*4; + total -= request->small32 * 32; } - request->small128 = 4; - - if(request->small32 > 0) - request->small128 -= small32; - - if(request->small128 > 4) { - request->small128 = 4; - request->small32 = 0; - } else if(request->small32 > 16) { - request->small128 = 0; - request->small32 = 16; + if(small128 > 0) { + request->small128 = atoi(&com[small128]); + total -= request->small128 * 128; } - - small128 = request->small128*4; - small32 = request->small32; - if((small128+small32) > 16) { + if(total < 0) { sprintf(error_string, - "please specify a complete split of a " - "Base Partion\n" - "(i.e. small32=4)"); + "You asked for %d more nodes than " + "are in a Midplane\n", total * 2); geoi = -1; + + } + +#ifndef HAVE_BGL + while(total > 0) { + if(total >= 256) { + request->small256++; + total -= 256; + } else if(total >= 128) { + request->small128++; + total -= 128; + } else if(total >= 64) { + request->small64++; + total -= 64; + } else if(total >= 32) { + request->small32++; + total -= 32; + } else if(total >= 16) { + request->small16++; + total -= 16; + } else + break; } +#else + while(total > 0) { + if(total >= 128) { + request->small128++; + total -= 128; + } else if(total >= 32) { + request->small32++; + total -= 32; + } else + break; + } +#endif request->size = 1; - +/* sprintf(error_string, */ +/* "got %d %d %d %d %d %d", */ +/* total, request->small16, request->small32, */ +/* request->small64, request->small128, */ +/* request->small256); */ } if(geoi<0 && !request->size) { @@ -858,6 +944,12 @@ request->conn_type=allocated_block->request->conn_type; request->rotate =allocated_block->request->rotate; request->elongate = allocated_block->request->elongate; + request->deny_pass = allocated_block->request->deny_pass; +#ifndef HAVE_BGL + request->small16 = allocated_block->request->small16; + request->small64 = allocated_block->request->small64; + request->small256 = allocated_block->request->small256; +#endif request->small32 = allocated_block->request->small32; request->small128 = allocated_block->request->small128; @@ -961,9 +1053,23 @@ conn_type = "MESH"; else { conn_type = "SMALL"; - xstrfmtcat(extra, " NodeCards=%d Quarters=%d", +#ifndef HAVE_BGL + xstrfmtcat(extra, + " 16CNBlocks=%d 32CNBlocks=%d " + "64CNBlocks=%d 128CNBlocks=%d " + "256CNBlocks=%d", + allocated_block->request->small16, + allocated_block->request->small32, + allocated_block->request->small64, + allocated_block->request->small128, + allocated_block->request->small256); +#else + xstrfmtcat(extra, + " 32CNBlocks=%d 128CNBlocks=%d", allocated_block->request->small32, allocated_block->request->small128); + +#endif } xstrfmtcat(save_string, "BPs=%s Type=%s", allocated_block->request->save_name, @@ -1226,12 +1332,28 @@ main_xcord, "NODES"); #endif main_xcord += 10; + +#ifndef HAVE_BGL + mvwprintw(text_win, main_ycord, + main_xcord, "16CN"); + main_xcord += 5; +#endif mvwprintw(text_win, main_ycord, - main_xcord, "NODECARDS"); - main_xcord += 11; + main_xcord, "32CN"); + main_xcord += 5; +#ifndef HAVE_BGL mvwprintw(text_win, main_ycord, - main_xcord, "QUARTERS"); - main_xcord += 10; + main_xcord, "64CN"); + main_xcord += 5; +#endif + mvwprintw(text_win, main_ycord, + main_xcord, "128CN"); + main_xcord += 6; +#ifndef HAVE_BGL + mvwprintw(text_win, main_ycord, + main_xcord, "256CN"); + main_xcord += 6; +#endif #ifdef HAVE_BG mvwprintw(text_win, main_ycord, main_xcord, "BP_LIST"); @@ -1249,7 +1371,7 @@ COLOR_PAIR(allocated_block->color)); mvwprintw(text_win, main_ycord, - main_xcord, "%c",allocated_block->letter); + main_xcord, "%c", allocated_block->letter); main_xcord += 4; if(allocated_block->request->conn_type==SELECT_TORUS) mvwprintw(text_win, main_ycord, @@ -1283,18 +1405,40 @@ main_xcord += 10; if(allocated_block->request->conn_type == SELECT_SMALL) { +#ifndef HAVE_BGL + mvwprintw(text_win, main_ycord, + main_xcord, "%d", + allocated_block->request->small16); + main_xcord += 5; +#endif mvwprintw(text_win, main_ycord, main_xcord, "%d", allocated_block->request->small32); - main_xcord += 11; + main_xcord += 5; +#ifndef HAVE_BGL + mvwprintw(text_win, main_ycord, + main_xcord, "%d", + allocated_block->request->small64); + main_xcord += 5; +#endif mvwprintw(text_win, main_ycord, main_xcord, "%d", allocated_block->request->small128); - main_xcord += 10; + main_xcord += 6; +#ifndef HAVE_BGL + mvwprintw(text_win, main_ycord, + main_xcord, "%d", + allocated_block->request->small256); + main_xcord += 6; +#endif } else - main_xcord += 21; - +#ifndef HAVE_BGL + main_xcord += 27; +#else + main_xcord += 11; +#endif + mvwprintw(text_win, main_ycord, main_xcord, "%s", allocated_block->request->save_name); @@ -1393,7 +1537,9 @@ _delete_allocated_blocks(allocated_blocks); ba_fini(); exit(0); - } if (!strcmp(com, "quit")) { + } + + if (!strcmp(com, "quit") || !strcmp(com, "\\q")) { break; } else if (!strncasecmp(com, "layout", 6)) { _set_layout(com); diff -Nru slurm-llnl-1.3.13/src/smap/job_functions.c slurm-llnl-1.3.15/src/smap/job_functions.c --- slurm-llnl-1.3.13/src/smap/job_functions.c 2008-12-29 22:56:17.000000000 +0100 +++ slurm-llnl-1.3.15/src/smap/job_functions.c 2009-03-24 18:41:50.000000000 +0100 @@ -195,7 +195,7 @@ main_xcord += 3; mvwprintw(text_win, main_ycord, main_xcord, "JOBID"); - main_xcord += 6; + main_xcord += 8; mvwprintw(text_win, main_ycord, main_xcord, "PARTITION"); main_xcord += 10; @@ -297,7 +297,7 @@ main_xcord += 3; mvwprintw(text_win, main_ycord, main_xcord, "%d", job_ptr->job_id); - main_xcord += 6; + main_xcord += 8; mvwprintw(text_win, main_ycord, main_xcord, "%.10s", job_ptr->partition); main_xcord += 10; @@ -323,7 +323,7 @@ job_state_string_compact(job_ptr->job_state)); main_xcord += 2; if(!strcasecmp(job_ptr->nodes,"waiting...")) { - sprintf(time_buf,"0:00:00"); + sprintf(time_buf,"00:00:00"); } else { time_diff = now_time - job_ptr->start_time; secs2time_str(time_diff, time_buf, sizeof(time_buf)); @@ -375,7 +375,7 @@ main_xcord = 1; main_ycord++; } else { - printf("%5d ", job_ptr->job_id); + printf("%8d ", job_ptr->job_id); printf("%9.9s ", job_ptr->partition); #ifdef HAVE_BG printf("%16.16s ", @@ -391,7 +391,7 @@ printf("%2.2s ", job_state_string_compact(job_ptr->job_state)); if(!strcasecmp(job_ptr->nodes,"waiting...")) { - sprintf(time_buf,"0:00:00"); + sprintf(time_buf,"00:00:00"); } else { time_diff = now_time - job_ptr->start_time; secs2time_str(time_diff, time_buf, sizeof(time_buf)); diff -Nru slurm-llnl-1.3.13/src/smap/partition_functions.c slurm-llnl-1.3.15/src/smap/partition_functions.c --- slurm-llnl-1.3.13/src/smap/partition_functions.c 2008-12-29 22:56:17.000000000 +0100 +++ slurm-llnl-1.3.15/src/smap/partition_functions.c 2009-02-05 01:06:53.000000000 +0100 @@ -944,6 +944,7 @@ static char* _convert_conn_type(enum connection_type conn_type) { +#ifdef HAVE_BG switch (conn_type) { case (SELECT_MESH): return "MESH"; @@ -953,7 +954,24 @@ return "SMALL"; case (SELECT_NAV): return "NAV"; +#ifndef HAVE_BGL + case SELECT_HTC_S: + return "HTC_S"; + break; + case SELECT_HTC_D: + return "HTC_D"; + break; + case SELECT_HTC_V: + return "HTC_V"; + break; + case SELECT_HTC_L: + return "HTC_L"; + break; +#endif + default: + return "?"; } +#endif return "?"; } diff -Nru slurm-llnl-1.3.13/src/smap/smap.c slurm-llnl-1.3.15/src/smap/smap.c --- slurm-llnl-1.3.13/src/smap/smap.c 2008-12-29 22:56:17.000000000 +0100 +++ slurm-llnl-1.3.15/src/smap/smap.c 2009-03-23 22:37:58.000000000 +0100 @@ -182,7 +182,7 @@ COLS, LINES); ba_fini(); - exit(0); + exit(1); } raw(); @@ -252,7 +252,7 @@ if(!params.commandline) endwin(); ba_fini(); - exit(0); + exit(1); break; #endif } diff -Nru slurm-llnl-1.3.13/src/squeue/opts.c slurm-llnl-1.3.15/src/squeue/opts.c --- slurm-llnl-1.3.13/src/squeue/opts.c 2008-12-03 00:38:47.000000000 +0100 +++ slurm-llnl-1.3.15/src/squeue/opts.c 2009-01-29 19:16:08.000000000 +0100 @@ -1,7 +1,7 @@ /****************************************************************************\ * opts.c - srun command line option parsing * - * $Id: opts.c 15808 2008-12-02 23:38:47Z da $ + * $Id: opts.c 16350 2009-01-29 18:16:08Z jette $ ***************************************************************************** * Copyright (C) 2002-2006 The Regents of the University of California. * Produced at Lawrence Livermore National Laboratory (cf, DISCLAIMER). @@ -68,7 +68,7 @@ /* FUNCTIONS */ static List _build_job_list( char* str ); -static List _build_part_list( char* str ); +static List _build_str_list( char* str ); static List _build_state_list( char* str ); static List _build_all_states_list( void ); static List _build_step_list( char* str ); @@ -107,6 +107,7 @@ {"states", required_argument, 0, 't'}, {"user", required_argument, 0, 'u'}, {"users", required_argument, 0, 'u'}, + {"account", required_argument, 0, 'U'}, {"verbose", no_argument, 0, 'v'}, {"version", no_argument, 0, 'V'}, {"help", no_argument, 0, OPT_LONG_HELP}, @@ -122,7 +123,7 @@ if ( ( env_val = getenv("SQUEUE_SORT") ) ) params.sort = xstrdup(env_val); - while((opt_char = getopt_long(argc, argv, "ahi:j::ln:o:p:s::S:t:u:vV", + while((opt_char = getopt_long(argc, argv, "ahi:j::ln:o:p:s::S:t:u:U:vV", long_options, &option_index)) != -1) { switch (opt_char) { case (int)'?': @@ -172,7 +173,13 @@ xfree(params.partitions); params.partitions = xstrdup(optarg); params.part_list = - _build_part_list( params.partitions ); + _build_str_list( params.partitions ); + break; + case (int) 'U': + xfree(params.accounts); + params.accounts = xstrdup(optarg); + params.account_list = + _build_str_list( params.accounts ); break; case (int) 's': if (optarg) { @@ -271,7 +278,13 @@ if ( ( params.partitions == NULL ) && ( env_val = getenv("SQUEUE_PARTITION") ) ) { params.partitions = xstrdup(env_val); - params.part_list = _build_part_list( params.partitions ); + params.part_list = _build_str_list( params.partitions ); + } + + if ( ( params.accounts == NULL ) && + ( env_val = getenv("SQUEUE_ACCOUNT") ) ) { + params.accounts = xstrdup(env_val); + params.account_list = _build_str_list( params.accounts ); } if ( ( params.states == NULL ) && @@ -838,12 +851,12 @@ } /* - * _build_part_list- build a list of partition names - * IN str - comma separated list of partition names - * RET List of partition names + * _build_str_list- build a list of strings + * IN str - comma separated list of strings + * RET List of strings */ static List -_build_part_list( char* str ) +_build_str_list( char* str ) { List my_list; char *part = NULL, *tmp_char = NULL, *my_part_list = NULL; diff -Nru slurm-llnl-1.3.13/src/squeue/print.c slurm-llnl-1.3.15/src/squeue/print.c --- slurm-llnl-1.3.13/src/squeue/print.c 2008-12-04 18:47:24.000000000 +0100 +++ slurm-llnl-1.3.15/src/squeue/print.c 2009-03-13 20:51:48.000000000 +0100 @@ -653,7 +653,9 @@ if (job == NULL) /* Print the Header instead */ _print_str("CPUS", width, right, true); else { - if (job->job_state == JOB_RUNNING) { + if ((job->num_cpu_groups > 0) && + (job->cpus_per_node) && + (job->cpu_count_reps)) { uint32_t cnt = 0, i; for (i=0; inum_cpu_groups; i++) { cnt += job->cpus_per_node[i] * @@ -1273,7 +1275,7 @@ ListIterator iterator; uint32_t *job_id, *user; enum job_states *state_id; - char *part; + char *part, *account; if (params.job_list) { filter = 1; @@ -1302,6 +1304,21 @@ if (filter == 1) return 2; } + + if (params.account_list) { + filter = 1; + iterator = list_iterator_create(params.account_list); + while ((account = list_next(iterator))) { + if ((job->account != NULL) && + (strcmp(account, job->account) == 0)) { + filter = 0; + break; + } + } + list_iterator_destroy(iterator); + if (filter == 1) + return 2; + } if (params.state_list) { filter = 1; diff -Nru slurm-llnl-1.3.13/src/squeue/sort.c slurm-llnl-1.3.15/src/squeue/sort.c --- slurm-llnl-1.3.13/src/squeue/sort.c 2008-07-30 19:39:47.000000000 +0200 +++ slurm-llnl-1.3.15/src/squeue/sort.c 2009-01-20 17:19:53.000000000 +0100 @@ -259,8 +259,13 @@ int diff; job_info_t *job1 = (job_info_t *) void1; job_info_t *job2 = (job_info_t *) void2; + char *val1 = "", *val2 = ""; - diff = strcmp(job1->name, job2->name); + if (job1->name) + val1 = job1->name; + if (job2->name) + val2 = job2->name; + diff = strcmp(val1, val2); if (reverse_order) diff = -diff; @@ -565,8 +570,13 @@ int diff; job_info_t *job1 = (job_info_t *) void1; job_info_t *job2 = (job_info_t *) void2; + char *val1 = "", *val2 = ""; - diff = strcmp(job1->partition, job2->partition); + if (job1->partition) + val1 = job1->partition; + if (job2->partition) + val2 = job2->partition; + diff = strcmp(val1, val2); if (reverse_order) diff = -diff; @@ -696,8 +706,13 @@ int diff; job_step_info_t *step1 = (job_step_info_t *) void1; job_step_info_t *step2 = (job_step_info_t *) void2; + char *val1 = "", *val2 = ""; - diff = strcmp(step1->partition, step2->partition); + if (step1->partition) + val1 = step1->partition; + if (step2->partition) + val2 = step2->partition; + diff = strcmp(val1, val2); if (reverse_order) diff = -diff; diff -Nru slurm-llnl-1.3.13/src/squeue/squeue.c slurm-llnl-1.3.15/src/squeue/squeue.c --- slurm-llnl-1.3.13/src/squeue/squeue.c 2008-07-09 19:48:23.000000000 +0200 +++ slurm-llnl-1.3.15/src/squeue/squeue.c 2009-03-23 22:37:58.000000000 +0100 @@ -63,13 +63,14 @@ ************/ static int _get_window_width( void ); static void _print_date( void ); -static void _print_job (void); -static void _print_job_steps( void ); +static int _print_job (void); +static int _print_job_steps( void ); int main (int argc, char *argv[]) { log_options_t opts = LOG_OPTS_STDERR_ONLY ; + int error_code = SLURM_SUCCESS; log_init(xbasename(argv[0]), opts, SYSLOG_FACILITY_USER, NULL); parse_command_line( argc, argv ); @@ -86,9 +87,9 @@ _print_date (); if ( params.step_flag ) - _print_job_steps( ); - else - _print_job( ); + error_code = _print_job_steps( ); + else + error_code = _print_job( ); if ( params.iterate ) { printf( "\n"); @@ -98,7 +99,10 @@ break; } - exit (0); + if ( error_code != SLURM_SUCCESS ) + exit (error_code); + else + exit (0); } /* get_window_width - return the size of the window STDOUT goes to */ @@ -126,7 +130,7 @@ /* _print_job - print the specified job's information */ -static void +static int _print_job ( void ) { static job_info_msg_t * old_job_ptr = NULL, * new_job_ptr; @@ -167,7 +171,7 @@ if (error_code) { slurm_perror ("slurm_load_jobs error"); - return; + return SLURM_ERROR; } old_job_ptr = new_job_ptr; if (job_id) @@ -189,12 +193,12 @@ print_jobs_array( new_job_ptr->job_array, new_job_ptr->record_count , params.format_list ) ; - return; + return SLURM_SUCCESS; } /* _print_job_step - print the specified job step's information */ -static void +static int _print_job_steps( void ) { int error_code; @@ -220,7 +224,7 @@ &new_step_ptr, show_flags); if (error_code) { slurm_perror ("slurm_get_job_steps error"); - return; + return SLURM_ERROR; } old_step_ptr = new_step_ptr; @@ -236,7 +240,7 @@ print_steps_array( new_step_ptr->job_steps, new_step_ptr->job_step_count, params.format_list ); - return; + return SLURM_SUCCESS; } diff -Nru slurm-llnl-1.3.13/src/squeue/squeue.h slurm-llnl-1.3.15/src/squeue/squeue.h --- slurm-llnl-1.3.13/src/squeue/squeue.h 2008-10-15 00:38:17.000000000 +0200 +++ slurm-llnl-1.3.15/src/squeue/squeue.h 2009-01-29 19:16:08.000000000 +0100 @@ -86,6 +86,7 @@ char* jobs; hostset_t nodes; char* partitions; + char* accounts; char* states; char* steps; char* users; @@ -94,6 +95,7 @@ List job_list; List part_list; + List account_list; List state_list; List step_list; List user_list; diff -Nru slurm-llnl-1.3.13/src/sreport/cluster_reports.c slurm-llnl-1.3.15/src/sreport/cluster_reports.c --- slurm-llnl-1.3.13/src/sreport/cluster_reports.c 2008-12-23 19:19:50.000000000 +0100 +++ slurm-llnl-1.3.15/src/sreport/cluster_reports.c 2009-02-26 01:15:51.000000000 +0100 @@ -103,10 +103,7 @@ } } - if(!end && !strncasecmp(argv[i], "where", - MAX(command_len, 5))) { - continue; - } else if(!end && !strncasecmp(argv[i], "all_clusters", + if(!end && !strncasecmp(argv[i], "all_clusters", MAX(command_len, 1))) { local_cluster_flag = 1; } else if(!end && !strncasecmp(argv[i], "withdeleted", @@ -115,8 +112,6 @@ set = 1; } else if(!end || !strncasecmp (argv[i], "WCKeys", - MAX(command_len, 3)) - || !strncasecmp (argv[i], "Names", MAX(command_len, 3))) { if(!wckey_cond->name_list) wckey_cond->name_list = @@ -212,13 +207,7 @@ } } - if (!strncasecmp (argv[i], "Set", MAX(command_len, 3))) { - i--; - break; - } else if(!end && !strncasecmp(argv[i], "where", - MAX(command_len, 5))) { - continue; - } else if(!end && !strncasecmp(argv[i], "all_clusters", + if(!end && !strncasecmp(argv[i], "all_clusters", MAX(command_len, 1))) { local_cluster_flag = 1; } else if (!end && !strncasecmp (argv[i], "Tree", @@ -319,19 +308,11 @@ } } - if (!strncasecmp (argv[i], "Set", MAX(command_len, 3))) { - i--; - break; - } else if(!end && !strncasecmp(argv[i], "where", - MAX(command_len, 5))) { - continue; - } else if(!end && !strncasecmp(argv[i], "all_clusters", + if(!end && !strncasecmp(argv[i], "all_clusters", MAX(command_len, 1))) { local_cluster_flag = 1; } else if(!end || !strncasecmp (argv[i], "Clusters", - MAX(command_len, 1)) - || !strncasecmp (argv[i], "Names", MAX(command_len, 1))) { slurm_addto_char_list(cluster_cond->cluster_list, argv[i]+end); diff -Nru slurm-llnl-1.3.13/src/sreport/job_reports.c slurm-llnl-1.3.15/src/sreport/job_reports.c --- slurm-llnl-1.3.13/src/sreport/job_reports.c 2008-12-12 22:04:09.000000000 +0100 +++ slurm-llnl-1.3.15/src/sreport/job_reports.c 2009-04-02 01:06:12.000000000 +0200 @@ -75,12 +75,14 @@ PRINT_JOB_DUR, PRINT_JOB_NODES, PRINT_JOB_SIZE, - PRINT_JOB_USER + PRINT_JOB_USER, + PRINT_JOB_WCKEY }; static List print_fields_list = NULL; /* types are of print_field_t */ static List grouping_print_fields_list = NULL; /* types are of print_field_t */ static int print_job_count = 0; +static bool flat_view = false; static void _destroy_local_grouping(void *object) { @@ -113,6 +115,58 @@ } } +/* + * Comparator used for sorting clusters alphabetically + * + * returns: 1: cluster_a > cluster_b + * 0: cluster_a == cluster_b + * -1: cluster_a < cluster_b + * + */ +extern int _sort_cluster_grouping_dec(cluster_grouping_t *cluster_a, + cluster_grouping_t *cluster_b) +{ + int diff = 0; + + if(!cluster_a->cluster || !cluster_b->cluster) + return 0; + + diff = strcmp(cluster_a->cluster, cluster_b->cluster); + + if (diff > 0) + return 1; + else if (diff < 0) + return -1; + + return 0; +} + +/* + * Comparator used for sorting clusters alphabetically + * + * returns: 1: acct_a > acct_b + * 0: acct_a == acct_b + * -1: acct_a < acct_b + * + */ +extern int _sort_acct_grouping_dec(acct_grouping_t *acct_a, + acct_grouping_t *acct_b) +{ + int diff = 0; + + if(!acct_a->acct || !acct_b->acct) + return 0; + + diff = strcmp(acct_a->acct, acct_b->acct); + + if (diff > 0) + return 1; + else if (diff < 0) + return -1; + + return 0; +} + /* returns number of objects added to list */ extern int _addto_uid_char_list(List char_list, char *names) { @@ -237,13 +291,7 @@ else command_len=end-1; - if (!strncasecmp (argv[i], "Set", MAX(command_len, 3))) { - i--; - break; - } else if(!end && !strncasecmp(argv[i], "where", - MAX(command_len, 5))) { - continue; - } else if(!end && !strncasecmp(argv[i], "all_clusters", + if(!end && !strncasecmp(argv[i], "all_clusters", MAX(command_len, 1))) { local_cluster_flag = 1; continue; @@ -251,6 +299,10 @@ MAX(command_len, 2))) { print_job_count = 1; continue; + } else if (!end && !strncasecmp (argv[i], "FlatView", + MAX(command_len, 2))) { + flat_view = true; + continue; } else if(!end || !strncasecmp (argv[i], "Clusters", MAX(command_len, 1))) { @@ -277,7 +329,7 @@ job_cond->usage_end = parse_time(argv[i]+end, 1); set = 1; } else if (!strncasecmp (argv[i], "Format", - MAX(command_len, 1))) { + MAX(command_len, 2))) { if(format_list) slurm_addto_char_list(format_list, argv[i]+end); } else if (!strncasecmp (argv[i], "Gid", MAX(command_len, 2))) { @@ -345,6 +397,14 @@ _addto_uid_char_list(job_cond->userid_list, argv[i]+end); set = 1; + } else if (!strncasecmp (argv[i], "Wckeys", + MAX(command_len, 2))) { + if(!job_cond->wckey_list) + job_cond->wckey_list = + list_create(slurm_destroy_char); + slurm_addto_char_list(job_cond->wckey_list, + argv[i]+end); + set = 1; } else { exit_code=1; fprintf(stderr, " Unknown condition: %s\n" @@ -444,6 +504,12 @@ field->name = xstrdup("User"); field->len = 9; field->print_routine = print_fields_str; + } else if(!strncasecmp("Wckey", object, + MAX(command_len, 1))) { + field->type = PRINT_JOB_WCKEY; + field->name = xstrdup("Wckey"); + field->len = 9; + field->print_routine = print_fields_str; } else { exit_code=1; fprintf(stderr, " Unknown field '%s'\n", object); @@ -610,20 +676,6 @@ goto end_it; } - memset(&assoc_cond, 0, sizeof(acct_association_cond_t)); - assoc_cond.id_list = job_cond->associd_list; - assoc_cond.cluster_list = job_cond->cluster_list; - assoc_cond.partition_list = job_cond->partition_list; - if(!job_cond->acct_list || !list_count(job_cond->acct_list)) { - job_cond->acct_list = list_create(NULL); - list_append(job_cond->acct_list, "root"); - } - assoc_cond.parent_acct_list = job_cond->acct_list; - - - assoc_list = acct_storage_g_get_associations(db_conn, my_uid, - &assoc_cond); - if(print_fields_have_header) { char start_char[20]; char end_char[20]; @@ -651,6 +703,22 @@ cluster_itr = list_iterator_create(cluster_list); group_itr = list_iterator_create(grouping_list); + if(flat_view) + goto no_assocs; + + memset(&assoc_cond, 0, sizeof(acct_association_cond_t)); + assoc_cond.id_list = job_cond->associd_list; + assoc_cond.cluster_list = job_cond->cluster_list; + /* don't limit associations to having the partition_list */ + //assoc_cond.partition_list = job_cond->partition_list; + if(!job_cond->acct_list || !list_count(job_cond->acct_list)) { + job_cond->acct_list = list_create(NULL); + list_append(job_cond->acct_list, "root"); + } + assoc_cond.parent_acct_list = job_cond->acct_list; + assoc_list = acct_storage_g_get_associations(db_conn, my_uid, + &assoc_cond); + if(!assoc_list) { debug2(" No assoc list given.\n"); goto no_assocs; @@ -747,19 +815,21 @@ /* here we are only looking for groups that * were added with the associations above */ - continue; -/* cluster_group = */ -/* xmalloc(sizeof(cluster_grouping_t)); */ -/* cluster_group->cluster = xstrdup(local_cluster); */ -/* cluster_group->acct_list = */ -/* list_create(_destroy_acct_grouping); */ -/* list_append(cluster_list, cluster_group); */ + if(!flat_view) + continue; + cluster_group = + xmalloc(sizeof(cluster_grouping_t)); + cluster_group->cluster = xstrdup(local_cluster); + cluster_group->acct_list = + list_create(_destroy_acct_grouping); + list_append(cluster_list, cluster_group); } acct_itr = list_iterator_create(cluster_group->acct_list); while((acct_group = list_next(acct_itr))) { - if(acct_group->lft != (uint32_t)NO_VAL - && job->lft != (uint32_t)NO_VAL) { + if(!flat_view + && (acct_group->lft != (uint32_t)NO_VAL) + && (job->lft != (uint32_t)NO_VAL)) { /* keep separate since we don't want * to so a strcmp if we don't have to */ @@ -772,34 +842,36 @@ list_iterator_destroy(acct_itr); if(!acct_group) { - //char *group = NULL; - //uint32_t last_size = 0; + char *group = NULL; + uint32_t last_size = 0; /* here we are only looking for groups that * were added with the associations above */ - continue; -/* acct_group = xmalloc(sizeof(acct_grouping_t)); */ -/* acct_group->acct = xstrdup(local_account); */ -/* acct_group->groups = */ -/* list_create(_destroy_local_grouping); */ -/* list_append(cluster_group->acct_list, acct_group); */ - -/* while((group = list_next(group_itr))) { */ -/* local_group = xmalloc(sizeof(local_grouping_t)); */ -/* local_group->jobs = list_create(NULL); */ -/* local_group->min_size = last_size; */ -/* last_size = atoi(group); */ -/* local_group->max_size = last_size-1; */ -/* list_append(acct_group->groups, local_group); */ -/* } */ -/* if(last_size) { */ -/* local_group = xmalloc(sizeof(local_grouping_t)); */ -/* local_group->jobs = list_create(NULL); */ -/* local_group->min_size = last_size; */ -/* local_group->max_size = INFINITE; */ -/* list_append(acct_group->groups, local_group); */ -/* } */ -/* list_iterator_reset(group_itr); */ + if(!flat_view) + continue; + + acct_group = xmalloc(sizeof(acct_grouping_t)); + acct_group->acct = xstrdup(local_account); + acct_group->groups = + list_create(_destroy_local_grouping); + list_append(cluster_group->acct_list, acct_group); + + while((group = list_next(group_itr))) { + local_group = xmalloc(sizeof(local_grouping_t)); + local_group->jobs = list_create(NULL); + local_group->min_size = last_size; + last_size = atoi(group); + local_group->max_size = last_size-1; + list_append(acct_group->groups, local_group); + } + if(last_size) { + local_group = xmalloc(sizeof(local_grouping_t)); + local_group->jobs = list_create(NULL); + local_group->min_size = last_size; + local_group->max_size = INFINITE; + list_append(acct_group->groups, local_group); + } + list_iterator_reset(group_itr); } local_itr = list_iterator_create(acct_group->groups); @@ -826,8 +898,12 @@ itr = list_iterator_create(print_fields_list); itr2 = list_iterator_create(grouping_print_fields_list); + list_sort(cluster_list, (ListCmpF)_sort_cluster_grouping_dec); list_iterator_reset(cluster_itr); while((cluster_group = list_next(cluster_itr))) { + + list_sort(cluster_group->acct_list, + (ListCmpF)_sort_acct_grouping_dec); acct_itr = list_iterator_create(cluster_group->acct_list); while((acct_group = list_next(acct_itr))) { @@ -905,9 +981,344 @@ if(assoc_list) { list_destroy(assoc_list); + assoc_list = NULL; + } + + if(cluster_list) { + list_destroy(cluster_list); + cluster_list = NULL; + } + + if(print_fields_list) { + list_destroy(print_fields_list); + print_fields_list = NULL; + } + + if(grouping_print_fields_list) { + list_destroy(grouping_print_fields_list); + grouping_print_fields_list = NULL; + } + + return rc; +} + +extern int job_sizes_grouped_by_wckey(int argc, char *argv[]) +{ + int rc = SLURM_SUCCESS; + acct_job_cond_t *job_cond = xmalloc(sizeof(acct_job_cond_t)); + acct_wckey_cond_t wckey_cond; + acct_wckey_rec_t *wckey = NULL; + int i=0; + + ListIterator itr = NULL; + ListIterator itr2 = NULL; + ListIterator cluster_itr = NULL; + ListIterator local_itr = NULL; + ListIterator acct_itr = NULL; + ListIterator group_itr = NULL; + + jobacct_job_rec_t *job = NULL; + cluster_grouping_t *cluster_group = NULL; + acct_grouping_t *acct_group = NULL; + local_grouping_t *local_group = NULL; + + print_field_t *field = NULL; + print_field_t total_field; + uint32_t total_time = 0; + sreport_time_format_t temp_format; + + List job_list = NULL; + List cluster_list = NULL; + List wckey_list = NULL; + + List format_list = list_create(slurm_destroy_char); + List grouping_list = list_create(slurm_destroy_char); + + List header_list = list_create(NULL); + +// sreport_time_format_t temp_time_format = time_format; + + print_fields_list = list_create(destroy_print_field); + + _set_cond(&i, argc, argv, job_cond, format_list, grouping_list); + + if(!list_count(format_list)) + slurm_addto_char_list(format_list, "Cl,wc"); + + if(!list_count(grouping_list)) + slurm_addto_char_list(grouping_list, "50,250,500,1000"); + + _setup_print_fields_list(format_list); + list_destroy(format_list); + + _setup_grouping_print_fields_list(grouping_list); + + /* we don't want to actually query by wckeys in the jobs + here since we may be looking for sub accounts of a specific + account. + */ + job_list = jobacct_storage_g_get_jobs_cond(db_conn, my_uid, job_cond); + + if(!job_list) { + exit_code=1; + fprintf(stderr, " Problem with job query.\n"); + goto end_it; + } + + memset(&wckey_cond, 0, sizeof(acct_wckey_cond_t)); + wckey_cond.name_list = job_cond->wckey_list; + wckey_cond.cluster_list = job_cond->cluster_list; + + wckey_list = acct_storage_g_get_wckeys(db_conn, my_uid, &wckey_cond); + + if(print_fields_have_header) { + char start_char[20]; + char end_char[20]; + time_t my_start = job_cond->usage_start; + time_t my_end = job_cond->usage_end-1; + + slurm_make_time_str(&my_start, start_char, sizeof(start_char)); + slurm_make_time_str(&my_end, end_char, sizeof(end_char)); + printf("----------------------------------------" + "----------------------------------------\n"); + printf("Job Sizes by Wckey %s - %s (%d secs)\n", + start_char, end_char, + (job_cond->usage_end - job_cond->usage_start)); + if(print_job_count) + printf("Units are in number of jobs ran\n"); + else + printf("Time reported in %s\n", time_format_string); + printf("----------------------------------------" + "----------------------------------------\n"); + } + total_time = job_cond->usage_end - job_cond->usage_start; + + cluster_list = list_create(_destroy_cluster_grouping); + + cluster_itr = list_iterator_create(cluster_list); + group_itr = list_iterator_create(grouping_list); + + if(!wckey_list) { + debug2(" No wckey list given.\n"); + goto no_assocs; + } + + itr = list_iterator_create(wckey_list); + while((wckey = list_next(itr))) { + while((cluster_group = list_next(cluster_itr))) { + if(!strcmp(wckey->cluster, cluster_group->cluster)) + break; + } + if(!cluster_group) { + cluster_group = + xmalloc(sizeof(cluster_grouping_t)); + cluster_group->cluster = xstrdup(wckey->cluster); + cluster_group->acct_list = + list_create(_destroy_acct_grouping); + list_append(cluster_list, cluster_group); + } + + acct_itr = list_iterator_create(cluster_group->acct_list); + while((acct_group = list_next(acct_itr))) { + if(!strcmp(wckey->name, acct_group->acct)) + break; + } + list_iterator_destroy(acct_itr); + + if(!acct_group) { + uint32_t last_size = 0; + char *group = NULL; + acct_group = xmalloc(sizeof(acct_grouping_t)); + acct_group->acct = xstrdup(wckey->name); + acct_group->lft = wckey->id; + acct_group->groups = + list_create(_destroy_local_grouping); + list_append(cluster_group->acct_list, acct_group); + while((group = list_next(group_itr))) { + local_group = xmalloc(sizeof(local_grouping_t)); + local_group->jobs = list_create(NULL); + local_group->min_size = last_size; + last_size = atoi(group); + local_group->max_size = last_size-1; + list_append(acct_group->groups, local_group); + } + if(last_size) { + local_group = xmalloc(sizeof(local_grouping_t)); + local_group->jobs = list_create(NULL); + local_group->min_size = last_size; + local_group->max_size = INFINITE; + list_append(acct_group->groups, local_group); + } + list_iterator_reset(group_itr); + } + list_iterator_reset(cluster_itr); + } + list_iterator_destroy(itr); +no_assocs: + itr = list_iterator_create(job_list); + + list_append_list(header_list, print_fields_list); + list_append_list(header_list, grouping_print_fields_list); + + memset(&total_field, 0, sizeof(print_field_t)); + total_field.type = PRINT_JOB_SIZE; + total_field.name = xstrdup("% of cluster"); + total_field.len = 12; + total_field.print_routine = sreport_print_time; + list_append(header_list, &total_field); + + print_fields_header(header_list); + list_destroy(header_list); + + while((job = list_next(itr))) { + char *local_cluster = "UNKNOWN"; + char *local_account = "UNKNOWN"; + + if(!job->elapsed) { + /* here we don't care about jobs that didn't + * really run here */ + continue; + } + if(job->cluster) + local_cluster = job->cluster; + if(job->account) + local_account = job->account; + + list_iterator_reset(cluster_itr); + while((cluster_group = list_next(cluster_itr))) { + if(!strcmp(local_cluster, cluster_group->cluster)) + break; + } + if(!cluster_group) { + /* here we are only looking for groups that + * were added with the associations above + */ + continue; + } + + acct_itr = list_iterator_create(cluster_group->acct_list); + while((acct_group = list_next(acct_itr))) { + if(!strcmp(job->wckey, acct_group->acct)) + break; + } + list_iterator_destroy(acct_itr); + + if(!acct_group) { + /* here we are only looking for groups that + * were added with the associations above + */ + continue; + } + + local_itr = list_iterator_create(acct_group->groups); + while((local_group = list_next(local_itr))) { + uint64_t total_secs = 0; + if((job->alloc_cpus < local_group->min_size) + || (job->alloc_cpus > local_group->max_size)) + continue; + list_append(local_group->jobs, job); + local_group->count++; + total_secs = (uint64_t)job->elapsed + * (uint64_t)job->alloc_cpus; + local_group->cpu_secs += total_secs; + acct_group->cpu_secs += total_secs; + cluster_group->cpu_secs += total_secs; + } + list_iterator_destroy(local_itr); + } + list_iterator_destroy(group_itr); + list_destroy(grouping_list); + list_iterator_destroy(itr); + +// time_format = SREPORT_TIME_PERCENT; + + itr = list_iterator_create(print_fields_list); + itr2 = list_iterator_create(grouping_print_fields_list); + list_sort(cluster_list, (ListCmpF)_sort_cluster_grouping_dec); + list_iterator_reset(cluster_itr); + while((cluster_group = list_next(cluster_itr))) { + list_sort(cluster_group->acct_list, + (ListCmpF)_sort_acct_grouping_dec); + acct_itr = list_iterator_create(cluster_group->acct_list); + while((acct_group = list_next(acct_itr))) { + + while((field = list_next(itr))) { + switch(field->type) { + case PRINT_JOB_CLUSTER: + field->print_routine( + field, + cluster_group->cluster, 0); + break; + case PRINT_JOB_WCKEY: + field->print_routine(field, + acct_group->acct, + 0); + break; + default: + field->print_routine(field, + NULL, + 0); + break; + } + } + list_iterator_reset(itr); + local_itr = list_iterator_create(acct_group->groups); + while((local_group = list_next(local_itr))) { + field = list_next(itr2); + switch(field->type) { + case PRINT_JOB_SIZE: + field->print_routine( + field, + local_group->cpu_secs, + acct_group->cpu_secs, + 0); + break; + case PRINT_JOB_COUNT: + field->print_routine( + field, + local_group->count, + 0); + break; + default: + field->print_routine(field, + NULL, + 0); + break; + } + } + list_iterator_reset(itr2); + list_iterator_destroy(local_itr); + + temp_format = time_format; + time_format = SREPORT_TIME_PERCENT; + total_field.print_routine(&total_field, + acct_group->cpu_secs, + cluster_group->cpu_secs, 1); + time_format = temp_format; + printf("\n"); + } + list_iterator_destroy(acct_itr); + } + list_iterator_destroy(itr); + +// time_format = temp_time_format; + +end_it: + if(print_job_count) + print_job_count = 0; + + destroy_acct_job_cond(job_cond); + + if(job_list) { + list_destroy(job_list); job_list = NULL; } + if(wckey_list) { + list_destroy(wckey_list); + wckey_list = NULL; + } + if(cluster_list) { list_destroy(cluster_list); cluster_list = NULL; diff -Nru slurm-llnl-1.3.13/src/sreport/job_reports.h slurm-llnl-1.3.15/src/sreport/job_reports.h --- slurm-llnl-1.3.13/src/sreport/job_reports.h 2008-06-18 01:01:05.000000000 +0200 +++ slurm-llnl-1.3.15/src/sreport/job_reports.h 2009-03-19 21:54:03.000000000 +0100 @@ -43,5 +43,6 @@ #include "sreport.h" extern int job_sizes_grouped_by_top_acct(int argc, char *argv[]); +extern int job_sizes_grouped_by_wckey(int argc, char *argv[]); #endif diff -Nru slurm-llnl-1.3.13/src/sreport/sreport.c slurm-llnl-1.3.15/src/sreport/sreport.c --- slurm-llnl-1.3.13/src/sreport/sreport.c 2008-12-30 21:48:54.000000000 +0100 +++ slurm-llnl-1.3.15/src/sreport/sreport.c 2009-04-02 01:06:12.000000000 +0200 @@ -75,7 +75,7 @@ int error_code = SLURM_SUCCESS, i, opt_char, input_field_count; char **input_fields; log_options_t opts = LOG_OPTS_STDERR_ONLY ; - + char *temp = NULL; int option_index; static struct option long_options[] = { {"all_clusters", 0, 0, 'a'}, @@ -99,6 +99,20 @@ quiet_flag = 0; log_init("sreport", opts, SYSLOG_FACILITY_DAEMON, NULL); + /* Check to see if we are running a supported accounting plugin */ + temp = slurm_get_accounting_storage_type(); + if(strcasecmp(temp, "accounting_storage/slurmdbd") + && strcasecmp(temp, "accounting_storage/mysql")) { + fprintf (stderr, "You are not running a supported " + "accounting_storage plugin\n(%s).\n" + "Only 'accounting_storage/slurmdbd' " + "and 'accounting_storage/mysql' are supported.\n", + temp); + xfree(temp); + exit(1); + } + xfree(temp); + while((opt_char = getopt_long(argc, argv, "ahnpPqs:t:vV", long_options, &option_index)) != -1) { switch (opt_char) { @@ -211,16 +225,22 @@ static void _job_rep (int argc, char *argv[]) { int error_code = SLURM_SUCCESS; + int command_len = strlen(argv[0]); - /* First identify the entity to add */ - if (strncasecmp (argv[0], "Sizes", 1) == 0) { + /* For backwards compatibility we just look at the 1st char + * by default since Sizes was the original name */ + if (!strncasecmp (argv[0], "SizesByAccount", MAX(command_len, 1))) { error_code = job_sizes_grouped_by_top_acct( (argc - 1), &argv[1]); + } else if (!strncasecmp (argv[0], + "SizesByWcKey", MAX(command_len, 8))) { + error_code = job_sizes_grouped_by_wckey( + (argc - 1), &argv[1]); } else { exit_code = 1; fprintf(stderr, "Not valid report %s\n", argv[0]); fprintf(stderr, "Valid job reports are, "); - fprintf(stderr, "\"Sizes\"\n"); + fprintf(stderr, "\"SizesByAccount, and SizesByWckey\"\n"); } if (error_code) { @@ -519,13 +539,13 @@ if (strncasecmp (format, "SecPer", MAX(command_len, 6)) == 0) { time_format = SREPORT_TIME_SECS_PER; - time_format_string = "Seconds/Percentange of Total"; + time_format_string = "Seconds/Percentage of Total"; } else if (strncasecmp (format, "MinPer", MAX(command_len, 6)) == 0) { time_format = SREPORT_TIME_MINS_PER; - time_format_string = "Minutes/Percentange of Total"; + time_format_string = "Minutes/Percentage of Total"; } else if (strncasecmp (format, "HourPer", MAX(command_len, 6)) == 0) { time_format = SREPORT_TIME_HOURS_PER; - time_format_string = "Hours/Percentange of Total"; + time_format_string = "Hours/Percentage of Total"; } else if (strncasecmp (format, "Seconds", MAX(command_len, 1)) == 0) { time_format = SREPORT_TIME_SECS; time_format_string = "Seconds"; @@ -537,7 +557,7 @@ time_format_string = "Hours"; } else if (strncasecmp (format, "Percent", MAX(command_len, 1)) == 0) { time_format = SREPORT_TIME_PERCENT; - time_format_string = "Percentange of Total"; + time_format_string = "Percentage of Total"; } else { fprintf (stderr, "unknown time format %s", format); return SLURM_ERROR; @@ -602,7 +622,7 @@ is different for each report type. \n\ cluster - AccountUtilizationByUser, UserUtilizationByAccount, \n\ UserUtilizationByWckey, Utilization, WCKeyUtilizationByUser \n\ - job - Sizes \n\ + job - SizesByAccount, SizesByWckey \n\ user - TopUsage \n\ \n\ are different for each report type. \n\ @@ -610,6 +630,8 @@ COMMON FOR ALL TYPES \n\ - All_Clusters - Use all monitored clusters default is \n\ local cluster. \n\ + - Clusters= - List of clusters to include in report \n\ + Default is local cluster. \n\ - End= - Period ending for report. \n\ Default is 23:59:59 of previous day. \n\ - Format= - Comma separated list of fields to display\n\ @@ -617,17 +639,35 @@ - Start= - Period start for report. \n\ Default is 00:00:00 of previous day. \n\ \n\ - cluster - Names= - List of clusters to include in report \n\ - Default is local cluster. \n\ + cluster - Accounts= - When used with the UserUtilizationByAccount,\n\ + or AccountUtilizationByUser, List of accounts\n\ + to include in report. Default is all. \n\ - Tree - When used with the AccountUtilizationByUser\n\ report will span the accounts as they \n\ in the hierarchy. \n\ + - Users= - When used with any report other than \n\ + Utilization, List of users to include in \n\ + report. Default is all. \n\ + - Wckeys= - When used with the UserUtilizationByWckey\n\ + or WCKeyUtilizationByUser, List of wckeys\n\ + to include in report. Default is all. \n\ \n\ job - Accounts= - List of accounts to use for the report \n\ - Default is all. \n\ - - Clusters= - List of clusters to include in report. \n\ - Default is local cluster. \n\ - - GID= - List of group ids to include in report \n\ + Default is all. The SizesbyAccount \n\ + report only displays 1 hierarchical level.\n\ + If accounts are specified the next layer \n\ + of accounts under those specified will be\n\ + displayed, not the accounts specified. \n\ + In the SizesByAccount reports the default\n\ + for accounts is root. This explanation \n\ + does not apply when ran with the FlatView\n\ + option. \n\ + - FlatView - When used with the SizesbyAccount \n\ + will not group accounts in a \n\ + hierarchical level, but print each \n\ + account where jobs ran on a separate \n\ + line without any hierarchy. \n\ + - GID= - List of group ids to include in report. \n\ Default is all. \n\ - Grouping= - Comma separated list of size groupings. \n\ (i.e. 50,100,150 would group job cpu count\n\ @@ -636,15 +676,19 @@ Default is all. \n\ - Partitions= - List of partitions jobs ran on to include\n\ in report. Default is all. \n\ - - PrintJobCount - When used with the Sizes report will print\n\ - number of jobs ran instead of time used. \n\ + - PrintJobCount - When used with the any Sizes report \n\ + will print number of jobs ran instead of \n\ + time used. \n\ - Users= - List of users jobs to include in report. \n\ Default is all. \n\ + - Wckeys= - List of wckeys to use for the report. \n\ + Default is all. The SizesbyWckey \n\ + report all users summed together. If \n\ + you want only certain users specify them \n\ + them with the Users= option. \n\ \n\ user - Accounts= - List of accounts to use for the report \n\ Default is all. \n\ - - Clusters= - List of clusters to include in report. \n\ - Default is local cluster. \n\ - Group - Group all accounts together for each user.\n\ Default is a separate entry for each user\n\ and account reference. \n\ @@ -654,6 +698,35 @@ Default is all. \n\ \n\ \n\ + Below are the format options for each report. \n\ + \n\ + Cluster \n\ + - AccountUtilizationByUser \n\ + - UserUtilizationByAccount \n\ + - Accounts, Cluster, CPUCount, Login, Proper, Used \n\ + - UserUtilizationByWckey \n\ + - WCKeyUtilizationByUser \n\ + - Cluster, CPUCount, Login, Proper, Used, Wckey \n\ + - Utilization \n\ + - Allocated, Cluster, CPUCount, Down, Idle, Overcommited, \n\ + Reported, Reserved \n\ + \n\ + Job \n\ + - Sizes \n\ + - Account, Cluster \n\ + \n\ + User \n\ + - TopUsage \n\ + - Account, Cluster, Login, Proper, Used \n\ + \n\ + \n\ + \n\ + Note, valid start/end time formats are... \n\ + HH:MM[:SS] [AM|PM] \n\ + MMDD[YY] or MM/DD[/YY] or MM.DD[.YY] \n\ + MM/DD[/YY]-HH:MM[:SS] \n\ + \n\ + \n\ All commands and options are case-insensitive. \n\n"); } diff -Nru slurm-llnl-1.3.13/src/sreport/user_reports.c slurm-llnl-1.3.15/src/sreport/user_reports.c --- slurm-llnl-1.3.13/src/sreport/user_reports.c 2008-12-30 20:05:06.000000000 +0100 +++ slurm-llnl-1.3.15/src/sreport/user_reports.c 2009-03-09 23:52:06.000000000 +0100 @@ -75,7 +75,7 @@ user_cond->assoc_cond->with_usage = 1; } assoc_cond = user_cond->assoc_cond; - + if(!assoc_cond->cluster_list) assoc_cond->cluster_list = list_create(slurm_destroy_char); for (i=(*start); iassoc_cond->without_parent_info = 1; + if(!list_count(format_list)) slurm_addto_char_list(format_list, "Cl,L,P,A,U"); diff -Nru slurm-llnl-1.3.13/src/srun/opt.c slurm-llnl-1.3.15/src/srun/opt.c --- slurm-llnl-1.3.13/src/srun/opt.c 2008-12-03 00:38:47.000000000 +0100 +++ slurm-llnl-1.3.15/src/srun/opt.c 2009-03-17 19:50:45.000000000 +0100 @@ -748,6 +748,7 @@ {"SLURM_KILL_BAD_EXIT", OPT_INT, &opt.kill_bad_exit, NULL }, {"SLURM_LABELIO", OPT_INT, &opt.labelio, NULL }, {"SLURM_LINUX_IMAGE", OPT_STRING, &opt.linuximage, NULL }, +{"SLURM_CNLOAD_IMAGE", OPT_STRING, &opt.linuximage, NULL }, {"SLURM_MLOADER_IMAGE", OPT_STRING, &opt.mloaderimage, NULL }, {"SLURM_NNODES", OPT_NODES, NULL, NULL }, {"SLURM_NSOCKETS_PER_NODE",OPT_NSOCKETS,NULL, NULL }, @@ -758,6 +759,7 @@ {"SLURM_OVERCOMMIT", OPT_OVERCOMMIT, NULL, NULL }, {"SLURM_PARTITION", OPT_STRING, &opt.partition, NULL }, {"SLURM_RAMDISK_IMAGE", OPT_STRING, &opt.ramdiskimage, NULL }, +{"SLURM_IOLOAD_IMAGE", OPT_STRING, &opt.ramdiskimage, NULL }, {"SLURM_REMOTE_CWD", OPT_STRING, &opt.cwd, NULL }, {"SLURM_STDERRMODE", OPT_STRING, &opt.efname, NULL }, {"SLURM_STDINMODE", OPT_STRING, &opt.ifname, NULL }, @@ -1034,8 +1036,10 @@ {"tasks-per-node", required_argument, 0, LONG_OPT_NTASKSPERNODE}, {"blrts-image", required_argument, 0, LONG_OPT_BLRTS_IMAGE}, {"linux-image", required_argument, 0, LONG_OPT_LINUX_IMAGE}, + {"cnload-image", required_argument, 0, LONG_OPT_LINUX_IMAGE}, {"mloader-image", required_argument, 0, LONG_OPT_MLOADER_IMAGE}, {"ramdisk-image", required_argument, 0, LONG_OPT_RAMDISK_IMAGE}, + {"ioload-image", required_argument, 0, LONG_OPT_RAMDISK_IMAGE}, {"reboot", no_argument, 0, LONG_OPT_REBOOT}, {"get-user-env", optional_argument, 0, LONG_OPT_GET_USER_ENV}, {"pty", no_argument, 0, LONG_OPT_PTY}, @@ -2152,14 +2156,24 @@ info("reboot : %s", opt.reboot ? "no" : "yes"); info("rotate : %s", opt.no_rotate ? "yes" : "no"); +#ifdef HAVE_BGL if (opt.blrtsimage) info("BlrtsImage : %s", opt.blrtsimage); +#endif if (opt.linuximage) +#ifdef HAVE_BGL info("LinuxImage : %s", opt.linuximage); +#else + info("CnloadImage : %s", opt.linuximage); +#endif if (opt.mloaderimage) info("MloaderImage : %s", opt.mloaderimage); if (opt.ramdiskimage) +#ifdef HAVE_BGL info("RamDiskImage : %s", opt.ramdiskimage); +#else + info("IoloadImage : %s", opt.ramdiskimage); +#endif info("network : %s", opt.network); info("propagate : %s", @@ -2217,14 +2231,19 @@ " [--ntasks-per-core=n] [--mem-per-cpu=MB]\n" #ifdef HAVE_BG /* Blue gene specific options */ " [--geometry=XxYxZ] [--conn-type=type] [--no-rotate] [--reboot]\n" +#ifdef HAVE_BGL " [--blrts-image=path] [--linux-image=path]\n" " [--mloader-image=path] [--ramdisk-image=path]\n" +#else +" [--cnload-image=path]\n" +" [--mloader-image=path] [--ioload-image=path]\n" #endif - " [--mail-type=type] [--mail-user=user] [--nice[=value]]\n" - " [--prolog=fname] [--epilog=fname]\n" - " [--task-prolog=fname] [--task-epilog=fname]\n" - " [--ctrl-comm-ifhn=addr] [--multi-prog]\n" - " [-w hosts...] [-x hosts...] executable [args...]\n"); +#endif +" [--mail-type=type] [--mail-user=user] [--nice[=value]]\n" +" [--prolog=fname] [--epilog=fname]\n" +" [--task-prolog=fname] [--task-epilog=fname]\n" +" [--ctrl-comm-ifhn=addr] [--multi-prog]\n" +" [-w hosts...] [-x hosts...] executable [args...]\n"); } static void _help(void) @@ -2358,12 +2377,22 @@ " --reboot reboot block before starting job\n" " --conn-type=type constraint on type of connection, MESH or TORUS\n" " if not set, then tries to fit TORUS else MESH\n" - " --blrts-image=path path to blrts image for bluegene block. Default if not set\n" - " --linux-image=path path to linux image for bluegene block. Default if not set\n" - " --mloader-image=path path to mloader image for bluegene block. Default if not set\n" - " --ramdisk-image=path path to ramdisk image for bluegene block. Default if not set\n" - "\n" +#ifndef HAVE_BGL + " If wanting to run in HTC mode (only for 1\n" + " midplane and below). You can use HTC_S for\n" + " SMP, HTC_D for Dual, HTC_V for\n" + " virtual node mode, and HTC_L for Linux mode.\n" + " --cnload-image=path path to compute node image for bluegene block. Default if not set\n" + " --mloader-image=path path to mloader image for bluegene block. Default if not set\n" + " --ioload-image=path path to ioload image for bluegene block. Default if not set\n" +#else + " --blrts-image=path path to blrts image for bluegene block. Default if not set\n" + " --linux-image=path path to linux image for bluegene block. Default if not set\n" + " --mloader-image=path path to mloader image for bluegene block. Default if not set\n" + " --ramdisk-image=path path to ramdisk image for bluegene block. Default if not set\n" #endif +#endif + "\n" "Help options:\n" " --help show this help message\n" " --usage display brief usage message\n" diff -Nru slurm-llnl-1.3.13/src/srun/srun.c slurm-llnl-1.3.15/src/srun/srun.c --- slurm-llnl-1.3.13/src/srun/srun.c 2008-12-04 19:59:39.000000000 +0100 +++ slurm-llnl-1.3.15/src/srun/srun.c 2009-03-30 23:01:49.000000000 +0200 @@ -406,6 +406,7 @@ if (slurm_step_launch(job->step_ctx, slurmctld_comm_addr.hostname, &launch_params, &callbacks) != SLURM_SUCCESS) { error("Application launch failed: %m"); + global_rc = 1; goto cleanup; } diff -Nru slurm-llnl-1.3.13/src/sstat/options.c slurm-llnl-1.3.15/src/sstat/options.c --- slurm-llnl-1.3.13/src/sstat/options.c 2008-12-22 22:46:26.000000000 +0100 +++ slurm-llnl-1.3.15/src/sstat/options.c 2009-02-03 18:30:54.000000000 +0100 @@ -66,6 +66,7 @@ printf("\n" "By default, sstat displays status data for job/step stated\n" "Options:\n" + "-a, --allsteps\n" "-C, --cluster\n" " Job is running on this cluster.\n" "-F , --fields=\n" @@ -265,6 +266,7 @@ log_options_t logopt = LOG_OPTS_STDERR_ONLY; static struct option long_options[] = { + {"allsteps", 0, 0, 'a'}, {"cluster", 1, 0, 'C'}, {"fields", 1, 0, 'F'}, {"help", 0, ¶ms.opt_help, 1}, @@ -283,11 +285,14 @@ opterr = 1; /* Let getopt report problems to the user */ while (1) { /* now cycle through the command line */ - c = getopt_long(argc, argv, "F:hj:Vv", + c = getopt_long(argc, argv, "aF:hj:Vv", long_options, &optionIndex); if (c == -1) break; switch (c) { + case 'a': + params.opt_all_steps = 1; + break; case 'F': if(params.opt_field_list) xfree(params.opt_field_list); diff -Nru slurm-llnl-1.3.13/src/sstat/sstat.c slurm-llnl-1.3.15/src/sstat/sstat.c --- slurm-llnl-1.3.13/src/sstat/sstat.c 2008-12-22 22:46:26.000000000 +0100 +++ slurm-llnl-1.3.15/src/sstat/sstat.c 2009-02-03 18:30:54.000000000 +0100 @@ -244,7 +244,23 @@ while((selected_step = list_next(itr))) { if(selected_step->stepid != NO_VAL) stepid = selected_step->stepid; - else + else if(params.opt_all_steps) { + job_step_info_response_msg_t *step_ptr = NULL; + int i = 0; + if(slurm_get_job_steps( + 0, selected_step->jobid, 0, + &step_ptr, SHOW_ALL)) { + error("couldn't get steps for job %u", + selected_step->jobid); + continue; + } + for (i = 0; i < step_ptr->job_step_count; i++) { + _do_stat(selected_step->jobid, + step_ptr->job_steps[i].step_id); + } + slurm_free_job_step_info_response_msg(step_ptr); + continue; + } else stepid = 0; _do_stat(selected_step->jobid, stepid); } diff -Nru slurm-llnl-1.3.13/src/sstat/sstat.h slurm-llnl-1.3.15/src/sstat/sstat.h --- slurm-llnl-1.3.13/src/sstat/sstat.h 2008-12-22 22:46:26.000000000 +0100 +++ slurm-llnl-1.3.15/src/sstat/sstat.h 2009-02-03 18:40:24.000000000 +0100 @@ -85,6 +85,7 @@ } type_t; typedef struct { + int opt_all_steps; /* --allsteps */ char *opt_field_list; /* --fields= */ int opt_help; /* --help */ List opt_job_list; /* --jobs */ diff -Nru slurm-llnl-1.3.13/src/sview/block_info.c slurm-llnl-1.3.15/src/sview/block_info.c --- slurm-llnl-1.3.13/src/sview/block_info.c 2008-12-29 22:56:17.000000000 +0100 +++ slurm-llnl-1.3.15/src/sview/block_info.c 2009-02-10 23:15:26.000000000 +0100 @@ -73,7 +73,9 @@ SORTID_RAMDISKIMAGE, SORTID_STATE, SORTID_UPDATED, +#ifdef HAVE_BGL SORTID_USE, +#endif SORTID_USER, SORTID_CNT }; @@ -195,6 +197,7 @@ static char* _convert_conn_type(enum connection_type conn_type) { +#ifdef HAVE_BG switch (conn_type) { case (SELECT_MESH): return "MESH"; @@ -204,7 +207,24 @@ return "SMALL"; case (SELECT_NAV): return "NAV"; +#ifndef HAVE_BGL + case SELECT_HTC_S: + return "HTC_S"; + break; + case SELECT_HTC_D: + return "HTC_D"; + break; + case SELECT_HTC_V: + return "HTC_V"; + break; + case SELECT_HTC_L: + return "HTC_L"; + break; +#endif + default: + return "?"; } +#endif return "?"; } diff -Nru slurm-llnl-1.3.13/src/sview/common.c slurm-llnl-1.3.15/src/sview/common.c --- slurm-llnl-1.3.13/src/sview/common.c 2008-03-20 00:10:58.000000000 +0100 +++ slurm-llnl-1.3.15/src/sview/common.c 2009-02-27 23:19:56.000000000 +0100 @@ -64,8 +64,7 @@ gtk_tree_model_get(model, a, sortcol, &name1, -1); gtk_tree_model_get(model, b, sortcol, &name2, -1); - if (name1 == NULL || name2 == NULL) - { + if (name1 == NULL || name2 == NULL) { if (name1 == NULL && name2 == NULL) goto cleanup; /* both equal => ret = 0 */ @@ -120,6 +119,52 @@ return ret; } +static int _sort_iter_compare_func_nodes(GtkTreeModel *model, + GtkTreeIter *a, + GtkTreeIter *b, + gpointer userdata) +{ + int sortcol = GPOINTER_TO_INT(userdata); + int ret = 0; + gchar *name1 = NULL, *name2 = NULL; + + gtk_tree_model_get(model, a, sortcol, &name1, -1); + gtk_tree_model_get(model, b, sortcol, &name2, -1); + + if (name1 == NULL || name2 == NULL) { + if (name1 == NULL && name2 == NULL) + goto cleanup; /* both equal => ret = 0 */ + + ret = (name1 == NULL) ? -1 : 1; + } else { + uint64_t int1 = atoi(name1); + uint64_t int2 = atoi(name2); + if(strchr(name1, 'K')) { + int1 *= 1024; + } else if(strchr(name1, 'M')) { + int1 *= 1048576; + } else if(strchr(name1, 'G')) { + int1 *= 1073741824; + } + + if(strchr(name2, 'K')) { + int2 *= 1024; + } else if(strchr(name2, 'M')) { + int2 *= 1048576; + } else if(strchr(name2, 'G')) { + int2 *= 1073741824; + } + + if (int1 != int2) + ret = (int1 > int2) ? 1 : -1; + } +cleanup: + g_free(name1); + g_free(name2); + + return ret; +} + static void _editing_started(GtkCellRenderer *cell, GtkCellEditable *editable, const gchar *path, @@ -546,7 +591,7 @@ treestore = gtk_tree_store_newv(count, types); if(!treestore) { - g_error("Can't great treestore.\n"); + g_error("Can't create treestore.\n"); return NULL; } @@ -567,13 +612,23 @@ break; case G_TYPE_STRING: - gtk_tree_sortable_set_sort_func( - GTK_TREE_SORTABLE(treestore), - display_data[i].id, - _sort_iter_compare_func_char, - GINT_TO_POINTER(display_data[i].id), - NULL); - break; + if(!strcasecmp(display_data[i].name, "Nodes")) { + gtk_tree_sortable_set_sort_func( + GTK_TREE_SORTABLE(treestore), + display_data[i].id, + _sort_iter_compare_func_nodes, + GINT_TO_POINTER(display_data[i].id), + NULL); + break; + } else { + gtk_tree_sortable_set_sort_func( + GTK_TREE_SORTABLE(treestore), + display_data[i].id, + _sort_iter_compare_func_char, + GINT_TO_POINTER(display_data[i].id), + NULL); + break; + } default: g_print("unknown type %d", (int)display_data[i].type); diff -Nru slurm-llnl-1.3.13/testsuite/expect/test12.2 slurm-llnl-1.3.15/testsuite/expect/test12.2 --- slurm-llnl-1.3.13/testsuite/expect/test12.2 2008-11-10 22:54:10.000000000 +0100 +++ slurm-llnl-1.3.15/testsuite/expect/test12.2 2009-03-19 17:03:24.000000000 +0100 @@ -269,12 +269,13 @@ set elapsed_time 0 spawn $sacct --noheader --job=$job_id.0 --fields elapsed expect { - -re "($number):(\[0-9\])(\[0-9\])" { - set mins $expect_out(1,string) - set sec_ten $expect_out(2,string) - set sec_one $expect_out(3,string) + -re "($number):($number):(\[0-9\])(\[0-9\])" { + set hours $expect_out(1,string) + set mins $expect_out(2,string) + set sec_ten $expect_out(3,string) + set sec_one $expect_out(4,string) set secs [expr $sec_ten * 10 + $sec_one] - set elapsed_time [expr $mins * 60 + $secs] + set elapsed_time [expr ($hours * 3600) + ($mins * 60) + $secs] exp_continue } timeout { diff -Nru slurm-llnl-1.3.13/testsuite/expect/test1.59 slurm-llnl-1.3.15/testsuite/expect/test1.59 --- slurm-llnl-1.3.13/testsuite/expect/test1.59 2008-05-01 00:15:52.000000000 +0200 +++ slurm-llnl-1.3.15/testsuite/expect/test1.59 2009-01-13 22:55:46.000000000 +0100 @@ -97,7 +97,6 @@ send_user "\nFAILURE: srun not responding\n" kill_srun set exit_code 1 - exp_continue } eof { wait diff -Nru slurm-llnl-1.3.13/testsuite/expect/test19.5 slurm-llnl-1.3.15/testsuite/expect/test19.5 --- slurm-llnl-1.3.13/testsuite/expect/test19.5 2008-08-19 22:39:30.000000000 +0200 +++ slurm-llnl-1.3.15/testsuite/expect/test19.5 2009-03-12 18:38:02.000000000 +0100 @@ -67,7 +67,7 @@ # # Build input script file and submit a job # -make_bash_script $file_in "sleep 60" +make_bash_script $file_in "$srun sleep 60" set job_id 0 spawn $sbatch --output=/dev/null -t2 $file_in