capture_persistent_env_contents: Assertion `r == 0' failed with tokudb_cache_size=1024

[Shahriyar Rzayev]

If you try to start Percona Server(TokuDB enabled) with tokudb_cache_size equal to 1024 (for eg.), it will fail to start.
Output from Error Log:

2015-10-31 19:29:55 15251 [Note] InnoDB: Percona XtraDB (http://www.percona.com) 5.6.26-74.0 started; log sequence number 876490089
/mnt/workspace/percona-server-5.6-debian-binary-new/label_exp/ubuntu-trusty-64bit/percona-server-5.6-5.6.26-74.0/storage/tokudb/PerconaFT/src/y
db.cc:587 capture_persistent_env_contents: Assertion `r == 0' failed (errno=2) (r=18446744073709451616)
: No such file or directory
Backtrace: (Note: toku_do_assert=0x0x7fe37a4d6ba0)
/usr/lib/mysql/plugin/ha_tokudb.so(_Z19db_env_do_backtraceP8_IO_FILE+0x1b)[0x7fe37a4d646b]
/usr/lib/mysql/plugin/ha_tokudb.so(+0x93593)[0x7fe37a4d6593]
/usr/lib/mysql/plugin/ha_tokudb.so(+0xa6602)[0x7fe37a4e9602]
/usr/lib/mysql/plugin/ha_tokudb.so(+0x131e35)[0x7fe37a574e35]
/usr/lib/mysql/plugin/ha_tokudb.so(+0x5102c)[0x7fe37a49402c]
/usr/sbin/mysqld(_Z24ha_initialize_handlertonP13st_plugin_int+0x48)[0x599a78]
/usr/sbin/mysqld[0x6e4ba1]
/usr/sbin/mysqld(_Z11plugin_initPiPPci+0x941)[0x6eb1f1]
/usr/sbin/mysqld[0x59236d]
/usr/sbin/mysqld(_Z11mysqld_mainiPPc+0x42c)[0x5933fc]
/lib/x86_64-linux-gnu/libc.so.6(__libc_start_main+0xf5)[0x7fe396d30ec5]
/usr/sbin/mysqld[0x5856ed]
.
.
.
Thread pointer: 0x0
Attempting backtrace. You can use the following information to find out
where mysqld died. If you see no messages after this, something went
terribly wrong...
stack_bottom = 0 thread_stack 0x30000
/usr/sbin/mysqld(my_print_stacktrace+0x2c)[0x8d037c]
/usr/sbin/mysqld(handle_fatal_signal+0x461)[0x6570f1]
/lib/x86_64-linux-gnu/libpthread.so.0(+0x10340)[0x7fe397904340]
/lib/x86_64-linux-gnu/libc.so.6(gsignal+0x39)[0x7fe396d45cc9]
/lib/x86_64-linux-gnu/libc.so.6(abort+0x148)[0x7fe396d490d8]
/usr/lib/mysql/plugin/ha_tokudb.so(+0x93598)[0x7fe37a4d6598]
/usr/lib/mysql/plugin/ha_tokudb.so(+0xa6602)[0x7fe37a4e9602]
/usr/lib/mysql/plugin/ha_tokudb.so(+0x131e35)[0x7fe37a574e35]
/usr/lib/mysql/plugin/ha_tokudb.so(+0x5102c)[0x7fe37a49402c]
/usr/sbin/mysqld(_Z24ha_initialize_handlertonP13st_plugin_int+0x48)[0x599a78]
/usr/sbin/mysqld[0x6e4ba1]
/usr/sbin/mysqld(_Z11plugin_initPiPPci+0x941)[0x6eb1f1]
/usr/sbin/mysqld[0x59236d]
/usr/sbin/mysqld(_Z11mysqld_mainiPPc+0x42c)[0x5933fc]
/lib/x86_64-linux-gnu/libc.so.6(__libc_start_main+0xf5)[0x7fe396d30ec5]
/usr/sbin/mysqld[0x5856ed]
You may download the Percona Server operations manual by visiting
http://www.percona.com/software/percona-server/. You may find information
in the manual which will help you identify the cause of the crash.

[George Ormond Lorch III (gl-az)]

Yes, TokuDB can do numerous bad things when you have an absurdly small cache size. There is no logic that tests cache size against node size either, allowing you to have a cache size that is smaller than a single node size. We should add a more rational minimum cache size and additional logic that validates cache_size > node_size(block_size) *10 > basement_node_size(read_block_size)*4 or something similar.

[Roel Van de Paar (roel11)]

Another occurrence:

2016-01-26T12:54:37.444140Z 0 [Note] InnoDB: Buffer pool(s) load completed at 160126 23:54:37
/git/PS-5.7_dbg/storage/tokudb/PerconaFT/src/ydb.cc:979 env_open: Assertion `r == 0' failed (errno=2) (r=18446744073709451616)
: No such file or directory
Backtrace: (Note: toku_do_assert=0x0x7ff370d5f0f6)

+bt
#0 0x00007ff43ddb1741 in __pthread_kill (threadid=<optimized out>, signo=6) at ../nptl/sysdeps/unix/sysv/linux/pthread_kill.c:61
#1 0x0000000001811920 in my_write_core (sig=6) at /git/PS-5.7_dbg/mysys/stacktrace.c:247
#2 0x0000000000e609e3 in handle_fatal_signal (sig=6) at /git/PS-5.7_dbg/sql/signal_handler.cc:223
#3 <signal handler called>
#4 0x00007ff43c14a5f7 in __GI_raise (sig=sig@entry=6) at ../nptl/sysdeps/unix/sysv/linux/raise.c:56
#5 0x00007ff43c14bce8 in __GI_abort () at abort.c:90
#6 0x00007ff370d5ee6f in toku_do_backtrace_abort () at /git/PS-5.7_dbg/storage/tokudb/PerconaFT/portability/toku_assert.cc:147
#7 0x00007ff370d5f028 in toku_do_assert_zero_fail (expr=18446744073709451616, expr_as_string=0x7ff370d67c24 "r", function=0x7ff370d69b96 <env_open(__toku_db_env*, char const*, unsigned int, int)::__FUNCTION__> "env_open", file=0x7ff370d67fd0 "/git/PS-5.7_dbg/storage/tokudb/PerconaFT/src/ydb.cc", line=979, caller_errno=2) at /git/PS-5.7_dbg/storage/tokudb/PerconaFT/portability/toku_assert.cc:177
#8 0x00007ff370c60e65 in env_open (env=0x7ff38a31b200, home=0x2bfa300 <mysql_real_data_home> "/sda/PS-19.01.16-mysql-5.7.10-1rc1-linux-x86_64-debug/data/", flags=11403457, mode=438) at /git/PS-5.7_dbg/storage/tokudb/PerconaFT/src/ydb.cc:979
#9 0x00007ff370c4e4ea in tokudb_init_func (p=0x7ff4377fa120) at /git/PS-5.7_dbg/storage/tokudb/hatoku_hton.cc:553
#10 0x0000000000eeb963 in ha_initialize_handlerton (plugin=0x7ff38a361528) at /git/PS-5.7_dbg/sql/handler.cc:870
#11 0x0000000001534d2e in plugin_initialize (plugin=0x7ff38a361528) at /git/PS-5.7_dbg/sql/sql_plugin.cc:1178
#12 0x00000000015358ea in plugin_init (argc=0x2bff4b0 <remaining_argc>, argv=0x7ff43a022300, flags=0) at /git/PS-5.7_dbg/sql/sql_plugin.cc:1490
#13 0x0000000000e4fa53 in init_server_components () at /git/PS-5.7_dbg/sql/mysqld.cc:4221
#14 0x0000000000e51012 in mysqld_main (argc=12, argv=0x7ff43a022300) at /git/PS-5.7_dbg/sql/mysqld.cc:4897
#15 0x0000000000e48600 in main (argc=12, argv=0x7fff6fb656a8) at /git/PS-5.7_dbg/sql/main.cc:25

[Roel Van de Paar (roel11)]

That was with --no-defaults --tokudb-cache-size=12

[George Ormond Lorch III (gl-az)]

see https://tokutek.atlassian.net/browse/DB-933

[Shahriyar Rzayev (rzayev-sehriyar)]

Percona now uses JIRA for bug reports so this bug report is migrated to: https://jira.percona.com/browse/PS-938