I tried executing the reproducible TC. Here are my observations:
* This bug is specific about use of --innodb-buffer-pool-populate which is not present in the bug-tcs.
* BugTC is trying to set "SET @@global.table_open_cache=TRUE;". Well table_open_cache is not a boolean but a numeric variable. Setting it to TRUE mean setting it to 1 that in other term means only 1 table can be open at any given time. Not sure if this is indented or miss by TC.
* TC crashes randomly when toku-db is involved. W/O toku-db involved I couldn't get the TC to crash. Infact last 2 crash where in toku-db code hitting the assert. See below for assert.
---------
Example of toku-db assert.
#2 0x00007fffdc2a221f in toku_do_backtrace_abort () at /opt/projects/codebase/5.6/storage/tokudb/PerconaFT/portability/toku_assert.cc:147
#3 0x00007fffdc2a2319 in toku_do_assert_fail (expr_as_string=0x7fffdc2cdb28 "txn->roll_info.num_rollentries == 0", function=0x7fffdc2ce070 <toku_txn_is_read_only(tokutxn*)::__FUNCTION__> "toku_txn_is_read_o
nly", file=0x7fffdc2cd880 "/opt/projects/codebase/5.6/storage/tokudb/PerconaFT/ft/txn/txn.cc", line=664, caller_errno=0) at /opt/projects/codebase/5.6/storage/tokudb/PerconaFT/portability/toku_assert.cc:168
#4 0x00007fffdc22c0ae in toku_txn_is_read_only (txn=0x7fff84026800) at /opt/projects/codebase/5.6/storage/tokudb/PerconaFT/ft/txn/txn.cc:664
#5 0x00007fffdc1b1d11 in locked_txn_commit_with_progress (txn=0x7fff840264d0, flags=0, poll=0x0, poll_extra=0x0) at /opt/projects/codebase/5.6/storage/tokudb/PerconaFT/src/ydb_txn.cc:267
#6 0x00007fffdc1b1e2b in locked_txn_commit (txn=0x7fff840264d0, flags=0) at /opt/projects/codebase/5.6/storage/tokudb/PerconaFT/src/ydb_txn.cc:317
#7 0x00007fffdc157398 in commit_txn (txn=0x7fff840264d0, flags=0) at /opt/projects/codebase/5.6/storage/tokudb/hatoku_defines.h:351
#8 0x00007fffdc15db42 in ha_tokudb::initialize_share (this=0x7fff84029280, name=0x7fff840138e8 "./test/t0", mode=2) at /opt/projects/codebase/5.6/storage/tokudb/ha_tokudb.cc:1664
#9 0x00007fffdc15e247 in ha_tokudb::open (this=0x7fff84029280, name=0x7fff840138e8 "./test/t0", mode=2, test_if_locked=2) at /opt/projects/codebase/5.6/storage/tokudb/ha_tokudb.cc:1765
#10 0x000000000064a9d2 in handler::ha_open (this=0x7fff84029280, table_arg=0x7fff84022c70, name=0x7fff840138e8 "./test/t0", mode=2, test_if_locked=2) at /opt/projects/codebase/5.6/sql/handler.cc:2743
#11 0x00000000008d1d65 in open_table_from_share (thd=0x20a9b40, share=0x7fff84013570, alias=0x7fff840056d0 "t0", db_stat=39, prgflag=44, ha_open_flags=0, outparam=0x7fff84022c70, is_create_table=false) at /o
pt/projects/codebase/5.6/sql/table.cc:2378
------------
Based on all these observations I would suggest you log a separate bug for toku-db using this TC and see if you can get the reproducible TC for existing bug with needed options.
I tried executing the reproducible TC. Here are my observations:
* This bug is specific about use of --innodb- buffer- pool-populate which is not present in the bug-tcs.
* BugTC is trying to set "SET @@global. table_open_ cache=TRUE; ". Well table_open_cache is not a boolean but a numeric variable. Setting it to TRUE mean setting it to 1 that in other term means only 1 table can be open at any given time. Not sure if this is indented or miss by TC.
* TC crashes randomly when toku-db is involved. W/O toku-db involved I couldn't get the TC to crash. Infact last 2 crash where in toku-db code hitting the assert. See below for assert.
---------
Example of toku-db assert.
#2 0x00007fffdc2a221f in toku_do_ backtrace_ abort () at /opt/projects/ codebase/ 5.6/storage/ tokudb/ PerconaFT/ portability/ toku_assert. cc:147 string= 0x7fffdc2cdb28 "txn->roll_ info.num_ rollentries == 0", function= 0x7fffdc2ce070 <toku_txn_ is_read_ only(tokutxn* )::__FUNCTION_ _> "toku_txn_is_read_o codebase/ 5.6/storage/ tokudb/ PerconaFT/ ft/txn/ txn.cc" , line=664, caller_errno=0) at /opt/projects/ codebase/ 5.6/storage/ tokudb/ PerconaFT/ portability/ toku_assert. cc:168 is_read_ only (txn=0x7fff8402 6800) at /opt/projects/ codebase/ 5.6/storage/ tokudb/ PerconaFT/ ft/txn/ txn.cc: 664 txn_commit_ with_progress (txn=0x7fff8402 64d0, flags=0, poll=0x0, poll_extra=0x0) at /opt/projects/ codebase/ 5.6/storage/ tokudb/ PerconaFT/ src/ydb_ txn.cc: 267 64d0, flags=0) at /opt/projects/ codebase/ 5.6/storage/ tokudb/ PerconaFT/ src/ydb_ txn.cc: 317 64d0, flags=0) at /opt/projects/ codebase/ 5.6/storage/ tokudb/ hatoku_ defines. h:351 :initialize_ share (this=0x7fff840 29280, name=0x7fff840138e8 "./test/t0", mode=2) at /opt/projects/ codebase/ 5.6/storage/ tokudb/ ha_tokudb. cc:1664 29280, name=0x7fff840138e8 "./test/t0", mode=2, test_if_locked=2) at /opt/projects/ codebase/ 5.6/storage/ tokudb/ ha_tokudb. cc:1765 29280, table_arg= 0x7fff84022c70, name=0x7fff840138e8 "./test/t0", mode=2, test_if_locked=2) at /opt/projects/ codebase/ 5.6/sql/ handler. cc:2743 from_share (thd=0x20a9b40, share=0x7fff840 13570, alias=0x7fff840 056d0 "t0", db_stat=39, prgflag=44, ha_open_flags=0, outparam= 0x7fff84022c70, is_create_ table=false) at /o codebase/ 5.6/sql/ table.cc: 2378
#3 0x00007fffdc2a2319 in toku_do_assert_fail (expr_as_
nly", file=0x7fffdc2cd880 "/opt/projects/
#4 0x00007fffdc22c0ae in toku_txn_
#5 0x00007fffdc1b1d11 in locked_
#6 0x00007fffdc1b1e2b in locked_txn_commit (txn=0x7fff8402
#7 0x00007fffdc157398 in commit_txn (txn=0x7fff8402
#8 0x00007fffdc15db42 in ha_tokudb:
#9 0x00007fffdc15e247 in ha_tokudb::open (this=0x7fff840
#10 0x000000000064a9d2 in handler::ha_open (this=0x7fff840
#11 0x00000000008d1d65 in open_table_
pt/projects/
------------
Based on all these observations I would suggest you log a separate bug for toku-db using this TC and see if you can get the reproducible TC for existing bug with needed options.