Bug #38474 | Maria crash in _ma_setup_live_state (pushbuild2) | ||
---|---|---|---|
Submitted: | 30 Jul 2008 21:09 | Modified: | 10 Mar 2009 14:33 |
Reporter: | Philip Stoev | Email Updates: | |
Status: | Closed | Impact on me: | |
Category: | MySQL Server: Maria storage engine | Severity: | S1 (Critical) |
Version: | 6.0-bzr | OS: | Any |
Assigned to: | Michael Widenius | CPU Architecture: | Any |
[30 Jul 2008 21:09]
Philip Stoev
[30 Jul 2008 22:04]
Philip Stoev
Test case for bug 38368
Attachment: bug38368.zip (application/x-zip-compressed, text), 4.21 KiB.
[30 Jul 2008 22:06]
Philip Stoev
Please find attached a test case for this bug. Please place the .txt files in mysql-test and the .test files in mysql-test/t and then run: $ perl ./mysql-test-run.pl \ --stress \ --stress-init-file=bug38368_init.txt \ --stress-test-file=bug38368_run.txt \ --stress-threads=10 \ --skip-ndb \ --mysqld=--log-output=file \ --stress-test-duration=60000 A crash will happen shortly after takeoff. The _init file may include tables and views that are not used in the actual test. Please let me know if an even more simplifed test case is required.
[31 Jul 2008 21:21]
Philip Stoev
Note: partitions are used in the test case.
[24 Aug 2008 13:22]
Michael Widenius
I tried the test on 2 different 64 bit linux systems, with and without debug, but couldn't repeat the problem withing 30 minutes/2 hours on either system. It's possible that my earlier fix for Bug#30816 fixed this case too.
[15 Sep 2008 9:26]
Philip Stoev
This bug is still present in 6.0.7
[28 Sep 2008 14:49]
Philip Stoev
Not present in 6.0-maria
[28 Sep 2008 16:22]
Philip Stoev
This just happened again on 6.0-maria with a different test. I will try to provide a simple test case.
[24 Nov 2008 19:58]
Guilhem Bichot
More precise post. It recently happened in 5.1-maria: guilhem@mysql.co... 2008-11-20 15:52:38 machine sol10 sparc32 sysqa test maria_stress: http://clustra.norway.sun.com/~bteam/pb2/web.py?action=archive_download&archive_id=138119&...; pretty=please serg@mysql.com-2... 2008-11-10 20:18:15 machine sol10 sparc32 sysqa test maria_stress: http://clustra.norway.sun.com/~bteam/pb2/web.py?action=archive_download&archive_id=122892&...; pretty=please The links above contain the exact log of what was done and stack traces.
[25 Nov 2008 10:08]
Guilhem Bichot
Also seen on Nov 22 when running tests on Linux 64 bit with default engine = Maria: main.events_restart_phase1 [ pass ] main.events_restart_phase2 [ pass ] main.events_restart_phase3 [ fail ] mysqltest: In included file "./include/wait_condition.inc": At line 53: Error running query 'select count(distinct name)=3 from execution_log': 2013 Lost connection to MySQL server during query (same stack trace).
[2 Dec 2008 15:23]
Sergei Golubchik
another test case: ./mtr --mysqld=-#f,new_trid,trnman_new_trn,trnman_end_trn,trnman_collect_transactions:t:d:O,dbug.log --mysqld=--default-storage-engine=maria mysqlslap this crashes almost always in_ma_setup_live_state(): while (trn->trid < history->trid && history->trid != ~(TrID)0) history= history->next; when the end of a list (history==0) is reached
[2 Dec 2008 15:31]
Guilhem Bichot
I can indeed get this crash in 5.1-maria: Program received signal SIGSEGV, Segmentation fault. 0x08610ce1 in _ma_setup_live_state (info=0xb696b4e8) at ma_state.c:94 94 while (trn->trid < history->trid && history->trid != ~(TrID)0) Current language: auto; currently c (gdb) p history $2 = (MARIA_STATE_HISTORY *) 0x0 by using Serg's exact mtr command-line in his last post, to which I add --mem.
[10 Dec 2008 16:49]
Michael Widenius
Fixed a bug that caused a crash at exactly this line in the MySQL-Maria tree. Will be merged to MySQL-6.0 shortly