Bug #17750 | MySQL service fails on systems with AMD processor(s) | ||
---|---|---|---|
Submitted: | 27 Feb 2006 18:59 | Modified: | 15 Mar 2006 9:47 |
Reporter: | Eugene Drobitko | Email Updates: | |
Status: | Duplicate | Impact on me: | |
Category: | MySQL Server: InnoDB storage engine | Severity: | S1 (Critical) |
Version: | 5.0.18 | OS: | Windows (Windows 2003) |
Assigned to: | CPU Architecture: | Any |
[27 Feb 2006 18:59]
Eugene Drobitko
[27 Feb 2006 19:36]
Jorge del Conde
Hi! Can you please give me instructions on how to reproduce this crash ? I have been running mysql under an AMD64 environment for quite some time, and have never experienced the problem you're describing. I use mysqld.exe and mysqld-max-nt.exe
[2 Mar 2006 5:54]
Jorge del Conde
Hi Eugene, Thanks for providing us the details to reproduce this bug. I experienced the behaviour you reported in my AMD64 machine, hence I was able to reproduce this bug.
[2 Mar 2006 5:55]
Jorge del Conde
I tested this under WinXP-64 and WinXP/SP2
[10 Mar 2006 9:42]
MySQL Verification Team
Are the tables InnoDB ? I might suspect assertion from btr_pcur_restore_position() is it's the case. what's in the mysql error log ??
[14 Mar 2006 5:26]
MySQL Verification Team
060313 18:32:00InnoDB: Assertion failure in thread 360 in file .\btr\btr0pcur.c
Attachment: 5.0.18-max-nt-error-log.txt (application/octet-stream, text), 23.49 KiB.
[14 Mar 2006 5:37]
MySQL Verification Team
Line 216 of btr0pcur.c is the below assertion. if (UNIV_UNLIKELY(cursor->old_stored != BTR_PCUR_OLD_STORED) || UNIV_UNLIKELY(cursor->pos_state != BTR_PCUR_WAS_POSITIONED && cursor->pos_state != BTR_PCUR_IS_POSITIONED)) { ut_print_buf(stderr, (const byte*)cursor, sizeof(btr_pcur_t)); if (cursor->trx_if_known) { trx_print(stderr, cursor->trx_if_known, 0); } ut_error; }
[15 Mar 2006 9:47]
Heikki Tuuri
Since you are using innodb_locks_unsafe_for_binlog, this is probably a duplicate of: http://bugs.mysql.com/bug.php?id=15650 Please test with 5.0.19.
[15 Mar 2006 11:08]
MySQL Verification Team
Heikki, 5.0.19 still crashes but I suspect it's a completely different crash, for which I will investigate and open another bug report if needed. 5.0.18 didn't crash when innodb_locks_unsafe_for_binlog=0.