Bug #48192 | MySQL Crash (InnoDB: We intentionally generate a memory trap.) | ||
---|---|---|---|
Submitted: | 20 Oct 2009 18:05 | Modified: | 17 Jul 2010 11:47 |
Reporter: | Tony Weber | Email Updates: | |
Status: | No Feedback | Impact on me: | |
Category: | MySQL Server: InnoDB storage engine | Severity: | S1 (Critical) |
Version: | 5.0.86-enterprise-nt-log | OS: | Windows (XP Pro and 2003 Server) |
Assigned to: | CPU Architecture: | Any |
[20 Oct 2009 18:05]
Tony Weber
[20 Oct 2009 19:40]
Tony Weber
I uploaded the error log and general query log to ftp://ftp.mysql.com/pub/mysql/upload/bug48192-logs.zip I uploaded the initial databases to ftp://ftp.mysql.com/pub/mysql/upload/bug48192-initial-databases.zip The test case was clean: restore the initial databases (`vfs` and `newsdd`), run our Virtual Factory application for about 31 hours. All queries after the restore are shown.
[21 Oct 2009 3:27]
Valeriy Kravchuk
Thank you for the problem report. Please, send your my.ini file content, the resuslt of SHOW GLOBAL STATUS and describe your hardware. Server was waiting too long for a buffer pool flush. So it was intentionally crashed. I'd say this is more a misconfiguration for your load than a bug. I'd also recommend to upgrade to 5.0.86 at least.
[21 Oct 2009 3:30]
Valeriy Kravchuk
Sorry, ignore my last suggestion. You are on 5.0.86 already it seems. Had you tried 5.1.x already?
[21 Oct 2009 5:43]
Tony Weber
my.ini and SHOW GLOBAL STATUS
Attachment: showglobalstatus.zip (application/x-zip-compressed, text), 6.28 KiB.
[21 Oct 2009 5:46]
Tony Weber
Machine name: VFS-SERVER Operating System: Windows XP Professional (5.1, Build 2600) Service Pack 3 (2600.xpsp_sp3_gdr.090206-1234) Language: Japanese (Regional Setting: Japanese) System Manufacturer: NEC System Model: Express5800/51Lb [N8000-503] BIOS: Default System BIOS Processor: Intel(R) Pentium(R) 4 CPU 2.40GHz Memory: 1016MB RAM Page File: 617MB used, 1828MB available Windows Dir: C:\WINDOWS
[21 Oct 2009 14:24]
MySQL Verification Team
See bug: http://bugs.mysql.com/bug.php?id=25506.
[21 Oct 2009 18:00]
Valeriy Kravchuk
Can you, please, try to monitor IO writes using perfmon during high load periods before the crash? Had you checked that older bug report Miguel had found, for some workaround ideas/similarities?
[22 Nov 2009 0:00]
Bugs System
No feedback was provided for this bug for over a month, so it is being suspended automatically. If you are able to provide the information that was originally requested, please do so and change the status of the bug back to "Open".
[30 Nov 2009 19:49]
Tony Weber
Others have reported this issue with InnoDB when innodb_file_per_table is on under Windows. For example http://bugs.mysql.com/bug.php?id=25506. We also have BLOB data and at some point InnoDB starts heavy I/O thrashing, resulting in simple queries being blocked for over 10 minutes. This results in InnoDB crashing itself. Some others reported the issue happening when anti-virus or file indexing services are turned on (and therefore trying to read the tablespace file), but I verified that this is not our situation. The situation does not occur when innodb_file_per_table is off. This is our current work-around.
[17 Jun 2010 11:47]
Valeriy Kravchuk
If you still have a chance, please, check if this still happens with a newer version, 5.0.91 or, even better, 5.1.48.
[15 Jul 2010 9:37]
Susanne Ebrecht
Bug #54720 could be a duplicate of this bug here.
[17 Jul 2010 23:00]
Bugs System
No feedback was provided for this bug for over a month, so it is being suspended automatically. If you are able to provide the information that was originally requested, please do so and change the status of the bug back to "Open".