Bug #49183 after unclean shutdown, mysqld halts with SIGSEGV; recovery impossible
Submitted: 29 Nov 2009 4:25 Modified: 19 Jul 2010 9:00
Reporter: joe barillari Email Updates:
Status: Can't repeat Impact on me:
Category:MySQL Server: InnoDB storage engine Severity:S3 (Non-critical)
Version:5.1.41 OS:Linux (Debian)
Assigned to: CPU Architecture:Any
Tags: innodb, segv, SIGSEGV

[29 Nov 2009 4:25] joe barillari
After a nasty crash (the laptop's battery died), MySQL refuses to start. InnoDB reports "Database page corruption on disk or a failed file read of page 1472705." It prints the message about forcing recovery (see the attached error log for full details), but then segfaults ("mysqld got signal 11"). It is not possible to use the innodb_force_recovery variable because the database never reaches the point where it will accept connections.

I'm not concerned about recovering the data (I have a full backup), but I wanted to report the SEGV.

Attachments: full error log from mysqld, stack trace with locals from GDB, my.cnf.

How to repeat:
Crash InnoDB nastily during a series of transactions.
[29 Nov 2009 4:25] joe barillari
My my.cnf

Attachment: my.cnf (application/octet-stream, text), 4.08 KiB.

[29 Nov 2009 4:26] joe barillari
Stderr from mysqld

Attachment: mysqld-stderr.txt (text/plain), 3.94 KiB.

[29 Nov 2009 4:28] joe barillari
GDB trace

Attachment: gdb-trace.txt (text/plain), 11.83 KiB.

[29 Nov 2009 4:37] joe barillari
Clarification: the crash reports came from mysql 5.1.41 as downloaded from mysql.com as mysql-5.1.41.tar.gz. I compiled it with gcc 4.3.4 using the configuration:

CC=gcc CFLAGS="-O2 -g" CXX=gcc CXXFLAGS="-O2 -g -felide-constructors     -fno-exceptions -fno-rtti" ./configure --prefix=/usr/local/mysql     --with-debug=full --with-extra-charsets=complex  --with-plugins=innobase && make

I actually experienced the crash using the Debian mysql package. I believe its version was 5.1.41-2, but I upgraded shortly after the crash, hoping to fix the problem, and I am not sure if the version at the time of the crash might have been slightly older.
[30 Dec 2009 23:52] Sveta Smirnova
Thank you for the report.

Could you please send us ib* files compressed?
[31 Dec 2009 14:04] joe barillari
Thanks for the follow-up! I would like to help, but after I hadn't heard anything for a while, I had to discard the ib* files to make room for a fresh copy of the database. My apologies.
[19 Jul 2010 9:00] Sveta Smirnova
Thank you for the feedback.

Status set to "Can't repeat", because I was not able to repeat the problem with generic test. If this crash happens next time please upload ib* files with configuration file to our ftp server, then add comment and reopen the report.