Bug #13967 5.0.13rc
Submitted: 12 Oct 2005 16:40 Modified: 18 Nov 2005 13:15
Reporter: Chris Horn Email Updates:
Status: No Feedback Impact on me:
None 
Category:MySQL Server Severity:S2 (Serious)
Version:5.0.13-rc OS:Linux (Linux)
Assigned to: CPU Architecture:Any

[12 Oct 2005 16:40] Chris Horn
Description:
The attached SQL will fail on the x86_64 build of 5.0.13rc, but not on the x86 build.

$mysql weblog < just_cookies.sql
ERROR 126 (HY000) at line 115: Incorrect key file for table  './weblog/log_cookie_fragment.MYI'; try to repair it

How to repeat:
The attached file is atomically repeatable on my server, which is CentOS 4 + mysql-standard-5.0.13-rc-linux-x86_64-glibc23.
[12 Oct 2005 16:41] Chris Horn
Inserts -> Table corrupt

Attachment: just_cookies.sql (text/x-sql), 12.65 KiB.

[15 Oct 2005 12:43] Valeriy Kravchuk
Thank you for a problem report. Sorry, but I was not able to repeat the crash you are writing about on the following x86_64 machine:

-bash-2.05b$ uname -a
Linux nocona.mysql.com 2.4.21-32.0.1.ELsmp #1 SMP Tue May 17 17:46:36 EDT 2005 x86_64 x86_64 x86_64 GNU/Linux
-bash-2.05b$ cat /etc/issue
Red Hat Enterprise Linux AS release 3 (Taroon Update 2)
Kernel \r on an \m

with 5.0.14-rc I built there myself (I note it is NOT 5.0.13-rc, but I have to verify with a current version in any case). Moreover, I see no real reasons for your test script to fail in any case. 

So, please, provide the additional information: your my.cnf content, any unusual messages from the error log and any ideas on why this problem can happen to you and how to repeat it.
[17 Oct 2005 16:37] Chris Horn
After rather a lot of painful testing, this config item appears to be the culprit:

key_cache_block_size = 10240

This has been reproduced on multiple machines with various mysql.com distributed binaries.
[18 Oct 2005 13:15] Valeriy Kravchuk
Sorry, still can't repeat, neither on 32-bit with 5.0.16-BK (that is what you described too), nor on nocona. I've just recreated the table and inserted all the rows from your script:

mysql> select count(*) from log_cookie_fragment;
+----------+
| count(*) |
+----------+
|      106 |
+----------+
1 row in set (0.00 sec)

mysql> select version();
+----------------------------+
| version()                  |
+----------------------------+
| 5.0.14-rc-nightly-20050919 |
+----------------------------+
1 row in set (0.00 sec)

mysql> show variables like 'key_cache%';
+--------------------------+-------+
| Variable_name            | Value |
+--------------------------+-------+
| key_cache_age_threshold  | 300   |
| key_cache_block_size     | 10240 |
| key_cache_division_limit | 100   |
+--------------------------+-------+
3 rows in set (0.00 sec)

mysql> exit
Bye
-bash-2.05b$ uname -a
Linux nocona.mysql.com 2.4.21-32.0.1.ELsmp #1 SMP Tue May 17 17:46:36 EDT 2005 x86_64 x86_64 x86_64 GNU/Linux

with your key_cache_block_size...

So, please, give the information I asked about (your my.cnf content, all of it, any unusual messages from the error logs), as well as all the versions that demonstrates the described behaviour.
[18 Oct 2005 13:59] Valeriy Kravchuk
The same result (can't repeat the behaviour you described) with standard, published 5.0.13-rc, built from the sources on this nocona x86_64 platform. Just want you to know...
[19 Nov 2005 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".