Bug #20998 Sql-server restart repeatedly

• Submitted: 12 Jul 2006 11:56
• Modified: 7 Sep 2006 14:24
• Reporter: Marco Bonfatti
• Status: No Feedback
• Category: MySQL Cluster: Cluster (NDB) storage engine
• Severity: S2 (Serious)
• Version: 5.1.11
• OS: Linux (Linux - Gentoo 2006.0)
• CPU Architecture: Any
• Tags: API, cluster, restart, sql server

[12 Jul 2006 11:56] Marco Bonfatti

Description:
I've a cluster with 4 data nodes (wl15,wl16,wl17,wl19) and 2 API nodes (wl14 and wl18); I use Tomcat5 and the mysql-connector-3.1.13.
I've 2 java-client that start to make query (each client do 1-5 query per second) to the Tomcat5 server; about 1 minute, the sql server restart (then I lost connection and query-response); if I wait that other connections starts make query to the sql-servers, the servers restart again.

Meantime I've a lot of errors because a query don't get me rows (empty set):
"SELECT * FROM AUTHOR, ITEM WHERE soundex(AUTHOR.A_LNAME) LIKE RPAD(soundex(?), 5, '%') AND ITEM.I_A_ID = AUTHOR.A_ID ORDER BY ITEM.I_TITLE LIMIT 50;"

How to repeat:
see above

[12 Jul 2006 11:59] Marco Bonfatti

Error log

Attachment: err.log (text/x-log), 1.94 KiB.

[12 Jul 2006 12:41] Jonathan Miller

Hi,

Usually if you have lost connection and the MySQLD has restarted, it is due to a crash. Can you please check your MySQLD error logs for a stack trace, or a core file in your data directory?

Thanks

[12 Jul 2006 13:52] Marco Bonfatti

This is the log file

Attachment: wl18.err (application/octet-stream, text), 2.18 KiB.

[12 Jul 2006 13:53] Marco Bonfatti

I've this log file, no other trace file was created for the crash.

[12 Jul 2006 17:10] Valeriy Kravchuk

You've got:

060712 15:49:15 [ERROR] Got error 121 when reading table './tpcw/ITEM'
060712 15:49:18 [ERROR] Got error 121 when reading table './tpcw/ADDRESS'

immediately before the crash (on COMMIT). Please, send the SHOW CREATE TABLE and SHOW TABLE STATUS results for both this tables.

It can be a hardware-related problem that lead to tables corruption.

[12 Jul 2006 18:14] Jonathan Miller

Hi,

Are you running TPC-W against the MySQL Cluster? If so please email me @ [e-mail].

Thanks,

[13 Jul 2006 7:47] Marco Bonfatti

SHOW CREATE TABLE *   and  SHOW TABLE STATUS

Attachment: tables.txt (text/plain), 4.71 KiB.

[13 Jul 2006 7:49] Marco Bonfatti

Yes, my intent was to use the suite TPC-W with MySql-Cluster.

[20 Jul 2006 3:12] Jason Downing

I had the same problem, and I got around it by re-installing Linux (Debian Sarge) on each machine in the cluster. I think there was something to do with the 5.0.20 version that had been there before I upgraded to 5.1.11

[24 Jul 2006 7:13] Marco Bonfatti

I've unpacked MySql-5.1.11 in a separate folder... I can't figure out how the old version can "disturb" the new version.
Now I cannot format and re-install the OS... I've to find out a different way.

[7 Aug 2006 14:24] Hartmut Holzgraefe

as the error log didn't contain a resolveable stack trace we'd really need a core dump for further analysis ...

[7 Sep 2006 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".