Strange behavior with MySQL (solved) [message #27654] |
Thu, 21 February 2008 22:30 |
bfrackie
Messages: 2 Registered: February 2008
|
Junior Member |
|
|
Hi,
i can reproduce a strange behavior on different container (running on several hardware nodes) running only MySQL Servers (see detail informations below).
A query with more than eleven LEFT JOINS did not respond, the mysql shell just hangs and the mysqld process reaches 100% CPU. After opening a new shell and submitting the same query in parallel, the CPU grows to 200% and so on. The same query on my notebook returns a result without any delay.
I am not able to isolate the problem, because a cat on user_beancounters shows no failcnt. I also set open-files-limit to 1024 (inside my.cnf) to avoid file descriptor problems. Also, no other informations are given in the mysql.log or syslog.
What else can i do to track the problem? Any hints? If you need more informations, just give me note.
Thanks for your help,
Bart
Hardware Node:
- Debian 4.0 with kernel from download.openvz.org
- Linux tropical 2.6.18-openvz-k7 #1 SMP Wed Apr 11 07:50:57 CEST 2007 i686 GNU/Linux
- vzctl version 3.0.18-1dso1
Container:
- Ubuntu 7.10
- MySQL 5.0.45
[Updated on: Mon, 25 February 2008 10:31] Report message to a moderator
|
|
|
|
Re: Strange behavior with MySQL (solved) [message #27732 is a reply to message #27659] |
Mon, 25 February 2008 10:30 |
bfrackie
Messages: 2 Registered: February 2008
|
Junior Member |
|
|
Hi Kir,
after hours of research, i have found that this is a bug in MySQL (a bug ticket on MySQL called this a feature). You have to to set
optimizer-search-depth = 0 (or another deepth, default is 62)
if you use more than 11 tables. Curios, because my notebook has the same value, but it works here. Anyway, thanks for support.
Bart
|
|
|