2 Quad Intel Xeon 2,2 Ghz - 12 GB memory OS : CentOS 5.8

Hi,

We would need more information on this, especially if, as you say, it only happens on some servers.
What is different about these servers comparing to those that are not affected?
Can you upload your schema and data dump? Can you enable general log on one of such servers?
If you can, then please:
when you have to restart the server again,

• create a data dump;
• enable general_log;
• run server until the memory leak becomes obvious;
• upload the previously created dump and the general log to our FTP server.

Thanks!

Hi,
I migrated many servers from MySQL 5.1 to MariaDB 5.5.30 & 5.5.31. The problem is only visible on servers with MariaDB 5.5.31
You can see on pictures the increasing memory consumption since 05/28/2013.

Hi,

I see the pictures and I believe you, but we still need more information to find the reason of the problem. If you can't provide the data dumps and the general logs, maybe at least some details about your schema and workflow? What engines you use, how big is the data, what kind of queries are typical, what is your configuration?
Could you execute
show variables;
show global status;
select engine,sum(data_length)/1024/1024 as DATA_MB,sum(INDEX_LENGTH)/1024/1024 as INDEX_MB from information_schema.tables group by engine;

and paste or attach the output?

Thanks.

• prdmutmys002.zip is the zip file of general log (prdmutmys002.log)
• engine_memory_uage.txt is the result of the request you provide me.

Hi,

I can't provide you a data dump but you'll find in as an attachment some answers to your questions.
Don't be surprise by the uptime in "show_global_status.txt" : we restarted the server this morning (too less memory available).
The file attach prdmutmys002.zip contains general log during 15 minutes. The queries reflect the typical activity of this server.

Thanks

Maybe this one? memory leak in xtradb + query cache:
https://bugs.launchpad.net/percona-server/+bug/1170103

Thanks, Patryk.

Hi,

could you please disable the query cache and see if it helps?

Thanks.

Hi,

I just disable query_cache (without restart, just by 'set global query_cache_size = 0') on one of our server and i observe his behavior during the day.

Thank you to both

It's better to set query_cache_type = 0 too.

Hi,

Consumption seems to stabilize since the deactivation of query_cache for 7 hours.
I'll tell you on Monday if the situation is stabilized.

Thanks

The server was migrated MariaDB 5.5 on 28/05, restarted Thursday 04/07.
Disabling the query_cache was dynamically applied Friday 05/07

Hi,

I confirm that the leak is stabilized on the server after 3 days (screenshot-1.jpg)
I apply the solution on other servers with the same problem.

Thank you

how is turning off the query cache a solution here? Sounds like removing an totally relevant part of the performance component of the server...

It's not a solution, it's a workaround till the next release (5.5.32) where the bug is supposed to be fixed.

Looks like you guys just release the version 5.5.32 and this fix is not on it. Do you have any certain release date for this bugifx? Also would you help guys if we provide us some more info on this matter ?

thanks

The fix should be in 5.5.32 release. The bug was fixed in XtraDB 5.5.32 (at least it is marked as such in https://bugs.launchpad.net/percona-server/+bug/1170103), and MariaDB 5.5.32 includes XtraDB 5.5.32.

We didn't have a reproducible test case to verify it, so please re-open the issue (or comment on it to get it re-opened) if you still have the problem with MariaDB 5.5.32,

excellent! perfect. Thanks for your time