Bug #38203 5.1 inserts spend a lot of time in the "closing table" thread state
Submitted: 17 Jul 2008 16:03 Modified: 29 Aug 2008 23:55
Reporter: Eric Jensen Email Updates:
Status: No Feedback Impact on me:
None 
Category:MySQL Server Severity:S5 (Performance)
Version:5.1.25 OS:Linux (x86_64)
Assigned to: Assigned Account CPU Architecture:Any
Tags: closing, table

[17 Jul 2008 16:03] Eric Jensen
Description:
I'm seeing 5x slower insert performance for my application on 5.1.25 vs. 5.0.51. One big difference I can see is that instead of the "update" state that I typically catch inserts at in 5.0, I only occasionally catch 5.1 insert threads in that state, but rather most often catch them in "closing table". I'm not sure if that's just a bad description of what's happening or not, but I can see in the status that I have far fewer tables open than my max cache, table opens counter is not growing, etc.

I realize this is not a well-encapsulated bug report.  I originally inquired about it on http://forums.mysql.com/read.php?24,216952,216952 but received no response.  It is preventing me from upgrading to 5.1

How to repeat:
The bottleneck in my application is bulk inserts (insert into ... values ... with 1500 or so values) into a one-to-many relation, i.e. a table that is only two integers:  a_id, b_id and has a single primary key on (a_id, b_id).  This bottleneck didn't change from 5.0 to 5.1, but the overall speed did.  It also appeared to be consuming more CPU than 5.0.
[17 Jul 2008 16:32] Eric Jensen
improved ticket metadata
[24 Jul 2008 16:27] MySQL Verification Team
Thank you for the bug report. Could you provide a test case (create table statements and data dump file)?. Thanks in advance.
[24 Jul 2008 17:38] Eric Jensen
See bug http://bugs.mysql.com/?id=38205 for my environment.  Thanks
[29 Jul 2008 23:55] MySQL Verification Team
Thank you for the feedback. Sorry but I can't repeat the behavior reported, the only way I can repeat this is comparing a 5.0 release server and 5.1 debug server which is expected, when doing against a 5.1 release server the performance is similar.
[30 Aug 2008 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".