[pgpool-general: 817] Re: pgpool causes high load

Benjamin Fras benjamin.fras at petafuel.de
Fri Jul 27 16:56:26 JST 2012


thank you for your reply. We tried out the latest snapshot of 3.1-stable tree like you suggested. Unfortunately this didn't solve the problem and we still have those weird load peaks. Any other ideas what might be causing this?

petaFuel GmbH
Benjamin Fras
IT Systemadministration

Tel.: +49 (0) 8161 4060 - 222

Geschäftsführer: Ludwig Adam, Dr. Peter Schönweitz
Amtsgericht München, HRB 133773

-----Ursprüngliche Nachricht-----
Von: Tatsuo Ishii [mailto:ishii at postgresql.org] 
Gesendet: Montag, 16. Juli 2012 12:12
An: Benjamin Fras
Cc: pgpool-general at pgpool.net
Betreff: Re: [pgpool-general: 739] pgpool causes high load

My guess is this:

The part which handles incoming events has been changed when 3.1 was released. Since then we have had several reports including near infinite loop which consumes CPU a lot. Though the problem reported in above is not similar to yours, I suspect the previous code could cause your problem. Can you try patch (can be downloaded from "patch" link or the latest snapshot of 3.1-stable tree?  (can be downloaded from
Tatsuo Ishii
SRA OSS, Inc. Japan
English: http://www.sraoss.co.jp/index_en.php
Japanese: http://www.sraoss.co.jp

> Dear list members,
> since we have updated our pgpools to the latest stable version (3.1.3), we have regular peaks of high load on our machines. Every 5 to 10 minutes the server load (average 1 min load) raises up to 50, but then recovers within the next minutes to normal values again. CPU usage and IO wait are unaffected and show normal values during the whole time, so we cannot identify the processes that cause the load. Do you have any ideas what might cause this behavior? We didn't have that problems with the previous version of pgpool (3.0.4).
> We suspected the root of the problem in child_life_time, which was set to 300. Thus we increased that value up to 7200 (2 hours), however it didn't improve the situation at all. Another measure was adjusting child_max_connections from 1500 to 1000 which also didn't solve the issue. Max_pool is set to 1.
> Our pgpools are running in master/slave mode and postgres streaming replication (2 postgres 9.1 backends).
> I want to add, that this problem is not that intense in another 
> environment. However this environment does not have as many 
> simultaneous connections (child_max_connections is set to 500),
> Apart from that strange load problems everything works fine.
> Regards,
> petaFuel GmbH
> Benjamin Fras
> IT Systemadministration
> Münchner Strasse 4
> 85354 Freising
> Tel.: +49 (0) 8161 4060 - 222
> Fax: +49 (0) 8161 4060 - 401            eMail: Benjamin.Fras at petafuel.de<mailto:Vorname.Nachname at petafuel.de>
> Geschäftsführer: Ludwig Adam, Dr. Peter Schönweitz Amtsgericht 
> München, HRB 133773
> www.petaFuel.de<http://www.petafuel.de/>

More information about the pgpool-general mailing list