[pgpool-general: 3962] Re: accept() scalability issues

Michael Renner michael.renner at amd.co.at
Thu Aug 13 23:15:22 JST 2015


> On 13 Aug 2015, at 15:21, Pablo Sanchez <pablo at blueoakdb.com> wrote:
> 
> On 08/13/2015 09:18 AM, Michael Renner wrote:
>> 
>> Thank you very much, that worked fine!
>> 
>> In a synthetic test with 1000 children and ~30 connections per
>> second this lowered the amount of context switches from ~38.000/sec to
>> ~1.000/sec.
>> 
> 
> Hey Michael,
> 
> Do you have any /sar -u 3 XXX/ data showing CPU usage between /before/
> and /after/?

No, not more than the vmstat output in my initial mail: http://nopaste.narf.at/show/25531/ <http://nopaste.narf.at/show/25531/>. And those values are quite dependent on the environment you’re running, so I find them not very interesting to compare.

You can easily reproduce the behaviour by running pgpool with a few hundred to thousands of pgpool children and a rate of 50-100 new connections per second.

best,
Michael
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.sraoss.jp/pipermail/pgpool-general/attachments/20150813/d7d853d5/attachment.html>


More information about the pgpool-general mailing list