[pgpool-general: 3960] Re: accept() scalability issues
michael.renner at amd.co.at
Thu Aug 13 22:18:43 JST 2015
> On 10 Aug 2015, at 11:25, Tatsuo Ishii <ishii at postgresql.org> wrote:
>> So if there’s nothing horribly wrong/out of the ordinary with our setup, I’d pretty much say that with a high number of children fencing around select/accept is dearly needed.
> Ok, here is the experimental patch to serialize select(2) for listen
> fds by using semaphore. Could you please try it out? The patch is
> against 3.3-stable.
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.
We’ll test it in production mid next week, will let you know how things go.
More information about the pgpool-general