Re: web app gets max processes about every three months

From: T. Schmidt <t.schmidt_at_rmv.spm.semantica.nl>
Date: Thu, 8 Aug 2002 14:35:20 +0200
Message-ID: <aitob3$9pc$1_at_reader09.wxs.nl>


"Joel Rees" <jreesmf_at_mac.com> wrote in message news:f0d5086.0208080142.1d574c54_at_posting.google.com...

> > # MaxRequestsPerChild: the number of requests each child process is
> > # allowed to process before the child dies. The child will exit so
> > # as to avoid problems after prolonged use when Apache (and maybe the
> > # libraries it uses) leak memory or other resources.
>
> Yeah, this is usually for memory leaks, but it might help with process
> allocation "leaks" as well.
>
> > You could try to set the "MaxRequestsPerChild" parameter (httpd.conf) to
 eg.
> > 1000.
> >
> > In conjunction with:
> > KeepAlive
> > MaxKeepAliveRequest
> > KeepAliveTimeout
> > MinSpareServers
> > MaxSpareServers
> > MaxClients
> >
> > Intensive tuning with these parameters solved our similar (??)
> > Apache<->Oracle- processes problems.
>
> Thanks for the suggestions. Question -- was your OS MSWindows? Also,
> what versions of the OS, Apache, and Oracle were you using?

Linux, Oracle 9iAS
Digital UNIX V4.0B [Rev. 564], Oracle RDBMS 7

At some moment db connections weren't being reused, and incomming httpd requestst couldn't be taken care of anymore. By periodically stopping the Apache child process (atfter eg. 100 MaxrequestperChild), thus ending the db connection the whole flow of serving a httpd request -> opening or reusing a db connection kept on going. Trade off: a db connection has to be (re)build after some time (= processor consuming). But we never faced the situation of "maximum processes allocated".

Good Luck,

-TS

  • to reply remove 'rmv.spm.' from e-mail address
Received on Thu Aug 08 2002 - 14:35:20 CEST

Original text of this message