This is the mail archive of the
cygwin@cygwin.com
mailing list for the Cygwin project.
Re: 1.3.16-1: Apache 1.3.24 not rotating/lockup problem
- From: "Elfyn McBratney" <elfyn-cygwin at exposure dot org dot uk>
- To: "cygwin" <cygwin at cygwin dot com>
- Date: Sun, 23 Feb 2003 08:46:36 -0000
- Subject: Re: 1.3.16-1: Apache 1.3.24 not rotating/lockup problem
- References: <3DE48797.5030004@atrove.com> <3DE4F8EE.8E7CAE83@wapme-systems.de> <20021127185855.O1398@cygbert.vinschen.de> <3DE59878.A9D64756@wapme-systems.de> <20021128101359.A23323@cygbert.vinschen.de> <b31tt5$cn9$1@main.gmane.org> <3E549A11.2C8ABB25@wapme-systems.de> <03b401c2db0a$d14b8ae0$666d86d9@webdev>
> > Dylan Cuthbert wrote:
> > >
> > > Any more information on this problem? Anyone solved it?
> >
> > unfortunatly no. I guessed the latest cygwin 1.3.20-x, which included
> > some TCP connection related fixes would be the cure, but it didn't.
> >
> > > server-status makes it obvious what's being said in this thread - ie.
> the
> > > children forked by apache aren't allowing themselves to be re-used on
a
> > > different connection properly, hence they all eventually lock up (once
> > > KeepAlive times out) and your web site goes down.
> > >
> > > Corinna - simply set KeepAlive on, and the time out value to 10. Then
> > > access your index page, wait 10 seconds and then access it again...
> continue
> > > this a few times and you will end up locking all the children up.
(they
> > > work fine until the keepalive system tries to reset them to another
> > > connection)
> >
> > thanks for the analysis. I'll try to scratch that down on one of our
> > installations here.
>
> Stipe, All,
>
> I thought I'd give the latest cygwin snapshot (Febury 20, 1.3.21s) a go
and
> see if that fixes the problem, as Corinna made quite a few fixes in the
> fhandler_* code. I've been running the Apache service for just over one
hour
> and twenty minutes, and although I can only access it from my machine due
to
> firewall restrictions, it is working quite well. I haven't done anything
out
> of the ordinary to get it up and running, unless this is considered out of
> the ordinary:
>
> - Changed the User/Group directives in httpd.conf to system/system.
>
> And as Dylan Cuthbert suggested, I kept the KeepAlive directive/set to
'on'
> and changed the Timeout directive to '10'. I'm accessing four documents at
a
> refresh interval of five seconds and have not seen an error what so ever.
> I'm strace'ing just in case, too.
>
> Just thought you'd like to know ;-)
Just an update: Apache has been running for over three hours now :-) It's
looking very good that this bug has well and truly died! ::::::-)
Regards,
Elfyn McBratney
elfyn at exposure dot org dot uk
www.exposure.org.uk
--
Unsubscribe info: http://cygwin.com/ml/#unsubscribe-simple
Bug reporting: http://cygwin.com/bugs.html
Documentation: http://cygwin.com/docs.html
FAQ: http://cygwin.com/faq/