Return-Path: Delivered-To: apmail-httpd-dev-archive@www.apache.org Received: (qmail 6647 invoked from network); 12 Jan 2009 13:50:00 -0000 Received: from hermes.apache.org (HELO mail.apache.org) (140.211.11.2) by minotaur.apache.org with SMTP; 12 Jan 2009 13:50:00 -0000 Received: (qmail 52102 invoked by uid 500); 12 Jan 2009 13:49:57 -0000 Delivered-To: apmail-httpd-dev-archive@httpd.apache.org Received: (qmail 51697 invoked by uid 500); 12 Jan 2009 13:49:56 -0000 Mailing-List: contact dev-help@httpd.apache.org; run by ezmlm Precedence: bulk Reply-To: dev@httpd.apache.org list-help: list-unsubscribe: List-Post: List-Id: Delivered-To: mailing list dev@httpd.apache.org Received: (qmail 51688 invoked by uid 99); 12 Jan 2009 13:49:55 -0000 Received: from athena.apache.org (HELO athena.apache.org) (140.211.11.136) by apache.org (qpsmtpd/0.29) with ESMTP; Mon, 12 Jan 2009 05:49:55 -0800 X-ASF-Spam-Status: No, hits=2.2 required=10.0 tests=HTML_MESSAGE,SPF_PASS X-Spam-Check-By: apache.org Received-SPF: pass (athena.apache.org: domain of trawick@gmail.com designates 74.125.46.162 as permitted sender) Received: from [74.125.46.162] (HELO yw-out-1516.google.com) (74.125.46.162) by apache.org (qpsmtpd/0.29) with ESMTP; Mon, 12 Jan 2009 13:49:49 +0000 Received: by yw-out-1516.google.com with SMTP id 4so4920238ywi.39 for ; Mon, 12 Jan 2009 05:49:28 -0800 (PST) MIME-Version: 1.0 Received: by 10.90.94.15 with SMTP id r15mr63242agb.2.1231768168734; Mon, 12 Jan 2009 05:49:28 -0800 (PST) Message-ID: <00163630ee1b24a2dc0460496078@google.com> Date: Mon, 12 Jan 2009 13:49:28 +0000 Subject: Re: Re: Graceful restart not so graceful? From: trawick@gmail.com To: dev@httpd.apache.org Content-Type: multipart/alternative; boundary=00163630ee1b24a2c90460496011 X-Virus-Checked: Checked by ClamAV on apache.org --00163630ee1b24a2c90460496011 Content-Type: text/plain; charset=ISO-8859-1; format=flowed; delsp=yes Content-Transfer-Encoding: 7bit On Jan 11, 2009 11:34am, Stefan Fritsch wrote: > Hi, > > > > thanks for following up on this and sorry for the late response. > > > > On Wednesday 07 January 2009, Jeff Trawick wrote: > > > Initial testing of your idea for a timeout was promising. > > > > I couldn't reproduce any hangs under linux with the patch you commited > > to trunk. Thanks for all of your work on this issue! > In my patch I tried to avoid that an idle apache wakes periodically > > for no good reasons. But if you don't think that is a problem, please > > backport your patch to 2.2.x. I'd also prefer that prefork doesn't wake up unless it has work, but I think the simpler, localized change which was committed will be less trouble in the long run. wrt backporting: I think that modern Linux and modern Solaris, perhaps the only platforms tested so far, both work for the same reason under the covers. Has anyone tested with regular poll()? If not, I can try to test another platform or at least dumb down APR to bypass epoll/event queue/etc. --00163630ee1b24a2c90460496011 Content-Type: text/html; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable On Jan 11, 2009 11:34am, Stefan Fritsch <sf@sfritsch.de> wrote:
= > Hi,
>
>
>
> thanks for following up= on this and sorry for the late response.
>
>
> <= br />> On Wednesday 07 January 2009, Jeff Trawick wrote:
>
> > Initial testing of your idea for a timeout was promising.
&= gt;
>
>
> I couldn't reproduce any hangs und= er linux with the patch you commited
>
> to trunk.
Thanks for all of your work on this issue!

> In my patch = I tried to avoid that an idle apache wakes periodically
>
>= ; for no good reasons. But if you don't think that is a problem, please=
>
> backport your patch to 2.2.x.

I'd also= prefer that prefork doesn't wake up unless it has work, but I think th= e simpler, localized change which was committed will be less trouble in the= long run.

w.r.t. backporting:
I think that modern Linux an= d modern Solaris, perhaps the only platforms tested so far, both work for t= he same reason under the covers. Has anyone tested with regular poll()? I= f not, I can try to test another platform or at least dumb down APR to bypa= ss epoll/event queue/etc. --00163630ee1b24a2c90460496011--