Mailing-List: contact dev-help@httpd.apache.org; run by ezmlm
Precedence: bulk
Reply-To: dev@httpd.apache.org
Received-SPF: pass (hermes.apache.org: domain of trawick@gmail.com designates
 64.233.170.200 as permitted sender)
DomainKey-Signature: a=rsa-sha1; q=dns; c=nofws;
        s=beta; d=gmail.com;
        h=received:message-id:date:from:reply-to:to:subject:in-reply-to:mime-version:content-type:content-transfer-encoding:references;
        b=fA+kYNlcN7Aq6gWrfrem+vtgDNHNslctuQBnQLLRdDdqVNQeCazrxbrioHLL0sDYkM1LBbXjYX8XpG9Lp03uacJqYGhJcn6xVclXq/sPr1FuNRnN7feo+5RjoTeQc6rr8J3zSGNaeOIqPFw3L3G5Z0ukb+473RUsNTLm0RfT5P0=
Message-ID: <cc67648e050317160175a8a153@mail.gmail.com>
Date: Thu, 17 Mar 2005 19:01:49 -0500
From: Jeff Trawick <trawick@gmail.com>
Reply-To: Jeff Trawick <trawick@gmail.com>
To: dev@httpd.apache.org
Subject: Re: do we still want sendfile enabled with our default conf files?
In-Reply-To: <20050317152754.GI5790@scotch.ics.uci.edu>
Mime-Version: 1.0
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: 7bit
References: <cc67648e05031704575741b217@mail.gmail.com>
	 <42398A5C.7000400@slive.ca>
	 <20050317141250.GA28555@castlerea.stdlib.net.>
	 <20050317152754.GI5790@scotch.ics.uci.edu>

On Thu, 17 Mar 2005 07:27:54 -0800, Justin Erenkrantz
<justin@erenkrantz.com> wrote:
> On Thu, Mar 17, 2005 at 02:12:50PM +0000, Colm MacCarthaigh wrote:
> > On Thu, Mar 17, 2005 at 08:47:08AM -0500, Joshua Slive wrote:
> > > +1, as before.
> > >
> > > From the users' perspective, sendfile results in unexplained corruption
> > > or uninterpretable error messages pretty much any time a network
> > > filesystem is used to host content (and random other times on win32).
> >
> > Judging by my inbox over the last year, similar behaviour on Linux with
> > IPv6 is becoming more and more prevalant, as adoption ramps up. So
> > a default of off would be a very great help there too.
> 
> These seem like broken OSes and not a suitable justification to disable
> sendfile.  We should fix the code - perhaps by teaching APR not to enable the
> sendfile-variants on these buggy platforms - not disable it entirely.  For
> those platforms that don't have bugs, disabling sendfile would be a ridiculous
> performance hit.  -1.  -- justin

Is that "-1" a vote, or a veto against the idea?  If the latter,
please explain in at least a little detail how a technical solution
can be implemented that will avoid some of the types of problems
triggered by the use of sendfile.  After a year or two that these
issues have been known, the only thing that anybody truly knows how to
do is to disable sendfile.