From owner-freebsd-hackers Fri Jun 20 07:30:44 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id HAA05461 for hackers-outgoing; Fri, 20 Jun 1997 07:30:44 -0700 (PDT) Received: from scanner.worldgate.com (scanner.worldgate.com [198.161.84.3]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id HAA05444; Fri, 20 Jun 1997 07:30:37 -0700 (PDT) Received: from znep.com (uucp@localhost) by scanner.worldgate.com (8.8.5/8.8.5) with UUCP id IAA02329; Fri, 20 Jun 1997 08:30:25 -0600 (MDT) Received: from localhost (marcs@localhost) by alive.znep.com (8.7.5/8.7.3) with SMTP id IAA17616; Fri, 20 Jun 1997 08:31:10 -0600 (MDT) Date: Fri, 20 Jun 1997 08:31:10 -0600 (MDT) From: Marc Slemko To: Josef Karthauser cc: questions@FreeBSD.ORG, hackers@FreeBSD.ORG, isp@FreeBSD.ORG Subject: Re: Apache 1.1.3 and 1.2.0 problems under FBSD2.2.2 In-Reply-To: <19970620104829.34842@pavilion.net> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-hackers@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk Please, limit your crossposting. _ONE_ mailing list is appropriate. If something is appropriate to another list, it is almost never appropriate to questions. Also it is appreciated if you don't ask a question publicly and submit it as an Apache bug at the same time. Ask it publicly first and wait to get answers. It avoids wasting limited developer time. On Fri, 20 Jun 1997, Josef Karthauser wrote: > The continuing saga of a sick web machine :( > > A brief history. We're running a P200pro FBSD2.2.2 machine here as > a virtual web server. It currently has 130ish web servers running, > each with their own IP address. > > Where did the problem start? I'm not sure, but it was some combination > of reaching 124ish web servers, and hitting a certain number of active > http connections. > > The web server is basically fine, until some limit gets reached and the > server then just hangs, with lots of sockets still active. You are running out of file descriptors. Just because you change the ulimit does _NOT_ mean you can actually use all of them. You are probably running into the FD_SETSIZE limit, which limits (among other things) the number of fds that select() can handle. It is still 255. Under 2.2, I think you should be able to just redefine FD_SETSIZE before compiling Apache to increase that. The reason that 1.1 works and 1.2 doesn't is probably because file descriptors are allocated in a different order in 1.2 for various reasons. You don't make it entirely clear, but I am assuming that just Apache is hanging, not the whole machine. -- Marc Slemko | Apache team member marcs@znep.com | marc@apache.org