Return-Path: Delivered-To: apmail-httpd-users-archive@httpd.apache.org Received: (qmail 71012 invoked by uid 500); 22 Aug 2002 01:33:54 -0000 Mailing-List: contact users-help@httpd.apache.org; run by ezmlm Precedence: bulk Reply-To: users@httpd.apache.org list-help: list-unsubscribe: list-post: Delivered-To: mailing list users@httpd.apache.org Received: (qmail 70990 invoked from network); 22 Aug 2002 01:33:53 -0000 Received: from unknown (HELO artemis.webmagix.net) (213.239.154.33) by daedalus.apache.org with SMTP; 22 Aug 2002 01:33:53 -0000 Received: from localhost (localhost [127.0.0.1]) by artemis.webmagix.net (Postfix) with ESMTP id 33AB6197398 for ; Thu, 22 Aug 2002 03:33:51 +0200 (CEST) Received: from kees (adsl219142.adsl.utwente.nl [130.89.208.240]) by artemis.webmagix.net (Postfix) with SMTP id 693B5197348 for ; Thu, 22 Aug 2002 03:33:50 +0200 (CEST) From: "Kees Hoekzema" To: Subject: RE: How to scale apache to deal with network outages Date: Thu, 22 Aug 2002 03:35:40 +0200 Message-ID: MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit X-Priority: 3 (Normal) X-MSMail-Priority: Normal X-Mailer: Microsoft Outlook IMO, Build 9.0.2416 (9.0.2911.0) X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000 Importance: Normal In-Reply-To: <3D6429EE.30405@expertcity.com> X-Virus-Scanned: by AMaViS snapshot-20020300 X-Spam-Rating: daedalus.apache.org 1.6.2 0/1000/N hey, At the moment we deal with it with low timeout values, the surrent timeout is 10 seconds. We have around 20M requests daily, server by 8 apache servers with the settings: Timeout 10 MinSpareServers 30 MaxSpareServers 50 StartServers 40 Okay, those servers have all 1G on the mainboard and a loadbalancer in front of them, but those settings should work with high loaded websites. -kees > -----Original Message----- > From: Steve Francis [mailto:steve@expertcity.com] > Sent: Thursday, August 22, 2002 2:02 AM > To: users@httpd.apache.org > Subject: How to scale apache to deal with network outages > > > How do large scale sites deal with network outages that consume all the > apache servers? > > We run apache 2.0.39 - normally there's about 50 active processes. If a > network glitch appears (internal ATT routing loop, BGP sesssions > flapping on a provider, etc) then the apache servers get stuck dealing > with the connection they are currently servicing, and wait for TCP > retransmissions and ultimately the connection to die before releasing > the connection and thus servicing the next client. > During these times apache can grow rapidly to the max servers, and then > stop servicing anyone until the interrupted connections expire. > > How do large sites deal with this - we can't just increase the max > servers, else we run out of physical memory. > > TIA > > > --------------------------------------------------------------------- > The official User-To-User support forum of the Apache HTTP Server Project. > See for more info. > To unsubscribe, e-mail: users-unsubscribe@httpd.apache.org > " from the digest: users-digest-unsubscribe@httpd.apache.org > For additional commands, e-mail: users-help@httpd.apache.org > --------------------------------------------------------------------- The official User-To-User support forum of the Apache HTTP Server Project. See for more info. To unsubscribe, e-mail: users-unsubscribe@httpd.apache.org " from the digest: users-digest-unsubscribe@httpd.apache.org For additional commands, e-mail: users-help@httpd.apache.org