From owner-freebsd-hackers Thu Apr 3 14:18:13 1997 Return-Path: Received: (from root@localhost) by freefall.freebsd.org (8.8.5/8.8.5) id OAA20206 for hackers-outgoing; Thu, 3 Apr 1997 14:18:13 -0800 (PST) Received: from news.IAEhv.nl (root@news.IAEhv.nl [194.151.64.4]) by freefall.freebsd.org (8.8.5/8.8.5) with SMTP id OAA20201 for ; Thu, 3 Apr 1997 14:18:10 -0800 (PST) Received: from LOCAL (uucp@localhost) by news.IAEhv.nl (8.6.13/1.63) with IAEhv.nl; pid 8112 on Fri, 4 Apr 1997 00:18:00 +0200; id AAA08112 efrom: devet@adv.IAEhv.nl; eto: UNKNOWN Received: by adv.IAEhv.nl (8.7.5/1.63) id VAA10059; Thu, 3 Apr 1997 21:10:27 +0200 (MET DST) Date: Thu, 3 Apr 1997 21:10:27 +0200 (MET DST) From: Arjan.deVet@adv.IAEhv.nl (Arjan de Vet) Message-Id: <199704031910.VAA10059@adv.IAEhv.nl> To: bakul@torrentnet.com Subject: Re: apache like preforking apps and high loads X-Newsgroups: list.freebsd.hackers In-Reply-To: <199704031653.LAA20772@chai.plexuscom.com> References: <199704031527.HAA03657@root.com> Organization: Internet Access Eindhoven, the Netherlands Cc: freebsd-hackers@freebsd.org Sender: owner-hackers@freebsd.org X-Loop: FreeBSD.org Precedence: bulk In article <199704031653.LAA20772@chai.plexuscom.com> you write: >To guido: For apache like apps one idea is to have one process be >the acceptor and have it pass a new socket to individual server >processes (including doing some initial processing to determine >which process should get this socket if not all server processes are >equivalent). That was one of the first approaches NCSA used for preforked HTTP daemons: one parent process which does an accept() and which passes the filedescriptor to one of the child processes. I don't know about more recent versions of NCSA but Apache now uses one parent process which only watches the number of free child processes. The child processes all block on accept(), waiting for a connection to arrive. Arjan