From owner-freebsd-hackers Sat Jan 11 18:10:10 1997 Return-Path: Received: (from root@localhost) by freefall.freebsd.org (8.8.4/8.8.4) id SAA24479 for hackers-outgoing; Sat, 11 Jan 1997 18:10:10 -0800 (PST) Received: from crh.cl.msu.edu (crh.cl.msu.edu [35.8.1.24]) by freefall.freebsd.org (8.8.4/8.8.4) with ESMTP id SAA24474 for ; Sat, 11 Jan 1997 18:10:07 -0800 (PST) Received: (from henrich@localhost) by crh.cl.msu.edu (8.8.4/8.8.4) id VAA09833; Sat, 11 Jan 1997 21:09:57 -0500 (EST) Date: Sat, 11 Jan 1997 21:09:57 -0500 (EST) From: Charles Henrich Message-Id: <199701120209.VAA09833@crh.cl.msu.edu> To: terry@lambert.org, freebsd-hackers@freebsd.org Subject: Re: mount -o async on a news servre Newsgroups: lists.freebsd.hackers References: <5b94ac$2863@msunews.cl.msu.edu> X-Newsreader: NN version 6.5.0 #1 (NOV) Sender: owner-hackers@freebsd.org X-Loop: FreeBSD.org Precedence: bulk In lists.freebsd.hackers you write: >If I were running a big site, I would be tempted to run a reading >host and a *seperate* posting host. >A reading host would be, by definition, a passive news spool with no >locally generated data; I'd run the FS for the news spool in async mode. >Using a seperate server for posting, I could run the posting host >with atime and sync... so my users postings would not be lost because >of a crash. That would be (mostly) useless. News servers spend (as a percentage of time) very little time writing, and massive amounts of time reading. Unless im mistaken, with noatime set, meta data is almost never created, and even then only updated in the case of the scheduled sync() call. Since this can also be pushed back to happen fairly infrequently (I run every 300 seconds, on advice from Joe, and scientific [heh] exploration by myself) it doesnt effect the news spool hardly at all. From my (admittedly almost nonexistant) knowledge of the FS, the only remaining metadata updating going on is on directory creation, which only happens at new newsgroup time (almost never relativly speaking). For performance reasons you would tend to break apart the standard "USER newsreader/poster" operation and "newsfeed" operations, both of which are read/write, both of which have many times the reads as writes. In other words, if you run as suggested back a few lines, there performance loss of running async should be damn close to zero.. -Crh -- Charles Henrich Michigan State University henrich@msu.edu http://pilot.msu.edu/~henrich