*BSD News Article 47819


Return to BSD News archive

Path: sserve!newshost.anu.edu.au!harbinger.cc.monash.edu.au!simtel!zombie.ncsc.mil!news.mathworks.com!fu-berlin.de!news.belwue.de!news.uni-stuttgart.de!informatik.uni-koblenz.de!ralf
From: ralf@informatik.uni-koblenz.de (Ralf Baechle)
Newsgroups: comp.unix.bsd.netbsd.misc
Subject: Re: NetBSD Filesystems
Date: 2 Aug 1995 16:25:05 GMT
Organization: Uni Koblenz, Germany.
Lines: 21
Distribution: world
Message-ID: <3vo8t1$5r1i@info4.rus.uni-stuttgart.de>
References: <1995Jul26.123455.28242@lssec.bt.co.uk> <MICHAELV.95Jul29005359@MindBender.HeadCandy.com> <3ve9jk$11b8@info4.rus.uni-stuttgart.de> <MICHAELV.95Jul30182230@MindBender.HeadCandy.com> <3vjcve$mah@wolfe.wimsey.com>
Reply-To: ralf@waldorf-gmbh.de
NNTP-Posting-Host: ozzy.uni-koblenz.de

In article <3vjcve$mah@wolfe.wimsey.com>, curt@cynic.portal.ca (Curt Sampson) writes:

|> >This has nothing to do with the filesystem.  NetBSD (and I believe
|> >FreeBSD) supports a flag that will let you mount the filesystem
|> >asynchronously (wrt metadata writes).  However, it is not the default,
|> >and most unix-seasoned people consider it just asking for trouble.

Oh, well, that depends what experiences one made.  I never really nuked a
filesystem unrecoverable except in one case - but WhatereverFS doesn't help
against faulty hardware.

|> It is asking for trouble. If you really need fast writes on lots
|> of small files (i.e., for a news filesystem or something like that)
|> the proper thing is to use the log filesystem or something similar.
|> 
|> Is LFS working in NetBSD?

What experience made people actually using logging filesystems about the
performance and reliability in real live?

  Ralf