OSEC

Neohapsis is currently accepting applications for employment. For more information, please visit our website www.neohapsis.com or email hr@neohapsis.com
 
Subject: Re: Scalability metrics?
From: Craig Sanders (castaz.net.au)
Date: Mon Feb 14 2000 - 18:04:17 CST

On Sat, Feb 12, 2000 at 11:05:13PM +0200, Liviu Daia wrote:
> > SGI's XFS does (and I have personally witnessed it handling 100,000
> > files in a single directory without any problems whatsoever), but
> > it is not yet available outside of SGI, except for some source code
> > that they've released but which has not yet been incorporated by
> > anyone I know of.
>
> Reiserfs is supposed to have both. It doesn't appear to be stable yet
> though.

reiserfs has been extremely stable in my testing of it so far (i'm using
it mainly for squid cache drives on large squid proxy servers, also
on one web server and one postgresql server). the proxies have been
extensively tested and are running as our main proxy servers, the web
server has had less testing, and i've only just started testing the
postgres server. i've had no problems at all so far.

i also know of people who have been using reiserfs routinely on their
workstations and servers for nearly a year.

IMO, it's stable enough. and it's damn fast. and it recovers from
simulated crashes (aka "pulling the power plug while the disks are
thrashing") in seconds - no more waiting ages for e2fsck to run...that's
a winner for me, to minimise downtime in the event of a disaster.

i'll be switching to it wherever it makes sense to do so. i'm already
planning to rebuild my main mail server with reiserfs (but first i'll
build a test mail server and write some scripts to simulate heavy mail
load on it :).

it also seems like people are starting to use it for postfix queues -
there was a bugfix a week or two ago for a small file fsync bug which
was exposed by postfix.

craig

ps: i've been following the discussions on NFS, and was wondering if
anyone had used CODA fs as a substitute? 

--
craig sanders