[26788] in Athena Bugs
Re: sun4 9.4.23: evolution
daemon@ATHENA.MIT.EDU (Camilla R Fox)
Wed Mar 29 01:31:07 2006
Message-Id: <200603290630.k2T6UvEp001889@red-herring.mit.edu>
To: "Richard J. Barbalace" <rjbarbal@mit.edu>
In-Reply-To: Your message of "Tue, 28 Mar 2006 18:40:42 EST."
<20060328184042.m38u4p008kwssk88@webmail.mit.edu>
Date: Wed, 29 Mar 2006 01:30:57 -0500
From: Camilla R Fox <cfox@mit.edu>
X-Spam-Score: 1.217
X-Spam-Level: * (1.217)
X-Spam-Flag: NO
Cc: bugs@mit.edu
Errors-To: bugs-bounces@mit.edu
> I experience the same when the number of messages is not too high; I currently
> have about 3400 messages and see similar delays. But when it gets higher, the
> performance deteriorates rapidly. It seems highly non-linear as the quota
> limit is approached.
> I have not such a dramatic slowdown with other applications I have tried,
> including ftp, web browsing, etc.
I've hit weird failures with a mozilla that's trying to write into
config files in an afs homedir from a machine behind the tunnel.
Sometimes running 'fs checks' a lot will bring it back. (I work around
it by caching a subset of stuff locally, and reading my mail ssh'd into
a machine on campus.)
I had chalked it up to being a combination of the tunnel and the
extremely asymmetrical bandwidth that comcast offers. (As well as AFS,
xzul on a big .anyone can wedge horribly.)
> Knowing relatively little about the network, how would NAT affect the problem?
One common NAT failure is that your afs client doesn't get callbacks
from one or more fileservers, when it's been too long since you've
connected to them. The result is that if you're editing a file
elsewhere, you'll see an older cached copy of the file on the machine
behind the NAT.
