[460] in arla-drinkers
Re: Excessive disk cache activity.
daemon@ATHENA.MIT.EDU (Magnus Ahltorp)
Tue Jan 5 22:28:33 1999
From owner-arla-drinkers@stacken.kth.se Wed Jan 06 03:28:32 1999
Return-Path: <owner-arla-drinkers@stacken.kth.se>
Delivered-To: arla-drinkers-mtg@bloom-picayune.mit.edu
Received: (qmail 28219 invoked from network); 6 Jan 1999 03:28:32 -0000
Received: from unknown (HELO sundance.stacken.kth.se) (130.237.234.41)
by bloom-picayune.mit.edu with SMTP; 6 Jan 1999 03:28:32 -0000
Received: (from majordom@localhost)
by sundance.stacken.kth.se (8.8.8/8.8.8) id EAA14322
for arla-drinkers-list; Wed, 6 Jan 1999 04:24:08 +0100 (MET)
Received: from turbot.pdc.kth.se (turbot.pdc.kth.se [130.237.221.42])
by sundance.stacken.kth.se (8.8.8/8.8.8) with ESMTP id EAA14318
for <arla-drinkers@stacken.kth.se>; Wed, 6 Jan 1999 04:24:03 +0100 (MET)
Received: (from d95-mah@localhost)
by turbot.pdc.kth.se (8.8.7/8.8.7) id EAA21076;
Wed, 6 Jan 1999 04:23:45 +0100 (MET)
To: Werner Krebs <werner.krebs@yale.edu>
Cc: arla-drinkers@stacken.kth.se
Subject: Re: Excessive disk cache activity.
References: <36919011.20DE@yale.edu> <ixd67ampdef.fsf@turbot.pdc.kth.se> <3692D232.D8A@yale.edu>
From: Magnus Ahltorp <map@stacken.kth.se>
Date: 06 Jan 1999 04:23:45 +0100
In-Reply-To: Werner Krebs's message of "Tue, 05 Jan 1999 22:02:10 -0500"
Message-ID: <ixd4sq5axtq.fsf@turbot.pdc.kth.se>
Lines: 31
X-Mailer: Gnus v5.6.45/Emacs 19.34
Sender: owner-arla-drinkers@stacken.kth.se
Precedence: bulk
> - When I would ls an AFS cell (I used stacken.kth.se cell because Yale
> does not have an AFS cell yet) I would get a lot of disk activity (the
> drive did an excessive amount of seeking very rapidly).
>
> This is obviously cache-related, but it is not a good sound (too much
> seeking in a small area of the drive). This seeking activity would
> continue for almost a minute after each simple ls.
It could be the precreation of cache nodes, but then it shouldn't just
happen after each ls, but rather a while after arlad was started. You
haven't enabled xfs debugging or something?
> Also, the operations I was doing (just ls'ing the same small directory a
> few times) should not have required using the on-disk cache at all after
> the first attempt.
It should not write something to the cache, but it will certainly read
from it, since you repeated the ls command.
> How do I add things to the root directory? (arlad -t seemed to mount the
> milko server (it was registering GetRoot fetches), but "ls /afs"
> returned "Not a directory" and I couldn't figure out how one goes about
> actually placing things in the volume.
If you have run "volmgr gen-ontop-vol <partition-name> <volume-id>",
you should have a volume with a top directory. Don't expect too much
though, milko is not much more than a proof-of-concept, and it will be
rewritten.
/Magnus
map@stacken.kth.se
