[8725] in Info-AFS_Redistribution
Re: automating 'vos release'
daemon@ATHENA.MIT.EDU (Jeffrey Hutzelman)
Wed Jun 13 02:43:55 2001
Date: Wed, 13 Jun 2001 02:35:51 -0400 (EDT)
From: Jeffrey Hutzelman <jhutz@cmu.edu>
To: Kim Kimball <kim@ccre.com>
cc: David R Boldt <dboldt@usgs.gov>, info-afs@transarc.com
In-Reply-To: <008c01c0cc0d$d5ab7a40$9436140a@worldnet.att.net>
Message-ID: <Pine.LNX.4.21L-021.0106130227370.901-100000@manticore.andrew.cmu.edu>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
On Mon, 23 Apr 2001, Kim Kimball wrote:
> 1. Vos release does "no op" but it takes quite a while for AFS to decide
> that all the ROs are up to date. In our environment, we release thousands
> of volumes weekly, automatically. We sometimes have 3-5 thousand requests
> in a single afternoon. The AFS-native "no op" decision took about 30
> seconds per volume. The mechanism below takes 3-10 seconds per volume. For
> large numbers of requests, the time savings is significant.
The reason this is so slow is because it is _not_ a no-op. A vos release
always results in a complete reclone of the RO copy located on the server
that houses the RW volume. Once the clone has been updated, an
incremental dump is done to update the other servers. The cloning
operation itself is somewhat expensive, and for a volume with lots of
directories, so is the incremental dump.
You're considerably better off if you skip the 'vos release' entirely for
volumes whose RW and RO volumes have the same update time.
-- Jeffrey T. Hutzelman (N3NHS) <jhutz+@cmu.edu>
Sr. Research Systems Programmer
School of Computer Science - Research Computing Facility
Carnegie Mellon University - Pittsburgh, PA
