[8463] in Info-AFS_Redistribution
RE: Can one find a mount point just knowing the volume name?
daemon@ATHENA.MIT.EDU (Peter Scott)
Thu Feb 8 12:59:32 2001
Message-Id: <4.3.2.7.2.20010208094946.00afe410@mail2a.jpl.nasa.gov>
Date: Thu, 08 Feb 2001 09:52:36 -0800
To: "Neulinger, Nathan R." <nneul@umr.edu>, "'Morris Strongson'" <mms@bnl.gov>,
info-afs@transarc.com
From: Peter Scott <Peter.J.Scott@jpl.nasa.gov>
In-Reply-To: <F349BC4F5799D411ACE100D0B706B3BB34259E@umr-mail03.cc.umr.e
du>
Mime-Version: 1.0
Content-Type: text/plain; charset="us-ascii"; format=flowed
At 11:20 AM 2/8/01 -0600, Neulinger, Nathan R. wrote:
>mount "point" is the problem... There isn't just one mount point for a
>volume, there can be 0, 1, or hundreds...
Bingo.
>If you feel like loading down your server, you can get that information from
>salvager, but it's slow. The alternative is to recursively descend through
>afs space looking for mount points.
I've done this (don't ask for the code, Caltech owns it). It's not easy;
you've got to look out for foreign cell mounts and cyclic symlinks,
symlinks to local filesystems, and make sure you don't go down another
mount to a volume you've already visited. If you have anything like a
large heterogeneous cell, you will be amazed (and disgusted) by some of the
perverse things people do that will trip you up.
--
Peter Scott
Peter.J.Scott@jpl.nasa.gov
