[166] in Info-AFS_Redistribution
Salvager (Re: Missing Backup Volumes)
daemon@ATHENA.MIT.EDU (John Gardiner Myers)
Mon Jun 24 12:06:07 1991
Date: Mon, 24 Jun 91 11:03:24 -0400 (EDT)
From: John Gardiner Myers <jm36+@andrew.cmu.edu>
To: Info-AFS@transarc.com
In-Reply-To: <9106181754.AA24709@pit.awdpa.ibm.com>
After a server gets disk corruption, we frequently find it necessary
to run a full salvage *twice*. The first salvage fixes the RW volumes,
but leaves the backup (and possibly readonly) volumes in a wierd
state. Of course, it marks the volume group as "clean".
The second full salvage removes the volume headers of these screwed-up
backup volumes.
And while I'm on the subject of the salvager...
A while back I modified the salvager to be able to salvage up to N
partitions in parallel (where N is user-settable on the command-line)
On a Sun 3/280 server with two Sabres and two Eagles, running with N
set to 4 cut the time for a full salvage from about 10 minutes to
about 6.
We've been running this modified salvager on production servers for a
few months now and it is working quite well. I sent the modifications
to Transarc and they've assigned them a number in the bug database.
--
_.John G. Myers Internet: John.G.Myers@andrew.cmu.edu
(412) 268-2314 LoseNet: ...!seismo!ihnp4!wiscvm.wisc.edu!give!up
