[OpenAFS-devel] cvs build 7FEB06 - stuck in salvaging filesystem mode

[Christopher D. Clausen]

I just built openafs from CVS (as recomened by Mr. Altman and with a 
patch from Mr. Brashear) on sun4x_510.

It installed fine, but as soon as afs started, a salvage started (normal 
operation) and seems to be in the "salvaging filesystem" state forever. 
Also, I don't see the salvager process running like I do on my other 
systems (during the salvage.)  If I run the salvager -force manually 
(not sure if this is good to do or not) it finishes after writing to the 
SalvageLog, but bos still thinks that the salvager is running.

If I then bos shutdown; pkill -9 bosserver; /usr/afs/bin/bosserver, the 
same thing happens again with bos seemingly stuck in "salvage mode."

There is no data on this fileserver, so I'm not in a huge hurry to get 
it working and I can always revert to an older build.  But can anyone 
recomend a way to usefully debug this issue?  Running salvager without 
the -force results in:
[root@alnitak:/usr/afs/bin]# ./salvager
Exiting immediately without salvage. Look into the FileLog to find 
volumes which really need to be salvaged!

I assume bosserver should note that message and immediately start the 
fileserver process, but that doesn't seem to happen.  And as I said, 
manually forcing a salvage of an empty vice partition doesn't seem to 
help either.

Could this be a result of changes within CVS?  Or the patch attached to:
http://rt.central.org/rt/Ticket/Display.html?id=25747

built using:
[cclausen@alnilam:/local/scratch]% CC -V
CC: Sun C++ 5.8 2005/10/13
[cclausen@alnilam:/local/scratch]% uname -a
SunOS alnilam.acm.uiuc.edu 5.10 Generic_118822-25 sun4u sparc 
SUNW,Ultra-4

with automake and autoconf packages from blastwave.org (to generate 
configure script using regen.sh)

CC=cc 
./configure --enable-debug-lwp --enable-bos-new-config --enable-fast-restart 
 --enable-supergroups --enable-bitmap-later --with-krb5-conf=/usr/local/bin/krb5-config 
 --enable-transarc-paths

Is there a way to force bosserver to start the filserver without running 
a salvage (I thought thats what the --enable-fast-restart option was 
for)?

-----

machine is currently up in this state:
[root@alnitak:/usr/afs/bin]# bos status afs1 -local
Instance fs, currently running normally.
    Auxiliary status is: salvaging file system.
Instance buserver, currently running normally.
Instance ptserver, currently running normally.
Instance vlserver, currently running normally.

The vlserver and ptserver appear to be working fine, there are no 
unusual messages in VLLog, BosLog, PtLog or BackupLog.  The SalvageLog 
contains only:
[root@alnitak:/usr/afs/logs]# cat SalvageLog
@(#) OpenAFS devel built  2006-02-07
02/07/2006 14:51:58 STARTING AFS SALVAGER 2.4 (/usr/afs/bin/salvager)

and no salvager process is currently running:
[root@alnitak:/usr/afs/logs]# ps -ef |grep "[/]usr/afs"
root   771   767   0 14:51:58 ? 0:00 /usr/afs/bin/vlserver
root   770   767   0 14:51:58 ? 0:00 /usr/afs/bin/ptserver
root   769   767   0 14:51:58 ? 0:00 /usr/afs/bin/buserver
root   767     1   1 14:51:58 ? 0:39 /usr/afs/bin/bosserver

-----

(If people would prefer this attached to either or both of the tickets 
in RT or on the -info list, feel free to yell at me and I'll do just 
that.)

<<CDC
-- 
Christopher D. Clausen
ACM@UIUC SysAdmin