[OpenAFS] Solaris 9, AFS 1.2.8 kernel panic: read volumeinfo

[Lee Damon]

> > We've just had 3 Solaris 9 systems panic with the same message. No
> > problems for a long time, then suddenly 3 panics within a 20 minute span.
> > I'm suspecting someone migrated from host to host without thinking
> > about it.  Any clues what I should be looking for?  I have a core dump
> > if anyone wants to look at it.
> 
> > May 20 10:25:54 barbet afs: [ID 908391 kern.notice] read volumeinfo
> in afs_UFSGetVolSlot:
> 
> 		tfile = osi_UFSOpen(volumeInode);
>                 code = afs_osi_Read(tfile, sizeof(struct fvolume) *
> tv->vtix,
>                                 &staticFVolume, sizeof(struct fvolume));
>                 if (code != sizeof(struct fvolume))
>                     osi_Panic("read volumeinfo");
>                 osi_UFSClose(tfile);
> 
> Is there any reason reading the cache partition would fail? What's the
> fstab entry for the partition it's on?

I can't think of any. AFS is the only thing that has any business in that
filesystem.  It's mounted as a dedicated FS.  Nothing has changed in
the way we mount the FS in the past 9 months or so.

: || barbet [1] ; grep vice /etc/vfstab
#device         device          mount           FS      fsck    mount   mount
/dev/dsk/c0t0d0s4       /dev/rdsk/c0t0d0s4      /usr/vice/cache ufs     2      
 yes     logging

We're now up to 5 hosts, the first one at 1:45am this morning, the most recent 
at
12:30pm this afternoon.

nomad
 -----------                       - Lee "nomad" Damon -          \
work: nomad@ee.washington.edu                                      \
play: nomad@castle.org    or castle!nomad                           \
                                                                    /\
Sr. Systems Admin, UWEE SSLI Lab                                   /  \
                "Celebrate Diversity"                             /    \