Possible Filesystem Corruption
dillon at apollo.backplane.com
Thu Aug 11 12:58:37 PDT 2005
: I am using HEAD last built about a week ago. For a while I have been
: getting a random number of disks in the line that says syncing disks
: when shutting down. Today I had a few more problems i.e. in /bin/csh
: and /bin/tcsh I got
: /bin/pwd : 1 syntax error: ``( '' unexpected
: also /usr/src was no longer seen as a directory and ls said it had a
: bad file descriptor for /usr/src. Then when shutting down I got a
: panic. I didn't know what to do at db> but typing panic ? got me a
It's possible that you may have caught the system in the middle of
the 'stat' update that Joerg made. That would seriously confuse
the shells if they were not completely recompiled.
The only way to tell is to do a complete buildworld/installworld/
buildkernel/installkernel sequence and to then see if the problem
: reboot and the subsequent fsck seems to have fixed everything I can
: check. There was something about a SUPERBLOCK but it was too fast for
: me to record. Not all filesystems were checked probably because the
: system was already in single user mode when it paniced ?.
: Here is the panic FYI, it is hand transcribed.
:syncing disks... 2
:panic: umount: dangling vnode
:Trace beginning at frame 0xc5423ae0
We need to get a core dump to track down this issue. This is
related to the shutdown code not being able to flush all the buffers.
It could very well be that this *IS* a driver issue, but in that case
I would have expected I/O errors to be reported on the console as well.
More information about the Users