S.J. Borrill wrote:
> 
> I'm getting an increasing amount of instability on my main RiscBSD server
> which has hitherto been faultlessly reliable. In the past few days I've
> had a couple of the following errors:
> 
> cleaned vnode: type VBAD, usecount 0, writecount 0, refcount 0
>         tag VT_UFS, ino 646002, on dev 16, 4
> panic: cleaned vnode isn't
> 
> Then it drops to the debug prompt.
> 
> It seems to have happened, since a bug in a certain NFSClient highlighted
> a bug in nfsd which gave rise to undeletable and unlistable files (bad
> file descriptor) though this might be a red herring.
> 
> The kernel is a 1.2E kernel (5328 IIRC)

Wierd :) Is it always that inode that's proving problematic? If so, you
could boot single user, clear it with iclr, then fsck the disc. Does
fsck seem happy with everything? fsck has a couple of outstanding
problems which can sometimes lead to a fault not being found when there
is one there, and you may have hit that.

Alternatively, if not, it could always be a hardware problem...

How can you create undeletable files? Unlistable? Wouldn't fsck pick
them up? If not, try and find their i-node number then iclr them and
refsck. If you can't find their number then I guess you could move the
other files out, remove the directory and then recreate the directory
whgich would leave them unreferenced which would hopefully mean that
fsck will zap them for you. If you can't remove the directory, try
taking the files out then iclring the directory.

Note I'm not a sysadmin and these are just some thoughts, I'd wait for
someone more knowledgable than I to back them up before trying. Scott,
does this make sense?!

	Neil