[OpenAFS-devel] File server consistency check and why was Nuke unhappy?

Harald Barth haba@pdc.kth.se
Thu, 20 Mar 2003 00:41:12 +0100 (CET)


Hi everyone.

Yesterday, I did a vos move volume duxserver a linuxserver a. During
that move someting went wrong (cred expired or something) and then I
was left with an incompletely moved volume. Target volume offline,
VLDB still pointing on the source. Ok, I thought, no harm done, let's
vos zap -force the off-lined one. Later it was noticed, that during
the vos zap I got a lot of stuff like this in the VolserLog:

Wed Mar 19 11:30:29 2003 Nuke: Failed to remove /vicepa/AFSIDat/k/kMc+U/+/+/M+++fpoMfI3
Wed Mar 19 11:30:29 2003 Nuke: Failed to remove /vicepa/AFSIDat/k/kMc+U/+/+/+
Wed Mar 19 11:30:29 2003 Nuke: Failed to remove /vicepa/AFSIDat/k/kMc+U/D1/b5/=wwgF
Wed Mar 19 11:30:29 2003 Nuke: Failed to remove /vicepa/AFSIDat/k/kMc+U/+/+/U+++fpoMfI3
Wed Mar 19 11:30:29 2003 Nuke: Failed to remove /vicepa/AFSIDat/k/kMc+U/+/+/+
Wed Mar 19 11:30:29 2003 Nuke: Failed to remove /vicepa/AFSIDat/k/kMc+U/+/+/+
Wed Mar 19 11:30:29 2003 Nuke: Failed to remove /vicepa/AFSIDat/k/kMc+U/+/+/+
Wed Mar 19 11:30:29 2003 Nuke: Failed to remove /vicepa/AFSIDat/k/kMc+U/+/+/+
Wed Mar 19 11:30:29 2003 Nuke: Failed to remove /vicepa/AFSIDat/k/kMc+U/+/+/+
Wed Mar 19 11:30:29 2003 Nuke: Failed to remove /vicepa/AFSIDat/k/kMc+U/+/+/+
Wed Mar 19 11:30:29 2003 Nuke: Failed to remove /vicepa/AFSIDat/k/kMc+U/+/+/+

The question now is: In spite of the volume being gone, are there any
dangling files that should be cleaned up? The volume in question was a
big one (data amd number of files). I want to know if vos zap did
leave trash behind. So how can I list all the /vicepx/AFSIDat/*
filenames of a volume? That way I could compare that list to the list
of files I actually have on disk. No, I don't want to run salvage just
now, because that means shutting down the server and I don't want to
do that without knowing if I have to.

This is current OpenAFS devel built 2003-03-18 on Linux 2.4.20. I'd be
very glad for some leads as I'd like to look into this later today
(MET) when awake again.

Thanks,
Harald.