[OpenAFS] VLDB corrupt? Help

[Russ Allbery]

Chaskiel M Grundman <cg2v@andrew.cmu.edu> writes:

> One possibility is that the hash chains in the vldb are corrupt. you can
> attempt to repair this using the vlclient tool (vlclient isn't
> installed, so you will have to build openafs from source and run
> vlclient from the build tree)

> run vlclient as 'vlclient -cell <cellname> -host <vldb.sync.site>',
> otherwise it will complain or be unable to update the vldb.  at the vl>
> prompt, type 'fixhash unused' and press return. it will take 2 minutes
> or so. You can run 'checkhash unused' first if you just want to see
> what, if any, entries are broken.

Do you think that would also be the solution to this problem?

(root) windlord:~> vos listvldb -server 171.64.14.164
VLDB entries for server 171.64.14.164 

user.rambi.bad 
    RWrite: 1936956182    Backup: 1936956184
    number of sites -> 1
       server 171.64.14.164 partition /vicepa RW Site 

Total entries: 1
(root) windlord:~> vos delentry user.rambi.bad
Could not delete entry for volume user.rambi.bad
You must specify a RW volume name or ID (the entire VLDB entry will be deleted)
VLDB: no such entry
Deleted 0 VLDB entries
(root) windlord:~> vos delentry 1936956182
Could not delete entry for volume 1936956182
You must specify a RW volume name or ID (the entire VLDB entry will be deleted)
VLDB: no such entry
Deleted 0 VLDB entries

I can vos changeloc the bad volume and even vos rename it, but I can't
manage to get the VLDB entry to go away.  I think I even tried moving it
to a live server and doing vos syncvldb at one point, although I might be
misremembering.

I've been hesitant to do anything that might possibly cause any sort of
production problem, since right now it's just an annoyance and not
actually causing any harm.  It's been like this for more than ten years,
and various people with my job have run across it, poked at it, and then
decided to ignore it some more.

-- 
Russ Allbery (rra@stanford.edu)             <http://www.eyrie.org/~eagle/>