[OpenAFS] AFS client not working any more with TABs in CellServDB

[Christophe BERNARD]

Hello.

I have been running an AFS server for a while. The server is multi-homed,
and used to be working. Recently, I noticed that the AFS server cannot
mount itself any more. Before considering asking to the list, I read the
archives, searched google for anything sensible, and tried a LOT of
workarounds, and got really desperate.

At startup, the afs client daemon failed, giving error messages like 
these:

  afs: Lost contact with file server 10.0.0.1 in cell domain.com (all 
  multi-homed ip addresses down for the server)

and later

  afs: file server 10.0.0.1 in cell domain.com is back up (multi-homed 
  address; other same-host interfaces may still be down)

afsd -verbose said:

  afsd: Mounting the AFS root on '/afs', flags: 0.
  afsd: Can't mount AFS on /afs(22)

This was a client problem: all other clients still could access the AFS
filesystem.

I checked name resolution, the hosts file, all NetCOnfig NetInfo files,
everything was OK, tried to clean-up the cache directory. Still "Can't
mount AFS".

After a lengthy comparison on config files between working clients and the
server's client config, I finally found the fix, which is worth
mentioning: the CellServDB file in /usr/vice/etc had a [TAB] character,
like this:

>domain.com[TAB]#cell name
10.0.0.1 #afs.domain.com

I replaced the [TAB] with spaces, and YES! it worked. So my 2c suggestion
to the openafs team: would it be possible to make CellServDB file parsing
by the client a wee more robust, like correctly skipping tabs like spaces,
and even possibly logging warnings when not-ascii characters are met?

Anyways, many thanks for the great piece of software,

Regards,

Christophe

PS: using redhat/linux
-- 
    Christophe BERNARD - Centre de Morphologie Mathématique
École des Mines de Paris - 35, rue Saint-Honoré - 77305 Fontainebleau cedex
           tél +33-1-64694775   - fax +33-1-64694707
        email bernard@cmm.ensmp.fr - http://cmm.ensmp.fr