[OpenAFS] Spooky /afs mounts

Eric Chris Garrison ecgarris@iupui.edu
Wed, 13 Jan 2010 10:43:44 -0500

Thank you very much for clearing this up.  I had noticed the difference in
the options on this machine compared to others, it was indeed missing
-dynroot.  I'll check into root.afs as well.

Thanks again,


Simon Wilkinson wrote:
> On 13 Jan 2010, at 15:02, Eric Chris Garrison wrote:
>> When I start the openafs client on one RHEL 4 server, it seems to ignore
>> what's in the CellServDB file, doesn't mount the standard list of sites
>> and our local one too.  It shows under /afs our local cell, a cell that
>> hasn't existed in a couple of years, and a test cell that's not running.
> There are two configurations that afsd can run in. The first is that it
> dynamically generates the contents of /afs (either from the DNS, or from
> your CelLServDB file). The second is that it looks for a volume called
> root.afs in the local cell (the one specified by 'ThisCell'). This
> behaviour is controlled by the
> -dynroot flag. Because using root.afs requires that the network is
> present when AFS starts, it has fallen out of favour in recent years.
> It sounds like your machine with the spooky mounts doesn't have -dynroot
> in its afsd options, and so is mounting root.afs, which has (by the
> sounds of it) bitrotted over the years.
> Cheers,
> Simon.

Eric Chris Garrison             | Principal Mass Storage Specialist
ecgarris@iupui.edu              | Indiana University - Research Storage
W: 317-278-1207 M: 317-250-8649 | Jabber IM: ecgarris@iupui.edu