On Dec 20, 2007 11:39 AM, Jean-Francois Malouin
<Jean-Francois.Malouin AT bic.mni.mcgill DOT ca> wrote:
> Dec 20 10:14:24 grumio automount[7198]: failed to mount /data/rbohbot
> Dec 20 10:14:24 grumio automount[7199]: failed to mount /data/bohbot/rbohbot1
> ...
> about 10 of them.
>
> On the client runtar was spawn at 10:14:24
>
> ~# cat /tmp/amanda-av/client/right2/runtar.20071220101424.debug
> runtar: debug 1 pid 7218 ruid 666 euid 0: start at Thu Dec 20 10:14:24 2007
> runtar: time 0.000: version 2.5.2p1
> /bin/tar version: tar (GNU tar) 1.14
>
> config: right2
> runtar: debug 1 pid 7218 ruid 0 euid 0: rename at Thu Dec 20 10:14:24 2007
> running: /bin/tar: 'gtar' '--create' '--file' '-' '--directory'
> '/export_raid12/bicadmin1/amanda-rsync' '--one-file-system'
> '--listed-incremental'
> '/opt/amanda/var/amanda/right2/gnutar-lists/grumio_export_raid12_bicadmin1_amanda-rsync_0.new'
> '--sparse' '--ignore-failed-read' '--totals' '.'
> runtar: time 0.061: pid 7218 finish time Thu Dec 20 10:14:24 2007

This 'runtar' invocation is for a different DLE --
/export_raid12/bicadmin1/amanda-rsync.  It's possible this is an error
in amanda or tar, but I would need to see a rundump, sendbackup, or
sendsize log that either includes the 'r', or includes the DLE without
the 'r' (in which case we can look to a bug in tar).  While I'm asking
for info -- what systems are these?

My suspicion remains with the automounter, and that it is for some
reason falling back to the 'r'-bearing fiename when the normal mount
fails.  If Amanda or tar were doing this, then it would not be
sporadic.

Dustin

-- 
Storage Software Engineer
http://www.zmanda.com