[OpenAFS] vos move "forgot" to remove .backup volume.

Harald Barth haba@kth.se
Fri, 13 Mar 2009 12:16:30 +0100 (CET)


A normal vos move looks like this:

vos move prj.sbc.abernsel13 trevally a valine a -v -local
------
Starting transaction on source volume 537080985 ... done
Allocating new volume id for clone of volume 537080985 ... done
Cloning source volume 537080985 ... done
Ending the transaction on the source volume 537080985 ... done
Starting transaction on the cloned volume 537097245 ... done
Setting flags on cloned volume 537097245 ... done
Getting status of cloned volume 537097245 ... done
Creating the destination volume 537080985 ... done
Setting volume flags on destination volume 537080985 ... done
Dumping from clone 537097245 on source to volume 537080985 on destination ... done
Ending transaction on cloned volume 537097245 ... done
Starting transaction on source volume 537080985 ... done
Doing the incremental dump from source to destination for volume 537080985 ...  done
Setting volume flags on old source volume 537080985 ... done
Setting volume flags on new source volume 537080985 ... done
Ending transaction on destination volume 537080985 ... done
Releasing lock on VLDB entry for volume 537080985 ... done
Setting forwarding pointer for volume 537080985 ... done
Deleting old volume 537080985 on source ... done
Ending transaction on old volume 537080985 on the source ... done
Creating transaction for backup volume 537080987 on source ... done
Setting flags on backup volume 537080987 on source ... done
Deleting the backup volume 537080987 on the source ... done
Ending transaction on backup volume 537080987 on source ... done
Starting transaction on the cloned volume 537097245 ... done
Deleting the cloned volume 537097245 ... done
Ending transaction on cloned volume 537097245 ... done
Volume 537080985 moved from trevally /vicepa to valine /vicepa

Observe the "Creating transaction for backup volume .... Ending transaction on
backup volume" information. Now I have a vos move some minutes later
which looks like this:

vos move prj.sbc.aronh09 trevally a valine a -v -local
------
Starting transaction on source volume 537084429 ... done
Allocating new volume id for clone of volume 537084429 ... done
Cloning source volume 537084429 ... done
Ending the transaction on the source volume 537084429 ... done
Starting transaction on the cloned volume 537097246 ... done
Setting flags on cloned volume 537097246 ... done
Getting status of cloned volume 537097246 ... done
Creating the destination volume 537084429 ... done
Setting volume flags on destination volume 537084429 ... done
Dumping from clone 537097246 on source to volume 537084429 on destination ... done
Ending transaction on cloned volume 537097246 ... done
Starting transaction on source volume 537084429 ... done
Doing the incremental dump from source to destination for volume 537084429 ...  done
Setting volume flags on old source volume 537084429 ... done
Setting volume flags on new source volume 537084429 ... done
Ending transaction on destination volume 537084429 ... done
Releasing lock on VLDB entry for volume 537084429 ... done
Setting forwarding pointer for volume 537084429 ... done
Deleting old volume 537084429 on source ... done
Ending transaction on old volume 537084429 on the source ... done
Creating transaction for backup volume 537084431 on source ... done
Starting transaction on the cloned volume 537097246 ... done
Deleting the cloned volume 537097246 ... done
Ending transaction on cloned volume 537097246 ... done
Volume 537084429 moved from trevally /vicepa to valine /vicepa

Here we have the "Creating transaction for backup volume ..." but then
the backup volume does not occur in the output from vos move any more.
And it is left:


# vos listvol trevally
Total number of volumes on server localhost partition /vicepa: 5 
...
prj.sbc.aronh09.backup            537084431 BK    9250606 K On-line
...

But the real volume is somewhere else:

# vos listvol valine
...
prj.sbc.aronh09                   537084429 RW    9250606 K On-line
prj.sbc.aronh09.backup            537084431 BK    9250606 K On-line
...


In the volser log, I have a lot of messages like this:

Fri Mar 13 02:47:06 2009 trans 248 on volume 537084431 is older than XXXX seconds

which end at XXXX = 7350. After that silence in the logfile.

So is there some assumption in vos that this transaction can not take
longer than X and is the silently aborted? 

Some info about the volume: diskused: 9250606KB, filecount: 2145921,
space used in vicepa ~13GB.

Should I add something to my volume move script that actually checks
if a .backup is really gone after a vos move? I'd rather not need to
do that, but I don't like to have 13GB junk chunks lying around
either. I bet this will screw things up if I tried to move the volume
back where it came from, too.

Harald.

PS, trivia: trevally is a fish and valine is an amino acid.