[OpenAFS] Problem with mass-moving volumes between 1.5.10 and .12 servers

Jakub Witkowski jpw@wszib.edu.pl
Wed, 10 Jan 2007 08:45:18 +0100

Content-Type: text/plain
Content-Transfer-Encoding: quoted-printable


I am trying to move 7000+ volumes off our main afs server to free it for
a radical update. Previously (i.e. with 1.2 series on the source and
1.4.2 on the target server) this only meant running a simple script
invoking vos move, and just keeping session's tokens fresh for the
duration of mass move.

With 1.5.12 installed on the target server, this procedure no longer
works well; after about 150 volumes moved vos bails out with=20

Failed to move data for the volume 536895796
   : No such file or directory
vos move: operation interrupted, cleanup in progress...


Failed to move data for the volume 536871217
   Possible communication failure
vos move: operation interrupted, cleanup in progress...

After that, running the script will move a few volumes, usually five too
fifteen, then bail out with the very same problem.

I noticed that restarting target server, be it via bos restart or via
bos salvage does reset that behaviour, and again I can move another
large batch of volumes.

A little about our enviroinment:
'source' server: HP DL380, Fedora Core 1, kernel, OpenAFS
1.5.10, all vice partitions are ext3

'target' server: HP DL140, Ubuntu 6.06, kernel 2.6.19, OpenAFS 1.5.15,
single 300GB sata2 disk (This might be part of the problem), jfs

Both servers are connected to a gigabit switch, there are no other
elements in the path.

I still have ~6000 volumes to move, so please, anything you can suggest
to determine the root of the problem?

Jakub Witkowski.

Content-Type: application/pgp-signature; name=signature.asc
Content-Description: To jest =?UTF-8?Q?cz=C4=99=C5=9B=C4=87?= listu
	podpisana cyfrowo

Version: GnuPG v1.4.3 (GNU/Linux)