Hi Gary, On 5/1/2018 3:23 PM, Gary Gatling wrote: > Hello, > > I was wondering if any users or developers have a solution for moving > all the volumes from one server to another?
As an AuriStorFS licensee you have access to the AuriStorFS vos command
and its "vos eachvol" command. To move RW volumes you need the "vos
move" operation and for RO volumes you need the "vos movesite" operation.
vos eachvol -server <server> -partition <partition> -ro \
-format "vos movesite -roid %n -fromserver %s -frompartition %p \
-toserver <server> -topartition <partition>" \
-execute
vos eachvol -server <server> -partition <partition> -rw \
-format "vos move -id %n -toserver <server> \
-topartition <partition>" \
-execute
will move all of the RO volumes and then all of the RW volumes.
> You have to remember to run it inside of screen or tmux. If you forget
> to do that, or you press CTRL-C to try to interrupt the script it passes
> that CTRL-C down to vos command and volumes get corrupted. Then you
> might have to go to tape backup.
If the volumes are becoming corrupted that is due to bugs in the
volserver not "vos". OpenAFS volservers do not respond well to volume
transactions being interrupted or when the partition fills before
completion. They leave behind volume clones and partially populated
volumes that must be manually cleaned up.
The AuriStorFS volserver is much more resilient and will automatically
clean up in the face of interrupted operations.
> Does anyone have a solution they use for this? Something that could be
> safely interrupted?
See above.
> The reason I ask is that I know of no other way to
> handle things such as Red Hat kernel upgrades where a reboot is
> required. Having an outage for kernel upgrades is problematic for us
> since so much other technology depends on having afs working at all
> times. The kernel updates seem to happen about every 2 weeks or so.
Jeffrey Altman
<<attachment: jaltman.vcf>>
smime.p7s
Description: S/MIME Cryptographic Signature
