On 4/24/2014 6:25 PM, Martinx - ジェームズ wrote:
If the Cinder volumes _are unattached_ (and are based on LVM2), you can do this:
Yes, I know. With nfs backend I copy attached boot volumes back and forth -- as long as the guest is powered off -- and then edit volumes.provider_location in the db and restart some services on the compute node. (Not sure which: last time I timed it to kernel upgrade so I did a full reboot.) Presumably it can be done with lvm too.
BTW Dimitri, what about BTRFS instead of ZFS?! :-P
Even better: I'm on centos, redhat's going to go btrfs standard soon. I said zfs b/c zfs support is already there, all they need to do is test it w/ openzfs on linux.
Dima _______________________________________________ Mailing list: http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack Post to : [email protected] Unsubscribe : http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack
