[Gluster-users] NIC died migration timetable moved up

[David Gossage]

So a NIC on one of my nodes died today.  Chance are good that  a reboot
would bring it back to life, but then I'd be down while 2-3TB of unsharded
VM's healed which wouldn't be fun.  I figured I'll run with 2 nodes while I
get sharding enabled and then bring in the 3rd new node I was going to
replace that one with anyway.

1) I am using oVirt3.6 with CentOS 7 and while I am going to confirm I
believe it does its communications over fuse mount still.  So I am thinking
it would help for me to move from 3.7.11 to 3.7.12 as the ligpfapi issues
shouldn't hit me.

2) enable sharding.  Do I need to completely move the VM(powered off) image
off the mount then back on for it to shad or can I rename VM images on the
fuse mount?

Typical disk dir has 3 files.  I'm thinking since only the large image will
shard that is only one I would need to move as others wouldn't shard?
-rw-rw----. 1 vdsm kvm  25G Apr 15 14:21
e7818fd2-7e2e-46e8-92b5-bc036850d88b
-rw-rw----. 1 vdsm kvm 1.0M Dec  2  2015
e7818fd2-7e2e-46e8-92b5-bc036850d88b.lease
-rw-r--r--. 1 vdsm kvm  320 Dec  2  2015
e7818fd2-7e2e-46e8-92b5-bc036850d88b.meta

3) Since I don't plan to re-enable the server with NIC issues can I just
rsync /var/lib/glusterd then give new server same IP other one used to peer
with?  Do I need to change UUID of new server?  Can I manually update the
info in /var/lib/glusterd/vols/*/info so options match after enabling
shards?

Any glaring gotchas I am overlooking?

*David Gossage*
*Carousel Checks Inc. | System Administrator*
*Office* 708.613.2284
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.gluster.org/pipermail/gluster-users/attachments/20160707/cddd8b94/attachment.html>