Adding relevant folks.
Sahina?
Hi,
I need some direction to make sure we won't make more mistake in recovering a 3-node self hosted engine with Gluster.
Someone carelessly, very, wiped out the first few MB of the LVM PV on one node. We changed the quorum and brought up the Gluster and oVirt. We ran vgcfgrestore on the 'bad node' and we can run pvscan, vgscan and lvscan to see the LV's.
If you have 2/3 bricks online - storage would have been available. Can you elaborate why you needed to change qourum
What should I do next to prevent more damage or corrupt the glusterfs on the other two nodes?
Will this work?
mkfs the LV's
bring up gluster
run gluster sync <goodNode>?
On the node where the bricks were wiped out,you need to
1. rebuild the bricks (i.e recreate the LVs and mount at the same brick path as before)
2. Restart glusterd (if not started)
3. Reset the brick - from Volume -> Bricks subtab, select the relevant brick