Hi:
we have a remote site which has a two-node drbd cluster. last week
there was a power outage. since we have ups for each server
configured, so the cluster auto shutdown cleanly.
unfortunately when utility power came back, one of the servers
didn't recover. it seems like the attached ups is dead or hang.
we need to start the survived node for service. I thought I just
need to set it as primary. but soon I found although the disk status
is consistent, it need to find peer to make sure if it is UpToDate. at
drbd8 there is command "drbdadm -- --overwrite-data-of-peer". but it
didn't work at drbd9. finally I found the new command "drbdadm
del-peer" and then I can force drbd primary.
I don't know what's the correct procedure under this kind of
situation. I read the drbd manual again and still has no clue. is
"drbdadm del-peer" enough for doing the work? will the secondary node
recover automatically when it comes back?
thanks a lot for help.
