[DRBD-user] Master-Master GFS2 fencing question

Pieter Steyn pieterste at gmail.com
Tue Sep 29 08:25:52 CEST 2009

Note: "permalinks" may not be as permanent as we would like,
direct links of old sources may well be a few messages off.

Hi guys,

I'm running a Master-Master configuration on a GFS2 partition.

Could someone please give me some advice regarding fencing?  I see
that if I power down one of my Master's then GFS2 starts logging

Sep 28 19:25:41 server2 fenced[2222]: fencing node "server2"
Sep 28 19:25:41 server2 fenced[2222]: fence "server2" failed

Once that happens, I am unable to access the partition on server1 (the
remaining one), until I run:

fence_ack_manual -n server2

Is there a way to automate this?  I know it's very risky when server2
isn't really dead.  Does DRBD have it's own fencing over and above
GFS2 in which case I can discard GFS2's fencing?

I'd like to do load balancing with automatic failover in case one of
the masters die, but the fencing is the only thing still in the way of

Pieter Steyn

More information about the drbd-user mailing list