160 Troubleshooting
Troubleshooting I/O fencing
Table C-1
I/O fencing scenarios
Event
Node A activity
The disk array
Operates as long as all systems
containing two of
stay in the VCS One cluster.
the three
coordinator disks
is powered off.
Races for a majority of
Node B leaves the
coordinator disks. It fails
VCS One cluster
because only one of three
and the disk array
coordinator disks is available.
is still powered
Removes itself from the VCS
off.
One cluster.
Node B activity
Operates as long as all systems
stay in the VCS One cluster.
Leaves the VCS One cluster.
Operator action
Power on the failed disk
array and restart the I/O
fencing driver. This lets
Node A register with all
coordinator disks.