One possible course
of action in this situation is to give each member a vote and
configure a quorum disk with 1 vote.
Expected votes is 4, and
quorum is 3.
If the quorum disk fails, remove the quorum disk
vote by using the
clu_quorum -d adjust 0
command.
Or, if a
member fails, remove the member vote by using the
clu_quorum -m
failed-cluster-member
0
command.
The resulting cluster (expected votes is 3; quorum is 2)
would survive another failure.