From: Roger Tsang <perj8@ho...> - 2004-12-23 00:59:29
If I reboot the initnode and forcefully power off the failover initnode
while it is in the process of failing over _and_ while the booting node is
in Running pre-root cluster initialization, the booting node can get stuck
in Running pre-root cluster initialization forever.
So even after I power up the failover initnode, there happens to be
split-brain and both potential initnodes are waiting for the other to go up
and both get stuck in Running pre-root cluster initialization. Apparently
they will wait forever in running pre-root cluster initialization until I
power cycle (hard reboot) all the nodes that are stuck in that state.