[Ocfs2-users] crash of node0 takes down node1 as well

[Michael Steinmann]

We are using ocfs2 1.1.7-SLES with a 2.6.5-7.236-smp x86_64 kernel on SLES9.
ocfs2-tools are 1.1.4.

OCFS2 works very smooth except for a major problem: a crash of node0 takes
down node1 as well.
Crash of node1 in turn leaves node0 running.

This happens even if node0 gets rebooted (init 6) when a shell sits on one
of the ocfs2 mountpoints and the shared mountpoint cannot be unmounted
cleanly.

We are using the /etc/init.d/ocfs2 initscript posted here
http://lists.suse.com/archive/suse-oracle/2005-Sep/0118.html. Standard
SLES script for mounting/unmounting OCFS2 volumes is missing from
ocfs2-tools-1.1.4-0.4.x86_64.rpm.

- what is the standard way of mounting/unmounting OCFS2 volumes?
- can fencing of node1 be avoided when node0 crashes?

--mike