I've been having cluster problems after applying SP6 to a 2-server

every (other) night, one of the servers abended eating a poison pill
during backup - and ONLY during backup. The surviving server was in a
somewhat "comatose" state: having no connectivity (no ping, "DISPLAY
SERVERS" = 0) and couldn't load any NLMs (console unresponsive on
loading any module).

One cluster resource was in state "loading", the other one in

The abend only occured during backup of the SYS: volume of one of the
servers, which is mirrored between a local IDE disk and a fibre channel
LUN. All other cluster resources could be backed up ok.

What I've tried already ...:

- load TSAFS /Cluster /NoCachingMode

Which is necessary (/Cluster) for BackupExec to recognize the cluster
service volumes, NoCachingMode significantly speeds up differential
backups. Have been playing around with other parameters to no avail.

- set cluster parameters a little bit more reluctant (20 instead of 8
secs timeout), which led me to recreation of SBD partition for some
strange reason

- changed hardware (from noname PC over HP DL 360 to VMware ESX 3.0.1
[which runs like a charm btw.])

Preliminary solution was to exclude SYS: from the backup, which of
course is not really a solution.

Recently, I saw an updated MM.NLM which talks about issues with RAID 1
and mirrored cluster partitions (which I don't have) - could this help?

If it's of interest for the Novell guys, I can post abend logs and
stack dumps here or mail them to you.

Best regards,