Tru64 Managers:
We have a 2-node ES40 cluster, running Tru64 5.1 patch 3.
When one node crashes and reboots, BOTH systems run painfully slow until we
reboot both cluster members.
My guess is that the MEMORY CHANNEL gets confused and doesn't come back to
life, and that IO is taking some roundabout ways. (There is no CPU
bottleneck during this slow performance period.)
I don't know how this MEMORY CHANNEL command looks on a "normal" basis, but
right now, it seems like it's not being used by the TruCluster:
# imcs -s
imcs: MEMORY CHANNEL is not initialised for user access
# imcs -f
imcs: MEMORY CHANNEL is not initialised for user access
My questions:
1 - Should 'imcs' provide meaningful information if the TruCluster is using
it correctly? (The 'man' page has some examples.)
2 - Are there tricks to getting the cluster back to full performance after a
member crash (other than rebooting both members)?
Thank you!
Rob Aldridge
AT&T Solutions
Alliance, Ohio
Received on Tue Jun 19 2001 - 16:02:44 NZST