Well, to correct it, you might start the scheduler for each node as a
cluster service that never fails over to the other node. That way, the
service won't start up until that cluster node is up and it will only see
the drives that you have assigned to the node.
Bill Smoldt SSSI
Storage Solutions Specialists, Inc.
www.storsol.com
-----Original Message-----
From: ADSM: Dist Stor Manager [mailto:[EMAIL PROTECTED]]On Behalf Of
Alex Mounayar
Sent: Friday, October 13, 2000 2:03 PM
To: [EMAIL PROTECTED]
Subject: Adsm scheduler on cluster node backing up cluster drives after
reboot
On our nt cluster, the tsm scheduler running on the individual nodes are
backing up cluster disks
as well as local disks. We have just setup a nt 4.0 sp 5 cluster. After
installing tsm to each local
node, a tsm scheduler was setup on both nodes for each cluster group.
Backup was running just fine,
the scheduler on each node was backing up the local drives (c d and e)
while the schedulers running
in each cluster group were backing up the cluster drives. Then the
cluster was rebooted, after the
reboot the schedulers running on each node started backing up both local
and what ever cluster drives
happened to be active on that node. There was no domain defined for the
local drive opt file.
Investigation revealed that TSM scheduler service for the local drives was
coming online before the
cluster service completed and all drives including the shared cluster
drives appeared local to the scheduler
and were backed up. After stopping and starting the scheduler only the
local drives are backed up.
Our cluster disks are large so we don't want to back them up twice every
day.
My options around this are manually start the scheduler or add domain
statement to the opt file.
I have added a domain line to the opt file specifying the local drives but
I don't think this should be necessary
and is an additional thing to deal with if any local drives are added.
Has anyone seen this behavior or have any ideas how to correct it?
I am a little concerned that there may be another problem with the cluster
service taking long to complete.
thanks,
Alex Mounayar
Shared Medical Systems
