I've got a customer who's got two MS SQL clusters, each with two
nodes. They're running the Windows client 4.1.3, with a TSM server on
AIX running 4.1.3.

We've got separate TSM clients on each of the two nodes. The active
node is set to back up all local drives and files, save the SQL files,
which they're using TDP to back up.

The rub is that until 1/31/02, the local drive list for both clusters
consisted of drives C:, E:, G:, H:, and Q:. C: and E: are the only
true local drives; the others are drives that are controlled by the
cluster. So, until 1/31, C:, E:, G:, H:, and Q: were all backed up,
with no special domain instructions.

On 1/31, something happened. Now, the TSM client will only list C: and
E: as local drives, and they became the only drives being backed up.
The spooky thing is that this happened on two separate clusters at the
same time. (I fixed the problem by specifying domains for each
cluster.)

No one will confess to having changed anything, either on the OS, the
cluster administration, or TSM. The only significant events concerning
TSM on 1/30 was a server downage, due primarily to too many clients
all being backed up at the same time and generating a 'server mount
not available' type of message, and the line

        schedlogretention 30

being added to the client option file for each node of the two
clusters.

Has anyone run across this?

--
Mark Stapleton (stapleton AT berbee DOT com)
Mark Stapleton (stapleton AT berbee DOT com)