Well, I'm stumped.
My customer has a 2-node Exchange 2003 active/passive cluster on Win2K3 R2 SP2.
TDP for Exchange is 6.1.3.3, Legacy backups.
Node 1 is usually the active node, and TSM works fine.
The TDPEX GUI on Node 1 works as expected. The client scheduler service for
Exchange is installed on node 1, is defined as a cluster resource, and works
fine.
Node 2 is usually the passive node, and TSM doesn't work at all.
The dsm.opt file, the information store, & the logs are on a cluster drive, I:.
When we fail Exchange over from Node 1 to Node 2, the I: drive becomes
available to Node 2, as it should, so the exchange dsm.opt file is used on both
nodes. CLUSTERNODE is set to YES, PASSWORDACCESS to GENERATE..
When we then start the TDP for Exchange GUI on Node 2, it creates a session
with the TSM server and authenticates OK (verified by checking the server
actlog). When you click restore, you can see the available backups to restore
as you would expect.
But when you click backup, you get the usual message "updating mailbox
history", then the GUI dies without a whimper. No message, the GUI just
vanishes.
If you invoke a backup via the command line from the \TDPEXCh directory::
tdpexcc backup * \tsmoptfile=dsm.opt \excserver=exch-foo \logfile=foo.log
EXACTLY the same thing happens. You see the session established with the
server, no password errors. You see the "updating mailbox history" message.
Then the cmd line exits. The exchange log foo.log shows the "updating mailbox
history" message, nothing else. No messages in dsierror.log, nothing in the
Windows event log. The server actlog shows only ANR0480W, indicating the
session was terminated by the client end. It's a mystery how it dies.
The box has been rebooted, and we've tried uninstalling and reinstalling the
TDP.
I'm not having any luck Googling for an error message I don't see - anybody
seen behavior before that makes the TDPEXch session vanish without a tra - um,
error? Suggestions for search words?
What am I missing here?
Thanks for any ideas!
Wanda
|
