Scheduled backup failures for win2k3

[jimperrin]

We are running TSM 5.5 on a linux server which runs just peachy. However we're seeing that the scheduled backups aren't running properly on windows systems via the TSM sheduler/client acceptor services. We've tried running these services as both local system accounts, and as Active directory users with admin rights to the system, however both methods fail. The error messages from the logs just say that the scheduled backup was missed and it's falling back to the next scheduled time. Anyone else seen this, or have suggestions for how to fix it? If we have an admin user logged in, everything runs fine. however I don't like leaving an admin account up on our servers.

 

[moon-buddy]

Any network ACLs that may hamper access?

How is you dsm.opt configured? Can you post a sample?

When you invoke dsmc, are you prompted for a password? have you reset the node's password? What does q sched give you at the dsmc prompt?

 

[jimperrin]

Thanks for following up!

dsmc works fine without password prompts (it's stored properly in the registry)

My dsm.opt is pretty basic, copied below:

ClusterNode Yes
NodeName MynodeNameHere
Domain R: Q:
CommMethod TcpIp
EnableLanFree No
GuiTreeViewAfterBackup Yes
LargeCommBuffers Yes
PasswordAccess Generate
SchedLogRetention 7 D
SchedLOgName R:\TSM\dsmsched.log
ErrorLogName R:\TSM\dsmerror.log
ErrorLogRetention 7 D
SchedMode Prompted
TcpBuffSize 512
TcpNoDelay Yes
TcpPort 1500
HTTPPort 1582
TcpServerAddress xxx.yyy.xy.xy
TcpWindowSize 63

I don't see anything staring at me as glaringly obvious here.

 

[moon-buddy]

Oh! You are running on a Cluster. Have you setup the Cluster environment using the MS Cluster manager utility?

Do you have separate TSM Schedulers for the node and the Cluster?

 

[jimperrin]

Yes, I'm sorry I should have stated that I'm running this in a clustered environment.

Yes. the cluster was set up using MS Cluster manager. I have tsm schedulers for the individual nodes, and then a scheduler running as a service in the cluster, which is dependent on the data in the cluster. The only issue that I had (per the documentation) was during the clustered service install, no matter what registry key I choose to add (per the setup documentation) I'm told it's an invalid key. The path is identical to what's listed in the documentation (substituting for my node name) but each time I'm told it's invalid.

 

[moon-buddy]

I know that there is a wrong entry on the Cluster setup instruction (a typo most probably) for the Registry entry. Can you post the link to the source document you are using, or if you can, post the exact entry you are using?

 

[jimperrin]

I'm following the guide at -> http://publibfp.boulder.ibm.com/epubs/pdf/c3201461.pdf, for both the local and MSCS setup.

 

[sgabriel62]

Jim
Make sure your TSM service dependency is set to Manual. Also although your paths may be correct in the registry - do you have the proper keys installed with the correct passwords. What I do - is after I set up my schedulers - I run the dsmcutil update pw /node: /password: /optfile ....options to update the registry. I do this on both active and passive nodes together in the event of a failover at any one point in time.
You can also get rid of the TCP sizing stuff - use the defaults assigned by the OS.

Hope this helps

 

[jimperrin]

Thanks for the advice. I'll give this a shot and remove the TCP bits. Thanks for the pointers. I appreciate your help!

 

[moon-buddy]

If Steven's suggestion does not work, take a second look at the registry entry:

From the guide (page 613 step 6), Take a second look at the registry entry: SOFTWARE\IBM\ADSM\CurrentVersion\BackupClient\Nodes\mscs-cluster-group-a\tsmsv1

I believe there is a typo here in which you should not include SOFTWARE\IBM\ADSM\CurrentVersion\ when entering the registry entry. (If this is not it, try a hit-and-miss - eliminate parts of SOFTWARE\IBM\ADSM\CurrentVersion: I can't exactly recall, and I don't have access to a server to compare with right now)