Tutorial for Install on Windows 2008 R2 Cluster Nodes?

[JustHall]

Is there a clear, detailed tutorial somewhere on how to install the TSM BA Client on a Windows Server 2008 R2 Ent. x64 failover cluster? Everything we've found is either outdated, not for Server 2008, or impossible for the inexperienced to follow. Even an outline of the entire process start-to-finish would be helpful. We tried to follow the TSM w/ Windows 2008 Clusters presentation, but it didn't give much detail and left us where we are now.

We want to backup the 2 nodes themselves (local disks) as well as the shared storage that may live on either node at any given time. As we understand it, the client acceptor and scheduler need to be associated with that shared storage somehow so that whichever node currently hosts the shared storage will also run the TSM backup. Is that even an accurate understanding of the process?

We tried installing the BA client (web client, client acceptor, and scheduler) on each node. Neither of them can see the shared storage drives. We added a Generic Service to the Windows cluster and pointed it at the TSM Client Acceptor and we can connect to the web client through the cluster network name. We didn't add a second service to the cluster for the scheduler, but we think we may need to - but we're not sure how to tie it to the other services so they all live on the same active node. Not sure where to go from here.

Thanks for any insight.

P.S.
The dsm.opt file on each node is virtually empty, it only contains 2 lines. Is that normal? We expected to see the options we configured during the installation process listed in there, but most of them were default so perhaps it omits those?

TSM BA Client 6.1.3
Windows Server 2008 R2 Enterprise 64-bit

 

[JustHall]

Bump. Even an outline of the process would be helpful. Anything at all...

Thanks.

 

[LED888]

Not sure if you found this doc or not.

http://www-01.ibm.com/support/docview.wss?uid=swg21383868

Good Luck,
Sias

 

[hyderchicagotsm]

I have done a active-passive cluster install on Windows 2008 and TSM BA 5.5.
In order to backup your local drives you just need to follow the regular BA Client install procedure. For shared drives, its good to have a separate dsm.opt and TSM scheduler for each cluster group that has a shared drive in it. Basically the only difference between the normal dsm.opt and the dsm.opt for a shared drive is that in the dsm.opt for shared drive you have to specify "CLUSTERNODE=YES". Also, the dsm.opt should preferably be on a shared drive and specify the drives to be backed up using the DOMAIN option. You first do this on the active node and then test your connectivity with the TSM Server and then failover and then test your connectivity from the passive node. This is important to do since if you do not do this and the password expires then in case of a real failover the service wouldnt come up on the passive node.Then you install your scheduler on both the nodes. Finally you install a Generic cluster resource for each cluster group that you have a dsm.opt for. The link which LED888 provided earlier explains this in detail.

 

[JustHall]

I guess we're hung up on a few things, including the "cluster group" vernacular. We've never setup a cluster before this and everyone seems to refer to things in Windows 2003 terminology but 2008 doesn't call similar components the same thing so we get lost easily.

Does the shared drive where the shared dsm.opt file lives need to be its own drive, dedicated to TSM, or is it supposed to live on a shared drive that we intend to backup (in this case, a SQL database data drive)?

Do we need to install the BA client multiple times per node somehow, or simply once per node and the dsm.opt files take care of the rest?

Sorry for our ignorance. Thank you so much for replying!

 

[JustHall]

Bump. Any further insight would be most appreciated.

Thanks.

 

[Project2501]

Not sure if you still care but here is how we do it on 2003R2 clusters.

The dsm.opt file lives on a cluster resource/disk as well as the log files associated with it.

The scheduler service is installed as tsm scheduler (node name) and added as a cluster resource associated with the opt file on shared disk.

You can, if you like, replicate the registry entries across clusters that hold the tsm client password and associated elements.

As started you will need to add the clusternode=yes line to your shared opt file and I also recommend adding clusterdiskonly=no

This should get you up and running.

Okay to sum up a bit of terminology, cluster group is the collection of resources that will fail over from server to server.

Cluster resource is a part of the cluster group.

Finally I really need to set up a 2008 cluster

One other thing, when you set the scheduler up as a cluster resource I recommend setting it to do not restart, it will save you some pain.

 

[JustHall]

Thanks for the reply, and yes we are still very interested.

So, the shared dsm.opt file needs to contain those two things you mentioned - does it need to contain a bunch of other stuff as well? (I'm not the TSM guy, just trying to relay to them what we need.)

Thanks.

 

[Project2501]

the opt file will need to contain more then just those 2 lines but your TSM guys should know what they need to backup what ever it is you are trying to back up.

 

[JustHall]

the opt file will need to contain more then just those 2 lines but your TSM guys should know what they need to backup what ever it is you are trying to back up.

Are there dsm.opt files on each server as well, or just the shared one?

 

[Project2501]

It depends on how you choose to set up your cluster.

I mostly deal with SQL clusters, so I have an opt file for file backups on the cluster disks, I have a second opt file to run the TDLSQL client on the cluster resources and I have an opt file that is attached to the physical box on non shared disk.

The non shared opt file is pointed at local disk and systemstate

the shared opt files are pointed at the disks associated with the cluster group or the SQL instance that is part of the cluster group.

Each opt file will be a little different, node names will be different, domain targets will be different and the exclusions are usually a little different.

I hope this helps.