Windows 2008 - TSM cluster client - no documentation!?!?

Bartacus

ADSM.ORG Member
Joined
Mar 14, 2005
Messages
34
Reaction score
0
Points
0
Website
Visit site
So we are starting to implement Window 2008 server clusters (the lovely Vista stuff). The cluster administrator has completely changed, rendering the old documentation for installing a clustered TSM client useless.

Problem is, I can't find ANY documentation on installing the 5.5.1.16 client in this new cluster administrator. From what I'm gathering, you now have to dedicate a clustered IP for TSM (which we never had to do before). The install wants its own IP and stops there. We've tried giving it the clusters VIP, but it won't take it. It wants its own, the greedy thing.

Anyone else run into this?

Bart
 
Hi Bartacus,

that sounds totally false to me.... I've installed TSM (better said: the TSM ressources) in a 2-Node-Cluster (Windows Server 2008 Enterprise) last week.

There is absolutely no need to assign a IP for TSM !!

All you have to do is installing the appropriate TSM-service (dsmcutil install scheduler.......). When finished this part you can do:

1. start Failover Cluster Management
2. click on the "+" on the left side of your Clustername (ex.: FILECLUST.example.com)
3. click on the "+" on the left side of "Services and Applications"
4. select the appr. Clustergroup

In the Middle-Window you see: Server Name, File Server and Disk Drives (maybe even Other Resources).

5. In the right part of this window click on "Add a resource"
6. Select "Generic Service"
7. A new Window pops up ... select the TSM Service created at the beginning
8. Click "Next" ... click "Next" ... click "Finish"

That's it !!!!! Much more easier than "before Windows 2008" :)

All you have to do at next is to bring this resource online (right-click on the resource and select this option).

Hope this helps
 
Thanks Obiwan2, but in the meantime we just forced IBM to provide proper documentation, something those goof-offs should have done MONTHS ago. :) For the money we spend on IBM garbage like TSM, you'd think documentation would be a given, but nooooo.

Bart
 
From what I'm gathering, you now have to dedicate a clustered IP for TSM (which we never had to do before). The install wants its own IP and stops there. We've tried giving it the clusters VIP, but it won't take it. It wants its own, the greedy thing.

Anyone else run into this?

Bart

I don't understand the issue with the clustered IP. In order to setup TSM on a clustered environment it must use the cluster's virtual IP. This is mandatory for the cluster to continue backing up shared resources in the case of a failover. Thus, TSM in the Clustered environment has a unique IP that is also the Cluster's IP Address.
 
Last edited:
I don't understand the issue with the clustered IP. In order to setup TSM on a clustered environment it must use the cluster's virtual IP. This is mandatory for the cluster to continue backing up shared resources in the case of a failover. Thus, TSM in a way TSM in the Clustered environment has a unique IP that is also the Cluster's IP Address.

Our only issue was you didn't need to do this before. That and the fact that IBM never bothered to provide new docs for the 2008 cluster install. We spend TONS of money with IBM, and a company this size should NEVER drop the ball like that.

Bart
 
How about the TSM Client Acceptor service for webclient restores, and what are the dependencies?
 
The best would be to install
- TSM CAD (Client Acceptor Daemon)
- TSM Scheduler

as Cluster-Ressources.

For using the Webclient you need to install the TSM RCA (Remote Client Agent) as a normal service ... NOT within the cluster. Additionally there should be the line "MANAGEDSERVICES webclient" in the OPT-File. That means that the TSM CAD will manage the TSM RCA. In our case the dependency of this service is the approp. harddisk (again a Cluster-Ressource), because it makes no sense to start this service without the underlying physical ressource.
 
DId you guy's create a seperate CAD and Schedule to manage each cluster resource. How do you handle your resources failing between nodes and still backing up?


We are having trouble when we failover to the other node the schedule doesn't re-start itself any one have any examples the instructions we found do not seem to help.

Thanks
 
Hi tsmhelum,

you need to configute the TSM ressources on both cluster-nodes !! Also don't forget to set the TSM Registry-Key on both nodes !

Sometimes it can happen that one of the TSM-Ressources will not start after a failover (I had this also a few times)... but not that often. You need to go sure that the password (stored in the registry) is set correctly. Maybe you have to delete the TSM-Services manually and recreate them. Before that you can delete the Registry-Key (HKEY_LOCAL_MACHINE/SOFTWARE/IBM/ADSM/CurrentVersion/Nodes/[YourNodeName] and re-register with the TSM-Server to store a new password in the registry. If the TSM-Ressources are then running again - repeat a failover to the other node. If this works - failover again.

If still no success - delete the TSM-Services, delete the registry-key again, recreate the TSM-Services, re-register with the TSM-Server. Start the whole failing-over procedure again to see if it now works.

I think i should have a document about the cluster-configuration and installation of TSM. If you need it .... drop me a line !
 
Back
Top