How to restart clients from TSM server

[illllm]

Hi everyone,

Everytime I patch and update the TSM server, I end up having to coordinate with the admins to restart the DSMCAD service on all the clients.

Two questions:

1. Is there a way to restart the client's CAD service from the TSM Server?
2. Is there some kind of fix for this issue? Its not a good thing to always ask the other admins to have to restart the CAD service always.

Thanks,

 

[ILCattivo]

Are you able to tell us what versions of the TSM Server & Client versions are affected here?

Have to say, that I have not come across this issue before when I've initiated patches and upgrades to the server infrastructure? For clients, the TSM Client Acceptor (CAD) controls the scheduler service.
You may want to verify that you have the following option within your dsm.sys/dsm.opt files on the clients

QUERYSCHEDPERIOD # <-- '#' relates to the time in hours the DSMCAD queries the Server for it's next schedule. Make this number smaller, like '2' for example

Alternatively, I have seen people use scheduled tasks within the OS itself which restarts the CAD service approx. 30 minutes before the (server) scheduled start time of that clients backup. [Like a scheduled OS initiated batch file or cron job] This would also prevent the user interaction required to restart any TSM related services, providing you do the patch or upgrade to the server way before the server side schedules are due to start.

 

[illllm]

The server is 8.1.10 and the clients are a mix of 7.x and 8.

 

[illllm]

is there a way to restart the DSMCAD service on the clients from the TSM server ?

 

[okriss]

is there a way to restart the DSMCAD service on the clients from the TSM server ?

PSEXEC?

I think you cannot manage TSM services through TSM admin console or with anything else what belongs to TSM...

 

[ILCattivo]

PSEXEC?

I think you cannot manage TSM services through TSM admin console or with anything else what belongs to TSM...

This is true.. as far as I know you cannot control TSM Client Services directly from the TSM Server or Operations Center.

In order to achieve a centralized ability to manage such things you would require a 3rd party tool like 'TSM Manager' to do it. There would however be specific ports and a remote service to be running on the client for it to work though. But essentially it's what you're looking for.

 

[scr1pt]

Why do you have to restart TSM services after downtime?

Do you have replication where the secondary server is never reachable for the clients (offloading/firewall for example)? If so then you can use the option "USEREPLICATIONFailover No" on the clients.

That is the only reason I can think of that requires client-side restart, because in a replicated environment if the primary server becomes unreachable for whatever reason; the clients will ask the secondary server if the primary server is available again - so if they are not able to reach the secondary server they will be in a loop trying to ask the secondary server if primary is available again. <-- This with "usereplicationfailover yes" which is the default.

See https://www.ibm.com/support/knowled....client.doc/r_opt_usereplicationfailover.html if that is the case.

 

[illllm]

That may be it !! I will check and let you know.

 

[illllm]

So ....a question about that.
The opt file has an automatic entry from TSM "My Replication Server" and a note "Do no modify this "

so if i add USEREPLICATIONSERVER NO to the opt file, i am telling it to no query the secondary server right ?

 

[TaSMania]

TSMCAD is managed by Client Acceptor right? So Client Acceptor should have Startup Type = Automatic. Do you have it set as manual services? When patch or update the server would reboot. Or I misunderstood the question?

 

[scr1pt]

So ....a question about that.
The opt file has an automatic entry from TSM "My Replication Server" and a note "Do no modify this "

so if i add USEREPLICATIONSERVER NO to the opt file, i am telling it to no query the secondary server right ?

(I am assuming you have a typo and you mean the option 'USEREPLICATIONFailover No' - I am not sure about that option you wrote)
Yes, I do think it does not even manage those options anymore then. Also I think the node replication table (NRTABLE) is not being updated client-side - but if it never would work you do not really need it.

 

[illllm]

Yes the entries in the .OPT file on the clients were updated automatically when installed the DR server. I will test this today:

1. USEREPLICATIONFailover No