Dear List,

it seems, it is solved:

On 12/12/2012 03:54 PM, Thomas Denier wrote:
...
> Are you using the style of script that does the following?
>
> * Store the process ID in a specific file after a successful
>   start.
> * Delete the ID file after stopping dsmc.
> * Refuse a request to start dsmc if the ID file exists and
>   contains the ID of a running dsmc process.
>
> If so, I suspect that the scripts are trying to use the same
> ID file for dmsc processes associated with different stanzas.
Yes, but we didn't know... The init-script uses the system tool
'start-stop-daemon' (part of Ubuntu?) to deamonize the dsmc processes.
Obviously the runtime behavior differs between the two boxes. Additional
options for the start-stop-daemon solved it. The error message
"/usr/bin/dsmc already running." is issued by the start-stop-daemon, not
by dsmc.

On 12/12/2012 05:51 PM, Alex Paschal wrote:
> Do they run if you start them manually?
> How about if you remove 6.2.4.4 and backlevel to 6.2.0.0
> to match the working box?
Thanks, this was the right hint. If we start the dsmc processes manually
everything works fine. So the different client versions aren't the
cause, see above (start-stop-daemon).

On 12/13/2012 09:21 AM, Maurice van 't Loo wrote:
> Multiple stanza's are quite normal to use, think about TDP's.
> Did you tried a different client port number for each scheduler?
We don't use different port numbers, should we?

Thank you very much.
I wish you a merry Christmas and a happy new year.

Best regards,

Ethan

> 2012/12/12 Ethan Günther <ethan.guenther AT rz.uni-augsburg DOT de>
> 
>> Dear List,
>>
>> for a client with large filespaces, we would like to split the backup up
>> into several nodes in order to balance the load. On the basis of the
>> TSM-client documentation we created several stanzas in the dsm.sys file
>> with different servername entries. In the init-script several dsmc
>> processes are started with the -servername option referring to those
>> stanzas. Now everything works fine with a 6.2.0.0 client (all processes
>> are started and are backing up as expected).
>>
>> On a very similar box with client version 6.2.4.4 something goes wrong:
>> Some dsmc processes are started successfully, some are not, the numbers
>> vary. So there could be some runtime condition. The error message
>> (printed on stdout) is: "/usr/bin/dsmc already running."
>>
>> Has anybody observed this issue and knows about a solution?
>>
>> Best regards,
>>
>> Ethan
>>