Understand linux process ID related to a specific TSM session number

tivolimistery

ADSM.ORG Member
Joined
Oct 10, 2009
Messages
58
Reaction score
0
Points
0
PREDATAR Control23

Hi All,

is there a way to detect the PID of the linux process (on TSM client machine) related to a specific TSM session number (on TSM server machine)?
Every day, a lot of users execute transparent recall of migrated files and I would know how many TSM sessions have been caused by each user.

Thank you in advance,
Mauro T.
 
PREDATAR Control23

There is no relationship with a system PROCESS ID vs SESSION ID.

However, you can see which users execute restores or backups manually through their USER ID - IF they have unique USER IDS. Also, they can only fire up one process at a time so looking at the TSM server side, you should be able to see the number of sessions from the TSM server side by USER ID.

Curiosity: If your intention is to determine the number of sessions, why are you going after the PID?
 
PREDATAR Control23

Hi Moon-Buddy,

thank you for your support. I'll try to explain better my problem providing you an example.

In our HPC cluster have been defined a lot of user, each one with a unique user id (on login server side!)
From the cluster login node, each user can do a lot of different actions:

- user A recalls data from /path/to/data/1
- user B recalls data from /path/to/data/2 and /path/to/data/3 and /path/to/data/4
- user C recalls data from /path/to/data/5
and so on...

Running "q ses f=d" command on TSM server, I can see only a great number of "running" sessions, but I can't understand the exact session related to a specific action executed by a cluster user. For example, If I want to cancel only actions related to user B I don't know how to do it.

I could try to do it killing linux process related to the specific user, but, often, this interruption caused an incomplete recall of the file (wrong file size).

What is your opinion/suggestion regarding this kind of problem?

Thank you again
 
PREDATAR Control23

Hi Moon-Buddy,

thank you for your support. I'll try to explain better my problem providing you an example.

In our HPC cluster have been defined a lot of user, each one with a unique user id (on login server side!)
From the cluster login node, each user can do a lot of different actions:

- user A recalls data from /path/to/data/1
- user B recalls data from /path/to/data/2 and /path/to/data/3 and /path/to/data/4
- user C recalls data from /path/to/data/5
and so on...

Running "q ses f=d" command on TSM server, I can see only a great number of "running" sessions, but I can't understand the exact session related to a specific action executed by a cluster user. For example, If I want to cancel only actions related to user B I don't know how to do it.

I could try to do it killing linux process related to the specific user, but, often, this interruption caused an incomplete recall of the file (wrong file size).

What is your opinion/suggestion regarding this kind of problem?

Thank you again

If user action is data related, why not just let these finish? After all, the role of a backup environment is to provide restore support when needed and I believe this one of such requirement.

If your issue is that there are so many sessions happening at the same time, why not create a process to queue each session so there order in the chaos? This will now be a procedural issue with a SLA in place.
 
Top