Steps to halt TSM 7.1 server on AIX cluster

[jqei]

Hi,

I need to stop a TSM 7.1 server that's running on AIX cluster. It's been ages since since I've dealt with TSM and that was version 5 on a Windows server.

Should all sessions just be cancelled and and then disabled? Then perform a "halt"? Can I somehow see when the TSM server is stopped? On Windows I just check the TSM server service. Can I look for some certain processes on AIX?

Also since this is a cluster setup? Can TSM just be failed over to passive node? Maybe a stupid question as I guess that's the whole idea about a cluster? But can failover just be done online and I perhaps don't need to halt TSM server first?

Many thanks!

 

[marclant]

Should all sessions just be cancelled and and then disabled? Then perform a "halt"?

You don't really need to cancel the sessions before, issuing the HALT command will cancel the sessions and processes before stopping the server.

n I somehow see when the TSM server is stopped?

Issue ps -ef | grep dsmserv to see if the process is still running and ps -ef | grep db2sysc to make sure DB2 also went down after dsmserv.

Also since this is a cluster setup? Can TSM just be failed over to passive node? Maybe a stupid question as I guess that's the whole idea about a cluster? But can failover just be done online and I perhaps don't need to halt TSM server first?

Not familiar enough with HA, I assume that yes you can failover, like you said, that's the whole point.

 

[RecoveryOne]

Don't have TSM in a HACMP cluster, but I a few other apps on AIX. Generally if you halt the app with the HALT command, hacmp will see the application terminate.

Depending on how your HA is configured you may need to unmanage resource groups if you do not wish to have automatic fail over.
Also, depending on how your scripts are setup for HACMP, you could move resource groups. That (if your scripts are setup correctly) should automatically halt dsmserv and db2, move the luns/other tasks, and then start up on the secondary node.

Please note that I've not read the docs on TSM and HACMP, so I could be way off in how TSM operates in a HA env.

 

[jqei]

Ok, thanks for all the feedback. Seems I'm good to go with all just a regular HALT and then check the TSM process afterwards. Thanks again. Much appreciated.

 

[RecoveryOne]

How versed are you with your HA Cluster? It should be fairly easy to find the start and stop scripts.

 

[jqei]

I'm a newbie. Just came from the Windows world and trying to find my way around. I'll try and reach out to our Unix admin team. They can probably help. Thanks.

 

[jqei]

Ok, so I was provided with the start and stop scripts. This is the stop:

# Set seconds to sleep.
secs=5
#
INSTANCE_DIR=/home/tsminst1/tsminst1
# TSM lock file
LOCKFILE="$INSTANCE_DIR/dsmserv.v6lock"
echo "Stopping the TSM server now..."
# Check to see if the dsmserv.v6lock file exists. If not then the server is not running
if [[ -f $LOCKFILE ]]; then
read PROGNAME J2 J3 PID REST < $LOCKFILE
kill -s PRE $PID
echo "Waiting for TSM server running on pid $PID to stop..."
# Make sure all of the threads have ended
while [[ `ps -m -o THREAD -p $PID | grep $PROGNAME | grep -c -w $PID` > 0 ]]; do
sleep $secs
done
fi
exit 0

I'm not sure what it does but it seems to terminate TSM in forced way? Not shutting down "properly" with issuing a HALT?

 

[marclant]

I'm not sure what it does but it seems to terminate TSM in forced way?

No, it's not doing a force kill. It sends a signal to the process to shut down gracefully. kill -s PRE is the equivalent to kill -36 which is one of the recommended ways to stop the server: https://www.ibm.com/support/knowledgecenter/SSEQVQ_8.1.9/srv.install/t_srv_stop-aix.html

 

[jqei]

Ok, so if I understand the kb correct it's best to run halt and if I can't connect to TSM then I can use the stop script? Thanks.