Tuesday, October 13, 2015

How to Start and Stop the Datastage Engine

Here are the instructions to Stop Information Server 8.x, verify all the processes are stopped and then restart Information Server (see instructions below for each step).

We will stop in this order:

  •       Stop DataStage
  •       Stop Node Agents
  •       Stop WebSphere

1.     Log into your Unix/Linux machine as root (you can run first part as dsadm). 
2.     Make sure that no one is using DataStage. 
3.     Check if there are any processes left by executing the following commands:

  • ps -ef | grep phantom - This would show processes from jobs
  • ps -ef | grep osh - This would show processes for parallel jobs
  • ps -ef | grep dsapi - This would show processes for client connection
  • ps -ef | grep dscs - This would show processes for client connection

These commands shouldn't return any results. If they return results, it means that there are still some processes. You need to wait for the users to stop their processes or kill the processes (using kill -9 PID) if you confirm that these are not longer valid connections .

4.     Check if DataStage RPC Daemon (dsrpc) is running by executing:

  • netstat -a | grep dsrpc

If dsrpc is running then you should see one line with the status LISTEN. If you get more lines it means that there are still some connections and you need to wait. If you see that dsrpc or other connections are in status CLOSE_WAIT then you also need to wait until these disappear (you can also reboot the machine)

5.     Stop DataStage (you have to run this from $DSHOME)

  • . ./dsenv
  • bin/uv -admin -stop

6.     Check to see if there are any shared memory segments left:

  • ipcs -m | grep ade

Note: If you are not running with the default itag of "ade" use that instead of "ade". For example if you have two DataStage instances one would be "ade" and the second would have another tag, say "801" or "efg".
If you get a result, take note of the ID and remove it (you will need root access for this part)
ipcrm -m ID

7.     Then check dsrpc again using


  • netstat -a | grep dsrpc

You shouldn't get any results. If you get any and/or the dsrpc is in CLOSE_WAIT then you need to wait until this it disappears (you can also reboot the machine)

8.     Stop the ASB Agent (you need to be root here)


  • Go to <InstallPath>/IBM/InformationServer/ASBNode/bin
  • Run ./NodeAgents.sh stop

Type 'yes' if you receive the following message: rm: remove write-protected regular file Agent.pid'?

9.     Check the agents have stopped:

  • ps -ef | grep java

To see full command in Solaris:

  • /usr/ucb/ps -auxww | grep java


If still see the Logging Agent or ASB Agent then you need to kill these. e.g. kill -9 PID

10.  Stop the WebSphere Server (you need to be root here)

  • Go to <InstallPath>/IBM/WebSphere/AppServer/bin
  • Run ./stopServer.sh server1

* if that does not work, try ./bin/stopServer.sh server1 -username wasadmin -password wasadmin
Now check that the WebSphere java processes is gone:

  • ps -ef | grep java

To see full command in Solaris:

  • /usr/ucb/ps -auxww | grep java

remove it if it exists with: kill PID (no -9 in the kill command)
Now, you can restart, in this order:


  •  Start WebSphere Server
  •   Start ASB Agent
  •  Start DataStage

1.     Start WebSphere Server as root

  • Go to <InstallPath>/IBM/WebSphere/AppServer/bin
  • Run ./startServer.sh server1

Wait until you get back a line saying: Server server1 open for e-business;

2.     Start ASB Agent as root

  • Go to <InstallPath>/IBM/InformationServer/ASBNode/bin
  • Run ./NodeAgents.sh start

3.     Start DataStage (as dsadm)


  • Go to $DSHOME
  • Run bin/uv -admin -start

3 comments:

tMap vs tJoin -Talend

  tMap is frequently used component for joins and lookup purpose, it is also use for verity of operations and transformations, whereas tJoin...