stop datastage service

[fmartinsferreira]

The correct steps to stop the datastage service are:

1 -> Connect to the server with user dsadm
2 -> ps -ef | grep dsapi
3 -> ps -ef | grep dscs
4 -> with any connection or job running kill by unix
5 -> go to DSHOME
6 -> ./uv -admin -stop

PS:
1 - dsadm is a administrator user, but he doen'st need to have Unix ID 0.

Is this correct?

Regards,

Fernando

[kcbland]

I prefer to use /etc/rc2.d/S99ds.rc STOP and START, but yes, the previous steps to remove all active connections and processes.

[chulett]

From $DSHOME, step #6 would actually be:

6 -> ./bin/uv -admin -stop

And some Operating Systems require dsadm to be in the 'root' group, but most don't. I think Tru64 may be the only one? It would be in the Install Guide.

[fmartinsferreira]

From $DSHOME, step #6 would actually be:
6 -> ./bin/uv -admin -stop

You are right.

But changing my step 5 to "-> go to $DSHOME/bin " then my step 6 works :D .

My Operating Systems is "Unix AIX 5.3 ".

Fernando

[csrazdan]

I prefer to run $ ps -eaf|egrep "dscs|dsapi|phantom|osh|orch" to check if any DataStage process is running.

However, it is also important to check if any client connections are open. You can check it by running $ netstat -aA | grep dsrpc

This command should display only DataStage listner running for you to safely bring down DataStage server.

You may face difficulty in bring up DataStage Server if the ports are left open at the time of bringing it down. Especially on AIX it takes it significant amount of time to realize that the connection for the port is broken and it need clan the port.

Hope it helps........

[fmartinsferreira]

To complete.

The correct steps to start the datastage service are:

start
1 -> Connect with user dsadm
2 -> set DSHOME enviroment variable (DSHOME=/datastage/Ascential/DataStage/DSEngine;export DSHOME)
3 -> go to DSHOME/bin
4 -> ./uv -admin -start

PS:
1 - dsadm is a administrator user, but he doen'st need to have Unix ID 0.

Is this correct?

Regards,

Fernando

[Ultramundane]

Especially on AIX it takes it significant amount of time to realize that the connection for the port is broken and it need clan the port.

You can set tcp_finwait2 to a lower value to take care of that. I set it to 60 which is 30 seconds. The default I believe is 1200, which would be 600 seconds or 10 minutes. Which is a very long time to wait to bring DS back up.

[fmartinsferreira]

Especially on AIX it takes it significant amount of time to realize that the connection for the port is broken and it need clan the port.

And

You can set tcp_finwait2 to a lower value to take care of that. I set it to 60 which is 30 seconds. The default I believe is 1200, which would be 600 seconds or 10 minutes. Which is a very long time to wait to bring DS back up.

Sorry but do you mean that after server re-start I need to wait 10 minutes to start the datastage service?

This if I haven't changed "tcp_finwait2".

Regards,

Fernando Martins

[Ultramundane]

Sorry but do you mean that after server re-start I need to wait 10 minutes to start the datastage service?

This if I haven't changed "tcp_finwait2".

Regards,

Fernando Martins

If you mean the datastage server and not the AIX server. Absolutely. It depends upon what netstat -a | grep dsrpc shows for active connections. If netstat -a | grep dsrpc shows active connections (do not count the listener) when you shutdown the server you will need to wait by default 10 minutes to restart the datastage server. If you don't wait until the connections in finwait2 are cleared, and try to restart, uv -admin -start will say that datastage started up fine. However, you will not be able to connect. You will be able to see the dsrpcd process running, but the listener would have failed to start.

[csrazdan]

Especially on AIX it takes it significant amount of time to realize that the connection for the port is broken and it need clan the port.

And

You can set tcp_finwait2 to a lower value to take care of that. I set it to 60 which is 30 seconds. The default I believe is 1200, which would be 600 seconds or 10 minutes. Which is a very long time to wait to bring DS back up.

Sorry but do you mean that after server re-start I need to wait 10 minutes to start the datastage service?

This if I haven't changed "tcp_finwait2".

Regards,

Fernando Martins

You can use rmsock command to clean the ports before bringing down DataStage server. Following is an example of netstat and rmsock commands:

$ netstat -aA | grep dsrpc

f1000d0001e62b90 tcp4 0 0 *.dsrpc *.* LISTEN
f1000d00019feb90 tcp4 0 0 ServerName.dsrpc phildcut0a.u.ardus FIN_WAIT_2
f1000d0000e98390 tcp4 0 0 ServerName.dsrpc 141.40.55.15.4232 FIN_WAIT_2

$ rmsock f1000d00019feb90 tcpcb
Run this command for each port except the listner.

You will to convinence UNIX administrator to give you access to rmsock command.

Hope this helps....

[fmartinsferreira]

Thanks a lot friends! :D

Best Regards,

Fernando

[Bryceson]

Folks,

I am running into the same problem . . . . as it has been describe above.

Where in DataStage do I need to set the tcp_finwait2??

I can't wait for 10 minutes to bring up DataStage Engine, I would like to set it to 30 seconds.

Thanks in advance

Bryceson

[chulett]

Nowhere in DataStage, that would need to be set by a UNIX Admin.