Page 1 of 2

Error in restarting the DS Engine

Posted: Tue Oct 07, 2008 9:14 pm
by neeraj
Hello friends,

I need your help.

When I tried to start the DS engine, it is not coming up.

When I fired ps -ef |grep ds command, I got the below mentioned details.

dsadm 241834 3113022 0 05:02:43 pts/0 0:00 ps -ef
root 266408 94398 0 Sep 30 - 0:18 /usr/sbin/rsct/bin/rmcd -a IBM.LPCommands -r
dsadm 798812 1388748 0 15:14:30 - 0:00 sed -e 1,/StackTrace/d
dsadm 1376318 1716354 2 0:00 <defunct>
dsadm 1388748 1 0 15:14:30 - 0:00 head -10
dsadm 2682988 1388748 0 0:00 <defunct>
dsadm 2810038 1388748 0 15:14:30 - 0:00 dbx -t 32000 -d 1000 -a 1376318
dsadm 3113022 2662428 0 05:00:49 pts/0 0:00 -ksh

This type of details I get very regularly . I am not able to understand what are these processes and why these processes are not allowing me to restart my DS enigne and how can I get rid of it and restart the DS engine.

Regards
Neeraj Mahajan

Posted: Tue Oct 07, 2008 9:30 pm
by meet_deb85
Hi Neeraj,

Please let us know the step by step processes which you are following to
restart the server.

Posted: Tue Oct 07, 2008 10:09 pm
by neeraj
Hello,

I fired the below mentioned commands.

1. uv -admin -stop

2. uv -admin -start

Regards
Neeraj Mahajan

Posted: Tue Oct 07, 2008 10:33 pm
by chulett
You can't just simply stop and immediately restart it, you need to check between those two steps and wait for it to come 'all the way down' and release all ports in use. Failure to do so can mean the engine comes up but the dsrpcd deamon cannot start and therefor you cannot connect to any of the projects.

Use netstat until all associated ports are released. Or try restarting dsrpcd in debug mode so it can tell you what's wrong. For an example, check this posting:

viewtopic.php?t=95994

Another sure-fire way to clear the hung ports is to bounce the server DataStage is running on. Unless you're on a Tru64 cluster, that is - much less sure-fire there. :wink:

Posted: Tue Oct 07, 2008 11:55 pm
by neeraj
Hello,

I fired below mentioned queries.
nohup dsrpcd -d9 > dsrpcd.log 2>&1 &
The result is:-

RPCPID=1847514 - 07:41:14 - uvrpc_debugflag=9 (Debugging level)
RPCPID=1847514 - 07:41:14 - In rpc_init()
RPCPID=1847514 - 07:41:15 - bind bombed errno=67
RPCPID=1847514 - 07:41:15 - listen failed

The I executed
netstat| grep dsr
The result is
tcp4 0 0 loopback.dsrpc loopback.35251 CLOSE_WAIT


I know the easiest solution is to restart the Unix box and this problem may get resovled. But Please let me know is there any alternative way to get rid of it.

Regards
Neeraj Mahajan

Posted: Tue Oct 07, 2008 11:57 pm
by chulett
Talk to your SysAdmin about clearing ports like that, see what tools are available for your O/S. They typically take 'root' access to use, so they would more than likely need to help in the process.

Posted: Wed Oct 08, 2008 12:09 am
by neeraj
Hello,

Thanks for the quick reply.

Can you please let me know the process/steps of doing it? I will login using root password and do the needful as suggested by you.

Regards
Neeraj

Posted: Wed Oct 08, 2008 12:17 am
by ray.wurlod
Have you undertake a Search of DSXchange to find the steps? I'm sure they've been documented here somewhere in the past.

Posted: Wed Oct 08, 2008 12:25 am
by neeraj
Hello,

I tried a lot. Might be I am not able to search it properly...

Please help me.

Regards
Neeraj

Posted: Wed Oct 08, 2008 7:28 am
by chulett
:? Are you a UNIX Admin? Root access is not a sword to be wielded lightly and I'm leery of giving you instructions that could potentially be lethal to your server. Never mind the fact that we don't know your O/S nor what software is installed there.

Find a SysAdmin and talk to them.

Posted: Wed Oct 08, 2008 7:36 am
by neeraj
Hello,

I really like it.

No. I am not working as a Unix admin. But being the lead person for datastage, I have got the access to root to look into datastage admin works. I can help my unix administrator in resolving this issue if you could help me. He is not very much clear about the process.

Regards
Neeraj

Posted: Thu Oct 23, 2008 10:22 am
by mmanes
If it's not coming up after your start you can do the following:

(using root user)

#uv -admin -stop

make sure that dsrpcd and monitor are not running

#kill -2 (or -9 if needed) all datastage processes
#$DSHOME/./scripts/DSEClearSockets.sh
#uv -admin -start

bye

Posted: Thu Oct 23, 2008 10:23 am
by mmanes
If it's not coming up after your start you can do the following:

(using root user)

#uv -admin -stop

make sure that dsrpcd and monitor are not running

#kill -2 (or -9 if needed) all datastage processes
#$DSHOME/scripts/DSEClearSockets.sh
#uv -admin -start

bye

Posted: Thu Oct 23, 2008 10:50 am
by ArndW
Except the "DSEClearSockets.sh" script is one that doesn't belong to the product but is user-written.

Posted: Thu Oct 23, 2008 10:54 am
by mmanes
written by Ascential (now IBM)