Problem in stopping Datastage server

[JoshGeorge]

Definitely not the right way of restarting datastage. CLOSE_WAIT states are waiting to close. You could wait a while and let them die by themselves / login into the uv account as dsadm, unlock all and then stop datastage instead of rebooting the operating systems.

[adasgupta123]

Hi ,
Thanks for your reply.
You have suggested that we can unlock all CLOSE_WAIT states logging
as dsadm.Does it mean releasing all the tcp sockets in CLOSE_WAIT state?


Regards
Avik

Definitely not the right way of restarting datastage. CLOSE_WAIT states are waiting to close. You could wait a while and let them die by themselves / login into the uv account as dsadm, unlock all and then stop datastage instead of rebooting the operating systems.

[asorrell]

Actually - the UNLOCK was referring to releasing all the internal locks in the DataStage environment. To do so:

1) Go to the DataStage home directory (shortcut - cd `cat /.dshome`) - note the quotes are the ones to the left of the '1' key on a US keyboard.

2) Execute the dsenv script to set environment variables (. dsenv)

3) Enter the DataStage environment (bin/dssh). Prompt will change.

4) Unlock all locks. Note: NEVER do this while the system is being used. (UNLOCK ALL).

5) QUIT out of DataStage Shell

If you wait for 10 - 15 minutes all the DataStage processes will eventually clear from netstat -a | grep dsrpc. At that point you can re-start DataStage.

[flashgordon]

Hi,

I believe I have something important to add to this discussion of netstat
processes blocking Datastage from coming up. There are cases where phantom
processes can keep Datastage from releasing the dsrpc port/socket. In this case
the netstat process below hung for 11+ hours.

$ netstat -a|grep dsrpc
localhost.dsrpc localhost.45671 49152 0 49152 0 CLOSE_WAIT
$

I was reading google discussions of netstat kills and I came across ps -aef. This command apparently shows parent and child process relationships in Unix. I ran it and noticed that there were still Datastage processes running even though the server was down. I had run the standard "ps -elf|grep dsapi" and that showed no running Datastage processes. So when I saw a Datastage process running I killed the phantom process. After I killed that phantom psdstage process, Datastage released the dsrpc port/socket that was keeping Datastage
from coming up immediately. This socket had been hung for 11 hours running. I was then able to start the Datastage server

myserv:/ps1/psdstage/Ascential/DataStage/DSEngine: ps -aef|grep dstage
psdstage 12368 12263 0 12:02:34 pts/2 0:00 -ksh
psdstage 13502 12368 0 12:04:12 pts/2 0:00 ps -aef
psdstage 13503 12368 0 12:04:12 pts/2 0:00 grep dstage
psdstage 22076 1 6 Dec 10 console 5459:30 phantom DSD.StageRun NW_OGL_EPM_F_NW_1G_MORGAD_SF. NW_OGL_EPM_F_NW_1G_MORGAD_SF
myserv:/ps1/psdstage/Ascential/DataStage/DSEngine: kill -9 22076
myserv:/ps1/psdstage/Ascential/DataStage/DSEngine: ps -aef|grep dstage
psdstage 12368 12263 0 12:02:34 pts/2 0:00 -ksh
psdstage 14217 12368 0 12:05:19 pts/2 0:00 ps -aef
psdstage 14218 12368 0 12:05:19 pts/2 0:00 grep dstage
myserv:/ps1/psdstage/Ascential/DataStage/DSEngine: netstat -a|grep dsrpc
myserv:/ps1/psdstage/Ascential/DataStage/DSEngine:
!!! a clear netstat -a means you can now bring up Datastage !!!

Releases where this occured:

more .dsrel
7.5.1.35
myserv:/ps1/psdstage/Ascential/DataStage/DSEngine: uname -a
SunOS myserv 5.10 Generic_125100-10 sun4u sparc SUNW,Sun-Fire-15000
myserv:/ps1/psdstage/Ascential/DataStage/DSEngine:


... Flash Gordon

[dannywcw]

Hi,

Referring to the above

Instruction

Step 3) Enter the DataStage environment (bin/dssh). Prompt will change.

what is the command to execute it?

is it ./dssh or ?

Please advice, my datastage has hung for 28hrs and the session with close_wait has not been release. thank you.

[chulett]

$DSHOME/bin/dssh

Get an SA involved if ports are hung and not clearing on their own, they have The Power.