Job control process (pid xxxx) has failed issue.

DSguru2B · Post by **DSguru2B** » Mon Mar 23, 2009 9:05 am

Hi folks,
I have been having an issue with our nightly batch cycles. The jobs, including all job control batches start failing with only one error and no other explanation. The error is :

Code: Select all

Project:XXXX (myServer)
Job name:PrtPartHistoryUpd
Event #:6253
Timestamp:3/23/2009 8:46:57 AM
Event type:Warning
User:fitsvcs\datastage
Message:
Job control process (pid 3304) has failed

When I reset the job, I get the following in "From previous run..."

Code: Select all

Project:XXXX (myServer)
Job name:PrtPartHistoryUpd
Event #:6255
Timestamp:3/23/2009 8:50:34 AM
Event type:Info
User:FITSVCS\DATASTAGE
Message:
From previous run 
DataStage Job 463 Phantom 3228
>>>>> UVRelVar failed to Disconnect
>>>>> UVRelVar failed to Disconnect

What is UVRelVar?
It seems almost like someone is logging on the server and killing all uvsh.exe threads.
Any help is appreciated.

attu · Post by **attu** » Mon Mar 23, 2009 9:33 am

See if you can get any info in Project's PH directoy.

DSguru2B · Post by **DSguru2B** » Mon Mar 23, 2009 9:45 am

When you reset a job, the message that pops up in the "From previous run..." is actually from the &PH& folder for that run. Nothing more I could find there.

chulett · Post by **chulett** » Mon Mar 23, 2009 9:53 am

Hmmm... something similar was posted 5 years ago but it got no responses, other than that there's no mention of "UVRelVar" here.

DSguru2B · Post by **DSguru2B** » Mon Mar 23, 2009 10:05 am

I know, I tried to search and a lonely thread by 'ogmios' showed up with no replies. He indicated it was harmless but its not harmless for me. Let see, maybe someone here has encountered it before.

priyadarshikunal · Post by **priyadarshikunal** » Mon Mar 23, 2009 11:20 am

We are getting similar problems but in PX 8x. We have a PMR open with IBM but no results yet.

In my case I am not getting any thing from &PH& except the phantom process number.

I think you should check if at that time core file was generated in project folder - which indicates the crash.

Code: Select all

file core

will give the process which was responsible for the core dump.

It might be due to the CPU limit and the priority of the processes generated by your user. (which is probably not the case with the server I am using).

Will post if I get any reply.

DSguru2B · Post by **DSguru2B** » Mon Mar 23, 2009 12:04 pm

What directory inside the project directory will this file be generated? Or is it just the project directory, period?
EDIT:No core file found anywhere in the project directory.

attu · Post by **attu** » Mon Mar 23, 2009 2:15 pm

A core file is an OS file which is generated when applications cannot access memory. It is created in the Projects directory.

DSguru2B · Post by **DSguru2B** » Mon Mar 23, 2009 2:37 pm

Please see reply above.

priyadarshikunal · Post by **priyadarshikunal** » Wed Mar 25, 2009 4:29 am

I haven't seen the windows tag on your post previously.

In unix its inside <basedir>/Projects/<project name>/core

for windows its defined inside control panel>system properties>advanced>startup and recovery.

check which path is defined there then use dumpchk.exe to validate it and then you can get the debugging tool from the link below to check that file.

http://www.microsoft.com/whdc/DevTools/ ... fault.mspx

priyadarshikunal · Post by **priyadarshikunal** » Wed Mar 25, 2009 4:55 am

or
http://support.microsoft.com/default.aspx?scid&H3Dkb ;en-us;286350

you can use ADPlus to retain the core of processes.

Hope this helps.
Cannot say much about windows

DSguru2B · Post by **DSguru2B** » Wed Mar 25, 2009 7:59 am

Thank you for the info. No dump file there either. For now the batches are running fine. It never happend, ever, the entire time I have been here. Other ETL folks have never seen it. I will wait for it to pop up its head again. Shall keep you folks posted.