Page 1 of 1

Dataset error: Parallel job reports failure (code 141)

Posted: Wed Oct 05, 2005 6:05 am
by ashwin2912
I am writing into a dataset and the job is aborting giving following error:
Info:
Contents of phantom output file =>
28191 Broken Pipe

Info:
Contents of phantom output file =>
DataStage Job 2490 Phantom 28192

Fatal:
Parallel job reports failure (code 141)

When I replace it with a sequential file it runs successfully. Any clue on what is happening?
Placing a dataset on any output link in the job gives the same result.

Posted: Wed Oct 05, 2005 6:10 am
by ArndW
Can you create another job with a DataSet output as a test to see if it is limited to your one job and one data contents? Does the job abort at or before row one or does it happen downstream? If you use a different APT_CONFIG file (with a different amount of nodes), does it work?

Posted: Wed Oct 05, 2005 6:55 am
by ashwin2912
I have deleted 7-8 stages and backtracked to the first stage and still the issue persisted. The job aborts immediately without processing any rows.

Then I wrote the dataset in sequential mode, and the job ran successfully!!
All other jobs in the same project are running fine with regard to this issue. I wonder what the problem is? :evil:

Posted: Wed Oct 05, 2005 7:06 am
by ArndW
Create a quickie job to write your generated sequential file into a dataset. Does that work?

Posted: Wed Oct 05, 2005 4:38 pm
by ray.wurlod
Have you checked your multi-node configuration file?

Posted: Fri Oct 07, 2005 5:06 am
by Eric
Have you checked that there are no previous dataset files that have the same name (from previous attempts)?

Posted: Mon Oct 10, 2005 11:02 pm
by ashwin2912
The problem disappeared mysteriously....But we realised the other day there was a change in the default configuration file used by multiple projects. Most probably this was the culprit.