SIGPIPE failure - How do you diagnosis cause?

Post questions here relative to DataStage Enterprise/PX Edition for such areas as Parallel job design, Parallel datasets, BuildOps, Wrappers, etc.

Moderators: chulett, rschirm, roy

Post Reply
Teej
Participant
Posts: 677
Joined: Fri Aug 08, 2003 9:26 am
Location: USA

SIGPIPE failure - How do you diagnosis cause?

Post by Teej »

Code: Select all

APT_CombinedOperatorController(1),2: Unexpected termination by Unix signal 13(SIGPIPE)
Right now, the best method I could do to diagnosis SIGPIPE failures are to eliminate all possible warning messages (even column propagation warnings at times can be fixable).

But I am at a loss at how to further diagnosis the actual cause of this cruelity to innocent processes by the big bad UNIX. Doing the above stuff would just make this message go away... without really understanding what in the job caused it.

So I'm calling out to the rest of you guys -- How do you deal with SIGPIPE failures? What do you do to diagnosis the cause to eliminate this error showing up again?

-T.J.
Developer of DataStage Parallel Engine (Orchestrate).
bigpoppa
Participant
Posts: 190
Joined: Fri Feb 28, 2003 11:39 am

SIGPIPE failure - How do you diagnosis cause?

Post by bigpoppa »

T.J.,

What are the output messages for the other "APT_CombinedOperatorController"s?

Also, have you tried turning off the combined operators env var? That would be a good step in debugging this error.

- BP
gbryans
Participant
Posts: 2
Joined: Sun Apr 11, 2004 8:12 pm

Re: SIGPIPE failure - How do you diagnosis cause?

Post by gbryans »

Enlarge the scratch area and ensure adequate page space. At least 1x physical memory. The reading process for a named pipe has been killed, probably space issue.
Disable operator combination with "APT_DISABLE_COMBINATION=1", force compile, run again.
Teej wrote:

Code: Select all

APT_CombinedOperatorController(1),2: Unexpected termination by Unix signal 13(SIGPIPE)
Right now, the best method I could do to diagnosis SIGPIPE failures are to eliminate all possible warning messages (even column propagation warnings at times can be fixable).

But I am at a loss at how to further diagnosis the actual cause of this cruelity to innocent processes by the big bad UNIX. Doing the above stuff would just make this message go away... without really understanding what in the job caused it.

So I'm calling out to the rest of you guys -- How do you deal with SIGPIPE failures? What do you do to diagnosis the cause to eliminate this error showing up again?

-T.J.
Post Reply