DataStage Jobs failure due to Broken Pipe

[saur_classic]

Hi,

I am working on Datastage 7.5 Release and the Underlying OS is LINUX SUSE.

All PX Jobs are Failing due to Broken pipe Error.

The Job contains
2-Dataset stage,1 Change Capture and 1 sequential file.

The Exact ERROR is

Fatal Prev_Snapshot_tbl_Ds,0: /bin/echo: write error: Broken pipe
Fatal Prev_Snapshot_tbl_Ds,0: Operator terminated abnormally: received signal SIGBUS

Is there any solution for the above problem?

Can Someone help to get around this Problem...?


Regards,

[ArndW]

All PX Jobs are Failing due to Broken pipe Error.

Usually broken pipes are not the cause of problems but the most visible symptoms. Something is causing your processes to fail, and then their side of the pipes are being closed down and the other side of the process reports this as an error.
There should be more errors in your log than these, look for an error or warning that occurs before the broken pipe messages. Can you use dataset management to view your datasets? Can you view the contents in the designer?

[madhukar]

sometimes it will be due to user permissions also.

[ashwin141]

Hi

Broken pipe may be because of incorrect metadata. Esnure that you are using the right metadata and you are not creating bigger data than waht your fields are. Check properly the metdata of datasets what you have defined and what you are reading here. Sigbus errors is mostly if you try reading an unallocated memory location, which you are not supposed to read.

Regards
Ashwin

[perfman]

When I changed to Load/Append method instead of Upsert, the problem solved. Is there any reason for that?

[ray.wurlod]

Probably that Load/Append finishes faster, giving less opportunity for the pipe to break (or to be broken).