err in job when the col gen stage execution mode is parallel

[rsrikant]

In my job i am writing from the lookup rejects link to a col generator and then to transformer and finally to a sequential stage.

I have a constraint on the transformer stage to abort the job after 1 row is rejected from lookup stage and this 1 row should be written into the sequential file before the job aborts.

lookup --------> col generator --------> transformer --------> Seq. File

This job runs inconsistently. Sometimes the job aborts by writing the 1 row in to the sequential file and sometimes it aborts with out writing into the sequential file. I am running it in multinode with $APT_BUFFERING_POLICY environment variable set to force buffering. The col generator stage execution mode is parallel. In the sequential stage options "Cleanup on failure" is set to false.

We have few more jobs like this but all of them runs fine except this one. I don't see any difference between this job and other jobs that are running fine.

Finally after some trial and error i changed the col generator stage from parallel to sequential. After changing this, the job aborts properly everytime by writing the 1 row in to the sequential file.

Any idea on what's happening?? Why it is not behaving as expected when the col generator execution mode is set to parallel?

Thanks in advance.

Srikanth

[mandyli]

Hi

This is not releated to col gen problem. I am also faced same kind of problem but I changed the col generator stage from parallel to sequential mode then no error and warning. I am not sure about $APT_BUFFERING_POLICY environment variable. What is use of $APT_BUFFERING_POLICY variable.

Thanks ..