Input record sequence number

[ray.wurlod]

Ask yourself whether uniqueness is sufficient. If it is, gaps in the numbers are completely irrelevant.

[arunkumarmm]

Run the parallel transformer in sequential way and use @INROWNUM

This is the current scenario Anbu. But the jobs run for a longer time. Say the job with sequential mode runs nearly an hour and making it parallel, it runs less than 10 minutes.

[arunkumarmm]

Ask yourself whether uniqueness is sufficient. If it is, gaps in the numbers are completely irrelevant. ...

No Ray. I want the Input records sequence number. The requirement is, we have an error processing and which ever record errors, I need to cature the Input record sequence number and store it in a table for further investigation/reporting.

And If I have only 175 records in today's file and say the last record errors. I cannot say them the record sequence number is 177, which will give no meaning to them.

[anbu]

Run the parallel transformer in sequential way and use @INROWNUM

This is the current scenario Anbu. But the jobs run for a longer time. Say the job with sequential mode runs nearly an hour and making it parallel, it runs less than 10 minutes.

Or finish your processing in one job without sequence number if it is ok for your requirement. In second job create seqeunce number using transformer running in sequential way

[arunkumarmm]

Or finish your processing in one job without sequence number if it is ok for your requirement. In second job create seqeunce number using transformer running in sequential way

I need to map this sequence number to the records which is errored out.

[ray.wurlod]

The only way to guarantee unbroken sequence is to run the stage that generates them - I'd use a Column Generator stage - in sequential mode.

[SBSA_DW2]

recieved this code, hope it helps:

Code: Select all

(1 + @INROWNUM + MOD(@PARTITIONNUM, @NUMPARTITIONS)) + ((@INROWNUM -1) * (@NUMPARTITIONS -1))