DSXchange

Details:-
Teradata V2r5
Datastage 7.5x2
OS Unix

I have a simple job of sequential file + Transform + Multiload.

Within MultiLoad Stage in the Output file path i mention a directory having permissions ( drwxr-xr-x) and no file is mentioned in the Report Files.

If one gives wrong password, the job hangs within Datastage and it does not abort within Datastage, however the multiload log file is generated mentioning Invalid Password.

I have tried with various combinations within
Sessions ( max - 4 , min - 1)
Error Counts
Tenacity

However the working does not change in either of the cases.

If one mentions a Report File Name the job successfully aborts within Datastage and the multiload log file is generated as usual.

I searched the forums however was unable to find a relevant link with such a problem.

I am unable to understand why it behaves so.


Thanks in advance,

May I offer one possible solution? It does not explain your error but may work for you anyway. We have had problems getting MultiLoad to work correctly here (different errors, though). As a result, we have switched to using the Teradata Enterprise stage for all read/write processing. My understanding is that it is faster and better supported (this recommendation came 'unofficially' from both IBM/Ascential and NCR).

The Enterprise stage uses FastExport to read data and FastLoad to write. Normally, the FastLoad utility requires that the target table is empty. However, I believe the way DataStage gets around this is to build a temporary table for loading and the data is then written to the designated target table. (Anyone can correct me if I am wrong).

HTH.

Brad.

Hi,
Yes, the Enterprise Stage does an FastLoad and Insert Select statement.
However, we had a requirement of Upsert hence were interested in using Multiload.
We were thinking along the lines that one can apply change capture to identify inserts and updates. However with Updates one is still stuck with either Multi Load or API stages.
Thanks for the reply,
Cheers

Had the same problem at the previous client and could never find a solution either. Ended up documenting the process of 'if the job is hanging check the multiload log file to see if there is a password error. If so abort job manually, set password correctly then restart job sequence' in the production support notes and leaving it at that. Not the nicest workaround.