Archives of the TeradataForum

Message Posted: Wed, 29 May 2002 @ 19:58:18 GMT

	<Prev	Next>		<<First	<Prev	Next>	Last>>

Hi Najam:

MultiLoad does not use the Transient Journal. It performs checkpointing on the input records instead.

If MLoad aborts in the middle of a run, you should be able to simply resubmit it for a restart to automatically take place.

If you cannot resubmit the job, my guess is that 1 of 3 things occurred:

1. The logon userid is different.

2. Enough changes were made to the script since the abort to cause Teradata to think it is a different Mload job.

3. You dropped your restart, error or worktables; don't do this! This action prevents a restart.

Other than these 3 things, it should work. My first suggestion is to use the same script that was initially run.

The only other option is to do a RELEASE MLOAD command for the locked table. HOWEVER, if you are in Phase 4, I think the software has been changed to prevent you from doing this (I hope). The problem with the RELEASE is that previously it left updated rows updated. There is no ROLLBACK (remember - no TJ because it updates a block at a time which makes it faster than a row at a time).

So, first try resubmitting the same script that aborted, especially if this is a production table. If it is a test table, you might not care and can try the RELEASE MLOAD. If you dropped your Mlaod tables, try the RELEASE. If it works, validate that your data is unchanged. If any of it is changed, you may need to reload it from your backup tape. Again, there exist very few, if any, times that you should ever drop your Mload tables in the middle of a run. Especially in production jobs.

Regards and good luck,

Mike

	<Prev	Next>		<<First	<Prev	Next>	Last>>