Archives of the TeradataForum

Message Posted: Thu, 29 Jul 2004 @ 22:23:21 GMT

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

The short answer is YES!

The long answer is:

As I said, I'm a bit fuzzy on the details of amp end-of-step coordination, but I know enough to say with confidence that an amp worker tasks which finish early (because it doesn't have many rows to deal with) will signal that it's done and go back to wait on its message queue without waiting for other amps to finish.

That's if no amp encounters an error, of course. If one of the amp worker tasks servicing a step encounters an error, there's a process - that I really don't know the details of - which gets all the amps to abort.

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