Archives of the TeradataForum

Message Posted: Fri, 14 Oct 2006 @ 00:43:27 GMT

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

Srikanth,

Here is how you want might want to code that, using a single performance group and multiple performance period/allocation group pairs:

WHAT YOU HAVE DEFINED

     schmon -p 8 L1 1 0 T 2323 5
     schmon -p 10 M1 1 2 T 400 6
     schmon -p 12 H1 1 4 T 600 7

POSSIBLE REDEFINITION

     schmon -p 8 L1 1 0 T 2323 5 400 6 600 7

Up until 2323 all queries mapping to L1 will use AG 5, from 2323 until 0400 hours they will use AG6, from 0400 hours to 0600 hours they will use AG7, from 0600 hours until 2323 they will use AG 5.

Something in your setup looks odd to me, if you don't mind my pointing it out. You have no reserved AMP worker tasks selected:

     schmon -w 0 80

Yet you have expedited one of your allocation groups (the "x" means the allocation group should use the reserved AMP worker task reserve pool):

     schmon -a 7 S X DEF 20

The result of that combination is that is that processes active in the expedited allocation froup (AG 7) will categorized as MSGWORKEIGHT work types, but will not have any reserves to draw from. They will use AMP worker tasks from the unassigned pool only. But that's not the main problem. The bad side-effect is that the additional reserve pools for MSGWORKNINE and MSGWORKTEN work types (for expedited spawned work and expedited spawned-spawned work) will still remain at 2 each, so you have reduced the unreserved AMP worker task count by 4, without getting any value from doing that.

Is it possible you had turned on an AMP worker task reserve at one point, and then turned it back to zero but forgot to take the "x" off the allocation group that you had expedited?

Thanks, -Carrie

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