Archives of the TeradataForum

Message Posted: Mon, 19 Jun 2000 @ 13:13:23 GMT

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

Claudio,

We are testing the scheduler at the moment, which means I can offer you advice based on what the manual says but not based on experience!

Using schmon:

To create an allocation group: 'start schmon -a 20 N ABS 10' will create a new allocation group with ID number 20, N denotes all processes belong to a single set, ABS for Absolute priority, 10 for weight.

To create a performance group, and associate the new allocation group number 20 to it: 'start schmon -p 10 LTest 1 0 R 0 20' where 10 is the performance group ID, LTest is the name, 1 is the resource partition, R denotes resource usage as opposed to time of day, 0 means no time changes and 20 is the allocation group to be mapped to this performance group.

Realistically, I don't think that quoting examples like this is very helpful. The syntax is so convoluted that I would strongly recommend that you refer to the 'Teradata User Utilities Guide', chapter 10.

In principle it is possible to do what you seem to want, i.e. vary priority by time of day. Thus:

start schmon -p 10 LBatch 1 0 T 0 20 800 30 1800 20

should change the batch jobs from allocation group 20 to 30 at 8 a.m. and back at 6 p.m. You would need to assign user batchusr with accountstring $LBatch$... To have this take effect.

You can also dynamically change priorities of running jobs and change allocation groups when a session has used a certain amount of CPU, e.g.:

start schmon -p 10 LBatch2 1 0 R 0 20 500 30 1000 40

This changes allocation groups at 500 and 1000 CPU seconds.

I hope this helps.

Regards,
John Dubery

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