[GE users] Machine Priority

Reuti reuti at staff.uni-marburg.de
Tue Oct 19 19:36:28 BST 2004


> If I understood everything said about this topic this morning, To 
> implement one of the ideas, you would change the scheduler to a seq_no 

Yes.

> sort.  Then define a queue for each set of machines Q1,Q2,Q3 and give 
> the ones you want to always have run a job if they are available a 
> seq_no=1, the ones to fill next, a seq_no=2, and the next set a seq_no=3 
> (assuming serial jobs)

You need only one cluster queue. Just add all three hostgroups and set 
the priority like:

qname                 serial
hostlist              @type1 @type2
seq_no                0,[@type1=50],[@type2=100]

> If this part is correct, can you still have the allq queue defined and 
> maybe make it subordinate to the others so if someone requests a 
> parallel environment that would require machines from all 3 queues it 
> can be assigned to the allq?

I would do it in the other way: maybe it makes no sense to mix the three 
different speeds for a parallel job, so having three queues to avoid 
mixing of the speed would be good. Unfortunately parallel jobs will also 
spread around queues (in 6.0).

<sidenote>

$ cat tester.sh.o133
NSLOTS=8
ic002 2 extra at ic002 UNDEFINED
ic001 2 extra at ic001 UNDEFINED
ic003 2 vast at ic003 UNDEFINED
ic004 2 vast at ic004 UNDEFINED

Andy: is this a bug or intended?

</sidenote>

If intended, then you have to define also three PEs (like in 5.3) with 
  similar names paralle1, parallel2 and parallel3 and request parallel* 
in the qsub.

The PEs should be set in the queues accordingly (i.e. para1 with @type1 
host only PE parallel1). - Reuti


---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe at gridengine.sunsource.net
For additional commands, e-mail: users-help at gridengine.sunsource.net




More information about the gridengine-users mailing list