[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Condor-users] Help with scheduling start/evict policy in condor_config




On 4/14/10 11:48 AM, Dan Bradley wrote:
> One likely source of trouble in this policy is that RANK is inherently
> a preemptive mechanism.  RANK is only relevant when deciding whether
> to preempt an existing job with a new better-ranked one.  This can
> lead to rapid cycles of preemption in some cases.

I can reform the question as follows: What needs to be done to make sure
that in each matching cycle idle job slots are considered first?  If no
matches can be done of available jobs with idle job slots, then eviction
is to be expected.  What we think we see now is that matching is done
against an arbitrary machine, whether it is idle or not, and the RANK
expression means that no consideration is given to a running job, even
when other idle nodes are available.

The current RANK expression is:

RANK = (regexp("group_cmsuser",TARGET.AccountingGroup) ||
regexp("group_cmsprod",TARGET.AccountingGroup) ||
regexp("group_cdf",TARGET.AccountingGroup) ||
regexp("group_monitor",TARGET.AccountingGroup) ||
regexp("group_mitlns",TARGET.AccountingGroup) ||
regexp("group_cmshi",TARGET.AccountingGroup))

Which appears to be a boolean expression.  Is the answer to append (I am
still not clear on when $(FOO) or just FOO should be used in classad
expressions):

|| ($(CondorLoadAvg) <= 0.2)

or

|| (CondorLoadAvg <= 0.2)

with the intention that this would prefer idle job slots?

Thanks,

Ian