Scheduler Setup - Every first run fails, every second works

[odo2063]

Yes, but if I start "short job" while "long job" is running, "short job" never gets nodes.

[Giri]

I don't think I understand what you are describing. You are not seeing nodes for cluster with short jobs when its jobs are running? Can you explain in more detail exactly what you think should happen? May be attach screenshots?

[odo2063]

ok...What I expect is
for FIFO behaviour:
when "long job" is finished, "short job" is being executed(gets nodes).

For "fair-use" behaviour:
while "long job" is running, "short Job" is running.(round robin)

for "LIFO" behaviour:
while "long job" is being executed, "long job" pauses for the execution of "short job".


What happens:
for all behaviours:
"short job" is never executed.

[Giri]

I will take a look at it this weekend.

[Giri]

I committed fix for this issue; change is in dispynode. Please confitrm if this works for you.

[odo2063]

First:
Sorry for the delay ...I had a full schedule

Second:
Seems to work fine now Thank you very much!!!

Third:
This is kind of a kindly feature request. Do you think it would be possible, with a reasonably timely effort, to implement a priority scheduler?

e.g.
if
dispy.SharedJobCluster(compute, ....., priority=0)
is running
and then
dispy.SharedJobCluster(somethingElse, ...., priority=10)
will be executed with respect to the priority.