AWS Thinkbox Discussion Forums

primary/secondary pools not respected?

We have the following setup:

N idle machines with pools like: 2d, 3d
and
M idle machines with pools like: 2dShared, 2d

We send a job with primary pool = 2dShared, secondary pool = 2d.

I would expect them to first go to the 2dShared machines, and only if those are all full, go to 2d. But maybe i misunderstand how it works? Idle 2d machines won’t be aware of the idle 2dShared boxes, so they just go right ahead and pick the job, right?

Right. Those jobs will spread to the N machines right away because they’re idle. When a slave looks for a job, it first scans the jobs using their primary pool, and then if it finds nothing, scans again using their secondary pool. If a 3d job would come along later, then those N machines would yield to it once they finish their current task for the 2d job.

Privacy | Site terms | Cookie preferences