This is strange,
but, I have a limit group set for Nuke jobs (I only have 5 Nuke render licenses). When all machines are offline (powered down via the Power Management) and submit a job, it wakes up only 5 machines, which is correct.
But, when the job is finished and the machines are still idle and I submit another Nuke job (via the same limit group), the Power Management wakes up some more machines, for no reason?!
Why is that? And can it be fixed?
I checked the code, and the only time Deadline won’t start up a machine for a job with a limit group is if that limit is reached. It doesn’t handle the case where there are already that many machines sitting idle that could use that limit group. We’ve logged this as a bug, and will look at fixing it. Unfortunately, there isn’t anything that can be done to workaround it for now.
No problem. 
At least you know about this scenario. Thanks!
Just had a quick question. How often does your power management check run? The default is every 5 minutes, and the reason I ask is because within that time frame, the 5 slaves that are already running should already be working on the job. Yes, there are the odd cases where the power management check will take place shortly after the job is submitted and extra slaves might get woken up, but this is probably acceptable.
Cheers,
I think I changed the defaults to either 3 or 10s, but, honestly, I don’t remember anymore and since I’m running Pulse without the UI I can’t check it out right now (don’t want to disturb the rendering).