Darin

@tarnfeld I removed the synchronized (this.scheduler) block in MesosTracker.idleCheck no more deadlock. I didn't spend a lot of time testing this and I may have introduced a race condition, I...

So, I spent a lot of time looking at the failure modes of this code lately. What we found was that since the executor continued to run, but the task...

OK, I can try that and report back.

I can confirm there's an issue I've got toree kernels for 1.6 (0.1.0.dev5) and 2.0 (0.2.0.dev1) running and the updating textbox fails to fall a function with the spark 2.0...

I like the idea but think the profiles should reflect the vCores of the NM, so maybe change vcoreMultiplier to vcoreFraction (vcoreRatio??), so a medium size NM would still have...

@taojieterry sorry if I wasn't clear, so I think the config should look like ``` profiles: zero: # NMs launched with this profile dynamically obtain cpu/mem from Mesos cpu: 0...

I think you're on the right track

@taojieterry see my comment about mem vs cpu on line 55 of myriad-scheduler/src/main/java/org/apache/myriad/scheduler/fgs/OfferUtils.java. Also, we'll need to address the vcore ratio for fine grained scaling. The files you'll need to...

@taojieterry I'll probably be testing you're other two PRs and merging the late this week. I'd like to merge this one at the same time but correctness is more important...

@taojieterry no worries, looking forward to the updates when they're ready!