> On Oct 10, 2018, at 12:07 PM, Noam Bernstein <noam.bernst...@nrl.navy.mil> > wrote: > > > slurmd -C confirms that indeed slurm understands the architecture, so that’s > good. However, removing the CPUs entry from the node list doesn’t change > anything. It still drains the node. If I just remove _everything_ having to > do with those counts from the node list item it just picks 1 cpu. >
Interestingly, it looks like maybe the only problem was that I had to manually set the state to Resume. I was playing around with different settings for various properties, and definitely had things screwed up in a way that would trigger a drain state at some point. Apparently after I fixed the situation slurm was actually OK with it, but just didn't reset the state properly. Seems to be OK now, once I got all the items set consistently.