Arvados

I tried spinning up 82 nodes today and only 68 spun up and it waited an hour and a half. I have a couple ideas.

If this many nodes are trying to spin up and they haven't after 30 minutes, support should get alerted to kill the job because nodes aren't being spun up and see the issue. (or some other failsafe so that resources aren't being wasted)

Show the user the maximum number of nodes that they can reliable spin up at this moment. If some nodes are broken or node manager can't see them or something, then this should let the user know that they shouldn't try getting more nodes than that number.