Re: Proposing change to the allocatable check in the allocator

Hi all, We had a nice discussion about this in the API working group meeting today. I agree that it's a good idea to do our best to make this change compatible with future updates to the Request call and/or quota. I think it would be beneficial to have a meeting in a few days to brainstorm some

Re: Proposing change to the allocatable check in the allocator

Instead of the master flag, why not a master API call. This will allow to update the value without restarting the master. Another thought is that we should explain operators how and when to use this knob. For example, if they observe a behavioural pattern A, then it means B is happening, and

Re: Proposing change to the allocatable check in the allocator

I would suggest we also consider the possibility of adding per framework control on `min_allocatable_resources`. If we want to consider supporting per-framework setting, we should probably model this as a protobuf, rather than a free form JSON. The same protobuf can be reused for both master

Proposing change to the allocatable check in the allocator

Hi: The allocatable check in the allocator (shown below) was originally introduced to help alleviate the situation where a framework receives some resources, but no cpu/memory, thus cannot