Fix in-place updates over ineligible nodes

[jorgemarey]

Fixes #12263

[hashicorp-cla]

All committers have signed the CLA.

[tgross]

Hi @jorgemarey! Sorry for the delay on reviewing this. We had an internal discussion over this about whether or not this was actually the expected behavior, and I think we've come to the conclusion that what you're looking for is correct. I need to review how we handle this case in the scheduler to make sure we're not missing anything, but once I've done so I'll come back to review this PR in detail. Thanks for your patience!

[tgross]

LGTM. I've added a changelog entry and rebased on main so we can ship in the upcoming Nomad 1.3.0

I need to review how we handle this case in the scheduler to make sure we're not missing anything, but once I've done so I'll come back to review this PR in detail.

I've had a chance to follow-up on that. The scheduler removes ineligible nodes from the "stack" when we do feasibility checking in the function readyNodesInDCs, which in turn calls (*Node) Ready() and that checks the eligibility. The eligibility of nodes for existing allocations isn't considered at all in the scheduler, which is how we end up with those in-place alloc updates landing on the plan applier as you've noted here.

[github-actions]

I'm going to lock this pull request because it has been closed for 120 days ⏳. This helps our maintainers find and focus on the active contributions.
If you have found a problem that seems related to this change, please open a new issue and complete the issue template so we can capture all the details necessary to investigate further.