Rework implementation for multiple admissions per cycle

[alculquicondor]

What would you like to be cleaned:

Re-design the logic that allows multiple admissions to happen per cycle without incurring in over admission.

I think a simpler approach would be along the lines of modifying the snapshot (add admitted workloads and remove preempted workloads within the cycle) and checking if the incoming workload still fits. There is a special case for workloads needing preemption where there are no candidates for preemption. I think we can add a smaller fake workload that reserves space for the workload without going beyond the nominal quotas.

Why is this needed:

The current logic has been reworked multiple times with various pros and cons.

• Fix preemption blocked by earlier pending Workload #1866
• Fix pending workload from StrictFIFO CQ doesn't block borrowing from other CQ #1399
• Consider priority across ClusterQueues in a cohort #1024
• Fix concurrent admission in cohort when borrowing #805
• Prevent borrowing when preemption could help #475

Now it's very hard to follow.

[alculquicondor]

cc @yaroslava-serdiuk

[k8s-triage-robot]

The Kubernetes project currently lacks enough contributors to adequately respond to all issues.

This bot triages un-triaged issues according to the following rules:

• After 90d of inactivity, lifecycle/stale is applied
• After 30d of inactivity since lifecycle/stale was applied, lifecycle/rotten is applied
• After 30d of inactivity since lifecycle/rotten was applied, the issue is closed

You can:

• Mark this issue as fresh with /remove-lifecycle stale
• Close this issue with /close
• Offer to help out with Issue Triage

Please send feedback to sig-contributor-experience at kubernetes/community.

/lifecycle stale

[alculquicondor]

/remove-lifecycle stale

[gabesaba]

/cc

[gabesaba]

/assign

[gabesaba]

in #2641, we enhance and simplify multiple preemption logic, by using the snapshot to account for usage. After promotion to beta in 0.9, we can delete the old code in 0.10

/close

[k8s-ci-robot]

@gabesaba: Closing this issue.

In response to this:

in #2641, we enhance and simplify multiple preemption logic, by using the snapshot to account for usage. After promotion to beta in 0.9, we can delete the old code in 0.10

/close

Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes-sigs/prow repository.