Michał Woźniak

Actually, the reported status in Kueue looks very much like created using Kueue before this fix: https://github.com/kubernetes-sigs/kueue/pull/3323 . Can you please confirm you observe the analogous status with the latest...

I think I understand the issue (for now only based on static code and logs analysis). It requires interaction between 3 components: job_controller, workload_controller, provisioning_controller. 0. (precondition) workload is admitted...

Since the nature of the issue is very racy it is a bit tricky to reliably repro. The two existing tests repro the issue but the failure rate is very...

/reopen There is a follow up issue that the ProvisioningRequests don't get create after second eviction, see repro by @PBundyra https://github.com/kubernetes-sigs/kueue/pull/6322

It is now tracked here: https://github.com/kubernetes-sigs/kueue/issues/6323 /close Let me close this one as it specifically mentions the state which should be not possible after the first fix: ``` Status: Admission...

/cc @andreyvelich @tenzen-y @mszadkow

/assign @trasc /cc @alculquicondor

I reviewed https://github.com/kubernetes-sigs/kueue/pull/2047, and I think we could follow the pattern here. The `AdmissionCheck` condition would be `CompatibleWithMultiKueue`, and the reason for inactive ClusterQueue could be `AdmissionCheckNonCompatibleWithMultiKueue`. We would do...

> What if other checks need similar semantics against others? Right, this approach cannot be used for arbitrary pairs of admission checks. However, MultiKueue seems more than an admission check....

I'm ok to close it until we revisit the design or some evidence for users running into this issue.