[fg91]

What type of PR is this?

/kind bug
/area multikueue

What this PR does / why we need it:

When a pod workload is admitted in a MultiKueue worker cluster, the cluster autoscaler in the management cluster triggers a scale up even though the pod will never run in the management cluster due to the scheduling gates. This PR prevents this.

Which issue(s) this PR fixes:

When a pod workload is created in the management cluster, the following is added to its spec and status:

  schedulingGates:
  - name: kueue.x-k8s.io/admission
  - name: kueue.x-k8s.io/topology

  conditions:
  - lastProbeTime: null
    lastTransitionTime: '...'
    message: Scheduling is blocked due to non-empty scheduling gates
    reason: SchedulingGated
    status: 'False'
    type: PodScheduled

Such a pod is not considered a scale up candidate by the cluster autoscaler.

However, once the pod is admitted in the worker cluster, the PodScheduled condition is currently overwritten with:

    conditions:
    - lastProbeTime: null
      lastTransitionTime: "..."
      message: '0/4 nodes are available: ...'
      observedGeneration: 2
      reason: Unschedulable
      status: "False"
      type: PodScheduled

The cluster autoscaler considers such a pod a scale up candidate despite the scheduling gates. Accordingly, I see a scale up event in the management cluster:

Pod triggered scale-up: [{https://... 0->1 (max: 1000)}]

This scale up in the management cluster doesn't make any sense as the workload will never run there. To prevent it, we must preserve the reason: SchedulingGated for the type: PodScheduled condition in the management cluster.

Does this PR introduce a user-facing change?

MultiKueue: Fixed a bug where admitted Pod workloads could trigger unnecessary Cluster Autoscaler scale-ups
in the manager cluster. Kueue now preserves the scheduling-gated PodScheduled condition for manager-cluster
Pods, since they are intended to run only in worker clusters.

Summary by CodeRabbit

Release Notes

• Bug Fixes

  • Improved Pod status synchronization to preserve the local PodScheduled condition when a Pod is scheduling-gated, while still keeping other status details in sync with the remote Pod.

• Tests

  • Added coverage to ensure scheduling-gated manager Pod PodScheduled remains unchanged during job status synchronization.

• E2E

  • Updated the multikueue end-to-end expectations to reflect the manager Pod staying scheduling-gated for its lifetime.

[k8s-ci-robot]

Hi @fg91. Thanks for your PR.

I'm waiting for a kubernetes-sigs member to verify that this patch is reasonable to test. If it is, they should reply with /ok-to-test on its own line. Until that is done, I will not automatically test new commits in this PR, but the usual testing commands by org members will still work.

Regular contributors should join the org to skip this step.

Once the patch is verified, the new status will be reflected by the ok-to-test label.

I understand the commands that are listed here.

Details

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.

[coderabbitai]

📝 Walkthrough

Walkthrough

syncLocalPodWithRemote in the MultiKueue pod adapter is updated to preserve the local PodScheduled condition when the local pod is gated, instead of overwriting it with the remote pod's condition. Two helper functions (findPodCondition, setPodCondition) are added for condition management, two new unit test cases validate the behavior, and an e2e test expectation is aligned to reflect the preserved scheduling-gated state.

Changes

Gated PodScheduled condition preservation in MultiKueue sync

Layer / File(s)	Summary
Sync logic and condition helpers pkg/controller/jobs/pod/pod_multikueue_adapter.go	findPodCondition and setPodCondition helpers added for condition lookup and upsert. syncLocalPodWithRemote is updated to preserve the pre-overwrite local PodScheduled condition when the pod is gated, preventing the remote status overwrite from replacing the kueueschedulinggate-owned condition.
Unit tests for gated PodScheduled preservation pkg/controller/jobs/pod/pod_multikueue_adapter_test.go	Two new TestMultiKueueAdapter cases assert that the manager pod's PodScheduled=False/PodReasonSchedulingGated condition is retained after SyncJob regardless of whether the remote worker pod reports unschedulable or fully scheduled status.
E2E test expectation for scheduling-gated pod test/e2e/multikueue/baseline/e2e_test.go	E2E test expectation updated to reflect that the manager-cluster pod's PodScheduled condition remains False with reason SchedulingGated for its lifetime during the admitted pod scenario.

Estimated code review effort

🎯 2 (Simple) | ⏱️ ~12 minutes

Poem

🐇 Hoppy news from the scheduler lane,
A gated pod need not bear the unscheduled blame!
The local condition is safely tucked away,
While remote status arrives to display.
No autoscaler confusion, no misplaced shame —
The rabbit kept conditions exactly the same! 🌸

🚥 Pre-merge checks | ✅ 4 | ❌ 1

❌ Failed checks (1 warning)

Check name	Status	Explanation	Resolution
Docstring Coverage	⚠️ Warning	Docstring coverage is 50.00% which is insufficient. The required threshold is 80.00%.	Write docstrings for the functions missing them to satisfy the coverage threshold.

✅ Passed checks (4 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title directly and clearly summarizes the main change: preserving SchedulingGated status to prevent cluster autoscaler scale-ups in MultiKueue manager clusters.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests

---

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

[mimowo]

/ok-to-test

[mimowo]

@fg91 that is a nice PR, thak you, please post a release note in the PR description in a dedicated block.

[fg91]

@fg91 that is a nice PR, thak you, please post a release note in the PR description in a dedicated block.

Done

[netlify]

✅ Deploy Preview for kubernetes-sigs-kueue canceled.

Name	Link
🔨 Latest commit	ca915b7
🔍 Latest deploy log	https://app.netlify.com/projects/kubernetes-sigs-kueue/deploys/6a30238445ddd10008cbfe76

[fg91]

I'll adapt pull-kueue-test-e2e-multikueue-baseline-main.

[mimowo]

/release-note-edit

MultiKueue: Fixed a bug where admitted Pod workloads could trigger unnecessary Cluster Autoscaler scale-ups
in the manager cluster. Kueue now preserves the scheduling-gated PodScheduled condition for manager-cluster
Pods, since they are intended to run only in worker clusters.

[mimowo]

Looks great, thank you for hardening the MultiKueue implementation 👍
/lgtm
/approve
/cherrypick release-0.18
/cherrypick release-0.17

[k8s-infra-cherrypick-robot]

@mimowo: once the present PR merges, I will cherry-pick it on top of release-0.17, release-0.18 in new PRs and assign them to you.

Details

In response to this:

Looks great, thank you for hardening the MultiKueue implementation 👍
/lgtm
/approve
/cherrypick release-0.18
/cherrypick release-0.17

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.

[k8s-ci-robot]

LGTM label has been added.

Details

Git tree hash: e0f9c1d48492f4a8eeab645e507e6ee47e317b10

[k8s-ci-robot]

[APPROVALNOTIFIER] This PR is APPROVED

This pull-request has been approved by: fg91, mimowo

The full list of commands accepted by this bot can be found here.

The pull request process is described here

Details

Needs approval from an approver in each of these files:

• pkg/controller/OWNERS [mimowo]
• test/OWNERS [mimowo]

Approvers can indicate their approval by writing /approve in a comment
Approvers can cancel approval by writing /approve cancel in a comment

[k8s-infra-cherrypick-robot]

@mimowo: new pull request created: #12272

Details

In response to this:

Looks great, thank you for hardening the MultiKueue implementation 👍
/lgtm
/approve
/cherrypick release-0.18
/cherrypick release-0.17

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.

[k8s-infra-cherrypick-robot]

@mimowo: new pull request created: #12273

Details

In response to this:

Looks great, thank you for hardening the MultiKueue implementation 👍
/lgtm
/approve
/cherrypick release-0.18
/cherrypick release-0.17

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.