kubernetes-sigs/kueue v0.15.3

on GitHub

Changes since v0.15.2:

Changes by Kind

Feature

• CLI: Support "kwl" and "kueueworkload" as a shortname for Kueue Workloads. (#8469, @kannon92)

Bug or Regression

• Add lws editer and viewer roles to kustomize and helm (#8515, @kannon92)

• FailureRecovery: Fix Pod Termination Controller's MaxConcurrentReconciles (#8665, @gabesaba)

• Fix ClusterQueue deletion getting stuck when pending workloads are deleted after being assumed by the scheduler. (#8548, @sohankunkerkar)

• Fix a bug that WorkloadPriorityClass value changes do not trigger Workload priority updates. (#8499, @ASverdlov)

• HC: Avoid redundant requeuing of inadmissible workloads when multiple ClusterQueues in the same cohort hierarchy are processed. (#8510, @sohankunkerkar)

• Integrations based on Pods: skip using finalizers on the Pods created and managed by integrations.

  In particular we skip setting finalizers for Pods managed by the built in Serving Workloads Deployments,
  StatefulSets, and LeaderWorkerSets.

  This improves performance of suspending the workloads, and fixes occasional race conditions when a StatefulSet
  could get stuck when deactivating and re-activating in a short interval. (#8573, @mbobrovskyi)

• JobFramework: Fixed a bug that allowed a deactivated workload to be activated. (#8438, @chengjoey)

• LeaderWorkerSet: Fixed a bug that prevented deleting the workload when the LeaderWorkerSet was scaled down. (#8673, @mbobrovskyi)

• MultiKueue now waits for WorkloadAdmitted (instead of QuotaReserved) before deleting workloads from non-selected worker clusters. To revert to the previous behavior, disable the MultiKueueWaitForWorkloadAdmitted feature gate. (#8600, @IrvingMg)

• MultiKueue: Fix a bug that the priority change by mutating the kueue.x-k8s.io/priority-class label on the management cluster is not propagated to the worker clusters. (#8574, @mbobrovskyi)

• MultiKueue: fix the eviction when initiated by the manager cluster (due to eg. Preemption or WairForPodsReady timeout). (#8402, @mbobrovskyi)

• ProvisioningRequest: Fixed a bug that prevented events from being updated when the AdmissionCheck state changed. (#8404, @mbobrovskyi)

• Revert the changes in PR #8599 for transitioning
  the QuotaReserved, Admitted conditions to False for Finished workloads. This introduced a regression,
  because users lost the useful information about the timestamp of the last transitioning of these
  conditions to True, without an API replacement to serve the information. (#8612, @mbobrovskyi)

• Scheduling: fix the bug that setting (none -> some) a workload priority class label (kueue.x-k8s.io/priority-class) was ignored. (#8584, @andrewseif)

• TAS: Fix a bug that MPIJob with runLauncherAsWorker Pod indexes are not correctly evaluated during rank-based ordering assignments. (#8663, @tenzen-y)

• TAS: Fixed an issue where workloads could remain in the second-pass scheduling queue (used for integration
  or TAS with ProvisioningRequests, and for TAS Node Hot Swap) even if they no longer require to be in the queue. (#8431, @skools-here)

• TAS: fix TAS resource flavor controller to extract only scheduling-relevant node updates to prevent unnecessary reconciliation. (#8453, @Ladicle)

• TAS: significantly improves scheduling performance by replacing Pod listing with an event-driven
  cache for non-TAS Pods, thereby avoiding expensive DeepCopy operations during each scheduling cycle. (#8484, @gabesaba)

Full Changelog: v0.15.2...v0.15.3