github/kubernetes
1
0
Fork 0
mirror of https://github.com/k3s-io/kubernetes.git synced 2026-07-18 04:54:54 +00:00
Code Issues Projects Releases Wiki Activity
136,827 Commits 42 Branches 9,765 Tags
caecddc9093881af9a2ceb3ad7d9dfd29a286321
Commit Graph

2104 Commits

Author SHA1 Message Date
Jon Huhn
71f02bb2f3 DRA: fix Pods in a scheduling gang sharing ResourceClaims 2026-03-18 17:30:19 -05:00
Kubernetes Prow Robot
7a3a6cf4be Merge pull request #136725 from pravk03/native-dra-2 
Introduce support of DRA for Native Resources
 2026-03-19 03:36:38 +05:30
Praveen Krishna
3cd22dcb4f Verify that assumed pod NodeAllocatableResourceClaimStatuses matches the status to patch. 2026-03-18 19:20:11 +00:00
Praveen Krishna
11c8bf1f13 Cleanup NodeAllocatableDRAClaimStatus in Unreserve 2026-03-18 19:20:11 +00:00
Praveen Krishna
cdcfc4eeb3 Add integration tests. 2026-03-18 19:20:10 +00:00
Praveen Krishna
a84f938404 Validate Pod resources cover container and DRA requests. 2026-03-18 19:20:10 +00:00
Praveen Krishna
628365297a Add validation check to disallow node-allocatable resource claim sharing. 2026-03-18 19:20:10 +00:00
Praveen Krishna
9e22d7b033 Include node-allocatable claims in kubelet admission check 2026-03-18 19:20:10 +00:00
Praveen Krishna
7c4f9ace63 Account for node-allocatable claims in NodeInfo cache 2026-03-18 19:20:10 +00:00
Praveen Krishna
af186d3534 Implement changes in DRA Plugin to handle node-allocatable claims 
* Filter: Calculate the pod's total resource footprint by combining spec requests and DRA allocations, respecting accounting policies. Perform node resource fit checks. Cache the resulting NodeAllocatableResourceClaimStatus in CycleState.
* PreBind: Retrieve result from CycleState and patch it into Pod.Status.
 2026-03-18 19:17:30 +00:00
vshkrabkov
5ea4532969 Add PodGroupPodsCount placement score scheduler plugin (#137488) 
* placement score plugin, grounded on count of scheduled pods and pods for placement

* align with v1alpha2 api changes

* addressing review comments

* plugin renaming
 2026-03-18 23:20:57 +05:30
Kubernetes Prow Robot
0d28578de1 Merge pull request #137795 from ttsuuubasa/dra-binding-conditions-beta 
DRA Device Binding Conditions: graduate to beta
 2026-03-18 19:18:33 +05:30
Kubernetes Prow Robot
5953c57c6b Merge pull request #137607 from 0xMH/fix/133531-dra-queueing-hint 
DRA: return error on filter timeout instead of unschedulable
 2026-03-18 14:26:31 +05:30
Tsubasa Watanabe
57e649ae03 DRA Device Binding Conditions: update unit and integration tests 
- update expectations for the default BindingTimeout in KubeSchedulerConfiguration
- DRA unit tests:
  - enable DRADeviceBindingConditions by default
  - add allocationTimestamp to expected ResourceClaims for PreBind cases
- disable DRADeviceBindingConditions when testing the stable allocator in TestSchedulerPerf

Signed-off-by: Tsubasa Watanabe <w.tsubasa@fujitsu.com>
 2026-03-18 13:44:30 +09:00
Hamza
67deec3bfa DRA: include node name in timeout error and rework FilterTimeout test setup 2026-03-18 04:33:24 +01:00
Bartosz
d9da8c7c4a Add scheduling constraints to v1alpha2 pod group api 
Add plugin to generate placements based on scheduling constraints

Co-authored-by: Antoni Zawodny <zawodny@google.com>
 2026-03-17 13:27:46 +00:00
Maciej Skoczeń
38f536c713 Use CycleState.IsPodGroupSchedulingCycle instead of NeedsPodGroupScheduling for pods 2026-03-17 09:10:52 +00:00
Kubernetes Prow Robot
c332baec33 Merge pull request #136155 from bhope/metrics-beta-scheduler 
Promote scheduler metrics to beta
 2026-03-17 04:47:37 +05:30
Kubernetes Prow Robot
d763ae8f6f Merge pull request #137073 from iomarsayed/snapshot-podgroup-state-before-scheduling-cycle 
Snapshot PodGroup state before initiating scheduling cycle
 2026-03-16 23:37:48 +05:30
Kubernetes Prow Robot
ac976cf6a1 Merge pull request #136652 from brejman/kep-5732-tas-placement-plugin-config 
Extend NodeResourcesFit plugin to score placements in the PodGroup scheduling cycle
 2026-03-16 23:37:40 +05:30
Bartosz
9533a4eddb Implement PlacementScorePlugin in NodeResourcesFit 2026-03-16 15:13:46 +00:00
Kubernetes Prow Robot
1817e10998 Merge pull request #136185 from tallclair/ndf-bitmap 
Optimize NodeDeclaredFeatures with a bitmap FeatureSet implementation
 2026-03-14 06:37:34 +05:30
Prathamesh Bhope
d66ae77959 Promote scheduler metrics to beta 2026-03-13 16:06:42 -07:00
Omar Sayed
e1b18e34ff snapshot pod group state before scheduling cycle and embed pod group manager into cache 2026-03-13 21:44:17 +00:00
Kubernetes Prow Robot
7f3a5ab96f Merge pull request #136579 from romanbaron/reuse-scheduling-signature 
Reuse pod scheduling signature for opportunistic batching
 2026-03-13 20:15:39 +05:30
Patrick Ohly
85bca3b684 DRA device taints: fix beta-enabled, alpha-disable configurations 
DeviceTaintRule is off by default because the corresponding v1beta2 API group
is off. When enabled, the potentially still disabled v1alpha3 API version was
used instead of the new v1beta2, causing the scheduler to fail while setting up
informers and then not scheduling pods.
 2026-03-13 09:20:57 +01:00
Tim Allclair
39a4f7654a Optimize NDF FeatureMapper.Unmap for sparse feature sets 2026-03-13 04:28:16 +00:00
Tim Allclair
b1b75f93d7 Make size explicit; switch to binary string format 2026-03-13 04:28:16 +00:00
Tim Allclair
1c2b07fa22 Avoid computing feature diff when not necessary 2026-03-13 04:28:16 +00:00
Tim Allclair
f91f641a65 Switch to bitmapped FeatureSet implementation. 2026-03-13 04:28:16 +00:00
Tim Allclair
e4521526b4 NodeDeclaredFeatures: Add global default NDF registry 2026-03-13 04:28:16 +00:00
Kubernetes Prow Robot
21305568b0 Merge pull request #137083 from brejman/generate-plugin 
Add placement generator plugin interfaces and logic for running them
 2026-03-13 06:15:33 +05:30
Roman Baron
6fcb95e72e scheduler: Moved TestQueuedPodInfo_UpdateInvalidatesSignature from queue/scheduling_queue_test.go to framework/types_test.go 2026-03-12 21:24:23 +02:00
Roman Baron
c0e973dc70 scheduler: Replaced context.Context and testing.T parameters with ktesting.TContext in scheduling_queue_test.go 2026-03-12 17:31:11 +02:00
Kubernetes Prow Robot
9874e76ac4 Merge pull request #137662 from tosi3k/revert-136254-extend-postfilter 
Revert "Extend PostFilterResult with a list of victim Pods"
 2026-03-12 17:33:42 +05:30
Bartosz
43c5d2a419 Add PlacementGeneratePlugin interface and runner 2026-03-12 09:33:05 +00:00
Antoni Zawodny
fa29c9db6a Revert "Extend PostFilterResult with a list of victim Pods" 2026-03-12 10:03:35 +01:00
Roman Baron
de1385fe1b scheduler: Added ObserveFrameworkDurationAsync to metrics recorder 2026-03-12 10:31:38 +02:00
Roman Baron
7b00255135 scheduler: Removed plugin stats from pod signing process 2026-03-12 10:31:04 +02:00
Roman Baron
2904e7f309 scheduler: replaced logger with HandleErrorWithLogger 2026-03-12 10:30:38 +02:00
Roman Baron
e6ee21c94f scheduler: Invalidate PodSignature in QueuedPodInfo Update method 2026-03-12 10:30:38 +02:00
Roman Baron
3a6c169034 scheduler: Reuse scheduling signature for opportunistic batching 2026-03-12 10:30:32 +02:00
Tsubasa Watanabe
30b811a99b DRA Device Binding Conditions: add metrics for prebind flow 
This commit introduces metrics and improves log outputs for
DRA Device Binding Conditions (KEP-5007):

- scheduler_dra_bindingconditions_allocations_total

  Counts the number of per-device scheduling attempts
  during PreBind where BindingConditions are in use

- scheduler_dra_bindingconditions_wait_duration_seconds

  Observes the time spent waiting for BindingConditions
  to be satisfied during PreBind.

Signed-off-by: Tsubasa Watanabe <w.tsubasa@fujitsu.com>
 2026-03-12 17:19:13 +09:00
Kubernetes Prow Robot
efc8cc256a Merge pull request #137201 from brejman/score-plugin 
Add placement scorer plugin interfaces and logic for running them
 2026-03-12 12:41:41 +05:30
Kubernetes Prow Robot
031f8ac9ed Merge pull request #136287 from abel-von/optimize-podgroupinfo 
scheduler: optimize podGroupInfo to minimize the lock time
 2026-03-12 12:41:34 +05:30
Kubernetes Prow Robot
802b3f744b Merge pull request #133622 from KunWuLuan/feat/volume-limit-acc 
Use indexer to acclerate volume limit plugin
 2026-03-12 11:27:34 +05:30
Abel Feng
2153ed1852 scheduler: optimize podGroupInfo to minimize the lock time 
Gang scheduler will add the pod into the podGroupInfo before pod
enqueue, if there are thousands of pods in a podGroupInfo, The
call of AssumedPods, AssignedPods and AllPods will hold the lock and
clone the map, so that new Pods should wait there for a long time to add into
the podGroupInfo, also it will be a long wait to enqueue the pod.

In our test, a pod will wait seconds to enqueue if we have 50000 pod in
a gang group.

In this PR, we can avoid the traverse and clone of the map by adding
AllPodsCount, AssumedPodsCount, AssignedPodsCount method, and we make
sure that assumed pods and assigned pods are disjoint.
 2026-03-11 17:21:36 +08:00
Hamza
0680960e5e Return Error from DRA filter on timeout to enable automatic retry 2026-03-11 02:20:30 +01:00
Kubernetes Prow Robot
d47f3f253b Merge pull request #137343 from gnufied/prevent-podscheduling-optin 
Add API changes to prevent pod scheduling via CSIDriver object
 2026-03-11 03:53:17 +05:30
Kubernetes Prow Robot
69144c9081 Merge pull request #137371 from pohly/dra-bind-claim-panic 
DRA scheduler: fix potential panic when DRABindingConditions are enabled
 2026-03-11 03:03:25 +05:30