v0.7.0
What's Changed
- refactor extended resource with
kubernetes.io
namespace by @zwzhang0107 in #350 - add multi-hierarchy-elastic-quota-management by @buptcozy in #398
- ci: rolling back to golangci-lint 1.47.3 by @jasonliu747 in #449
- adjusted the description of elastic quota by @hormes in #448
- koord-scheduler: adapt to different cpuset scheduling protocols by @ZiMengSheng in #452
- apis: add allocateOnce in reservation by @saintube in #447
- rename prefix of extended resource by @zwzhang0107 in #453
- koord-descheduler: complete configuration by @eahydra in #456
- ci: extend golangci-lint timeout to 10m by @jasonliu747 in #461
- fix: use uint32 to record process id by @jasonliu747 in #455
- ci: remove cache when setup go by @jasonliu747 in #463
- ci: add codeql analysis in workflows by @jasonliu747 in #240
- feat(deps): bump github.com/prometheus/client_golang from 1.12.2 to 1.13.0 by @dependabot in #454
- ADOPTERS: add iQIYI by @wangxiaoq in #465
- koord-scheduler: export NodeNUMAInfoCache and getAvailableCPUs by @ZiMengSheng in #469
- koord-scheduler: abstract GetPodQoSClass by @ZiMengSheng in #471
- add extended resource schedule plugin for compatibility by @zwzhang0107 in #472
- set kubernetes.io/batch-cpu as milli-core by @zwzhang0107 in #473
- koord-descheduler: init evictor plugin first by @eahydra in #458
- feature: support read args from KubeletConfiguration by @ZYecho in #470
- koord-scheduler: fix stale reservation status by @saintube in #474
- bugfix: fix exec exit with 126 by @ZYecho in #446
- feat(deps): bump github.com/spf13/cobra from 1.1.3 to 1.5.0 by @dependabot in #279
- koord-scheduler: fix typo by @jasonliu747 in #476
- koord-descheduler: add tests for framework runtime by @eahydra in #475
- ci: remove codeql from workflows by @jasonliu747 in #479
- docs: format markdown by @jasonliu747 in #480
- ci: add codeql in pipeline workflows by @jasonliu747 in #481
- koord-scheduler: reservation support allocateOnce by @saintube in #478
- feat(deps): bump go.uber.org/atomic from 1.9.0 to 1.10.0 by @dependabot in #483
- koord-runtime-proxy: regard empty string for CpusetCpus and CpusetMem… by @honpey in #484
- refactor batch node resource update by @zwzhang0107 in #485
- ci: use fast-test to skip redundant steps in workflows by @jasonliu747 in #486
- mv container extend resource to independent file by @zwzhang0107 in #488
- koord-descheduler: enable allocateOnce when creating Reservation by @eahydra in #494
- support node extended resource plugin by @zwzhang0107 in #491
- fix metric-expire-seconds usage by @huiwq1990 in #495
- adopters: add Quwan by @zshmmm in #497
- koord-descheduler: fix incorrect handling of Reservation's status by @eahydra in #500
- koord-scheduler: fix succeeded reservation GC and unschedulable retry by @saintube in #501
- koord-runtime-proxy: support inject env in proxy by @ZYecho in #503
- refactor: move GeneratePodPatch under util package by @jasonliu747 in #504
- fix: correct misspells in pkg/koordlet/metriccache by @huiwq1990 in #506
- koord-runtime-proxy: add more test cases by @honpey in #511
- apis: add DeviceAllocations and methods by @jasonliu747 in #513
- koord-scheduler: fix succeeded reservation in scheduler cache by @saintube in #516
- koord-descheduler: stop migration when reserved on the same node by @eahydra in #519
- koord-descheduler: support evict flow control to ensure availability by @eahydra in #520
- feat: not update status field when create NodeMetric by @huiwq1990 in #514
- koord-manager: record configmap and pod qos unmarshal failed events by @songtao98 in #518
- koord-scheduler: refactor NodeNUMAResource plugin by @eahydra in #507
- koord-descheduler: control the maximum number of migrations to ensure availability by @eahydra in #522
- apis: update Reservation kubebuilder:printcolumn comments by @eahydra in #524
- feature: support inject NVIDIA_VISIBLE_DEVICES env by @ZYecho in #517
- fix: Walkfunc should check the err by @j4ckstraw in #460
- runtime-hooks: add auto register hooks by @songtao98 in #505
- qos grep control extension by @zwzhang0107 in #529
- Add runtime hooks stage config flag by @songtao98 in #526
- docs: update fine grained device proposal by @jasonliu747 in #531
- chore: update tools version by @jasonliu747 in #535
- docs: reformat adopters list by @jasonliu747 in #537
- add quotaTreeWrapper and elasticQuota API by @xulinfei1996 in #515
- koord-scheduler: optimize the process after reservation scheduling fails by @eahydra in #532
- fix bug: modify unit test file path to temp dir by @songtao98 in #540
- ci: support aliyun hangzhou registry by @jasonliu747 in #542
- koord-scheduler: add coscheduling plugin args by @Wenshiqi222 in #538
- Update 20220629-fine-grained-device-scheduling.md by @hormes in #543
- koord-scheduler: export getDefaultNUMAAllocateStrategy by @ZiMengSheng in #547
- koord-scheduler: support API Services by @eahydra in #545
- chore: downgrade CUDA version to 11.2.2 by @jasonliu747 in #548
- koord-scheduler: support custom SharedLister by @ZiMengSheng in #549
- fix container cgroup comment by @huiwq1990 in #553
- koord-scheduler: fix CPUTopologyManager get&update race condition by @eahydra in #555
- koord-scheduler: CPU allocate algorithm supports maxRefCount by @ZiMengSheng in #551
- apis: update Device CRD fields by @jasonliu747 in #554
- koord-scheduler: fix coSchedulingArgs json name by @Wenshiqi222 in #558
- support Reservation select order by @eahydra in #552
- koord-descheduler: fix failed evict with StatefulSet Pod by @eahydra in #559
- koord-descheduler: fix UT conflict by @eahydra in #563
- koord-scheduler: improve reservation compatibility by @saintube in #550
- kood-scheduler: fix Reservation Score divide by zero by @eahydra in #564
- koord-scheduler: fix Reservation Score divide by zero with some empty… by @eahydra in #565
- chore: add comments about CUDA version in dockerfile by @jasonliu747 in #567
- Fix concurrency conflicts in states informer by @zwzhang0107 in #568
- cpu suppress: only limits cpuset at besteffot container level if kube… by @zwzhang0107 in #556
- koord-runtime-proxy: add more tests by @honpey in #575
- ci: ignore apis & pkg/client in codecov by @jasonliu747 in #576
- test: add Test_DumpJSON() by @jasonliu747 in #578
- koord-runtime-proxy: Introduce envs injecting under contianerd scenario by @honpey in #544
- feature: report gpu resource by @ZYecho in #410
- docs: Update details about compatible existing GPU Pods by @eahydra in #571
- koodlet: fix gpu reporter build tag by @zwzhang0107 in #580
- feat: get kubelet pid through kubeletEndpoint port by @dongzx666 in #521
- test: add tests for reservation utils by @saintube in #582
- add groupQuotaManager by @xulinfei1996 in #541
- koord-scheduler: implement enhanced coscheduling by @Wenshiqi222 in #510
- koodlet: fix resource in cri proxy response maybe nil for kubelet 1.18 by @zwzhang0107 in #587
- add extension plugin framework for nodeslo by @zwzhang0107 in #579
- koord-scheduler: fix Reserve Pod concurrent map write fatal error by @eahydra in #588
- koord-scheduler: fix Reservation normalize score by @eahydra in #589
- koord-scheduler: support device(GPU/RDMA/FPGA) sharing feature by @jasonliu747 in #502
- test: add ut for pkg/util/config.go by @jasonliu747 in #591
- koordlet: startup flag replace int with time.Duration in reporter by @jasonliu747 in #593
- koord-scheduler: fix DeviceShare UT by @eahydra in #594
- QoS Manager Framework by @stormgbs in #590
- koordlet: report kubelet reserved cpus by @eahydra in #592
- koord-scheduler: support kubelet reserved cpus by @eahydra in #595
- chore: add aliyun hangzhou registry when publish release by @jasonliu747 in #598
- ci: add aliyun hangzhou registry credential by @jasonliu747 in #599
- feat(deps): bump github.com/evanphx/json-patch from 4.11.0+incompatible to 5.6.0+incompatible by @dependabot in #573
- Revert "QoS Manager Framework" by @jasonliu747 in #601
- chore: remove deprecated io/ioutil by @guilhem in #605
- fix: fix panic when NodeTopo is nil by @guilhem in #606
- koord-scheduler: support debug scores by @eahydra in #604
- feat(deps): bump github.com/lluissm/license-header-checker from 1.3 to 1.4 by @jasonliu747 in #611
- koord-descheduler: support plugins customize MigrationJob via context by @eahydra in #609
- feat(deps): bump github.com/docker/docker from 20.10.17+incompatible to 20.10.18+incompatible by @dependabot in #608
- koordlet: change evict to v1beta1 for compatible with 1.21- by @zwzhang0107 in #613
- koordlet: fix incompatibility with kubelet v1.18 flags by @eahydra in #612
- chore: update license checker repo by @eahydra in #616
- koord-scheduler: add elastic quota plugin by @xulinfei1996 in #584
- feat(deps): bump github.com/NVIDIA/go-nvml by @jasonliu747 in #615
- koord-scheduler: fix quota ut failed by @xulinfei1996 in #620
- add restful api for coscheduling by @Wenshiqi222 in #603
- koord-scheduler: optimize ElasticQuota UT by @xulinfei1996 in #625
- koord-scheduler:refine gang restful api for coscheduling by @buptcozy in #624
- koord-scheduler: support score schedulingPhase hook and multiple hook by @ZiMengSheng in #627
- koord-scheduler: add quota debug restful api by @buptcozy in #628
- koordlet: fix nvml init error by @eahydra in #635
- koordlet: fix failed update existing Device CRD object by @eahydra in #634
- Update and rename support-request.md to question.md by @hormes in #623
- fix coScheduling ut by @Wenshiqi222 in #633
- koord-scheduler: add device scheduling debug api by @buptcozy in #637
- koord-scheduler: change NodeNUMAResource API endpoints by @eahydra in #638
- koord-scheduler: fix schedulingPhaseHook registerd log nil pointer bug by @ZiMengSheng in #641
- koordlet: cpu suppress check whether nodeslo enable by @jasonliu747 in #640
- koord-scheduler: optimize ElasticQuotaPluginUT by @xulinfei1996 in #639
- fix: panic on node nil by @guilhem in #631
- koord-scheduler: fix quota panic if missing QuotaInfo by @eahydra in #642
- bugfix: not report when no gpu device by @ZYecho in #636
- fix CoScheduling ut:wg count negative bug by @Wenshiqi222 in #643
- koord-scheduler: fix ElasticQuota panic when init by @xulinfei1996 in #645
- chore: change ElasticQuota log level by @eahydra in #646
- koordlet: fix memory evict for nil podMetric by @saintube in #648
- just tidy up some logic by @lucming in #644
- chore: update resource names in Fine-grained Device Scheduling proposal by @eahydra in #649
- chore: rename Gang Scheduling proposal by @eahydra in #647
- koord-manager: add elastic quota webhook by @xulinfei1996 in #585
- koord-manager: revise node batch resource update by @saintube in #650
- koord-scheduler: fix ElasticQuota controller panic when update annotations by @eahydra in #658
- koord-scheduler: fix PodGroupController panic when init by @eahydra in #659
New Contributors
- @wangxiaoq made their first contribution in #465
- @huiwq1990 made their first contribution in #495
- @zshmmm made their first contribution in #497
- @songtao98 made their first contribution in #518
- @xulinfei1996 made their first contribution in #515
- @dongzx666 made their first contribution in #521
- @guilhem made their first contribution in #605
- @lucming made their first contribution in #644
Full Changelog: v0.6.2...v0.7.0