v0.5.4

What's Changed

• fix reclaimed-large-pod.yaml in examples by @flpanbin in #600
• fix(borwein): use indicator original value when it cannot be updated by borwein by @csfldf in #610
• refactor(async): support limited async workers by @cheney-lin in #605
• feat(general): audit manager framework by @zzzzhhb in #572
• feat(agent): refactor eviction/reporter healthz rule by @zzzzhhb in #614
• feat(qrm): 1. refine timemonitor until 2. fallback when cpu advisor is unhealthy by @csfldf in #611
• refine and fix checkLabelSelectorOverlap by @luomingmeng in #599
• refine GetSiblingNumaInfo by @luomingmeng in #612
• fix(util): fix deadlock by @cheney-lin in #619
• fix(qrm): always allocate reclaim pool lastly by @csfldf in #595
• feat(lifecycle): emitter cluster health rate by @zzzzhhb in #615
• feat(audit): eventbus metrics by @zzzzhhb in #613
• feat(qrm): support shared_cores with numa_binding by @csfldf in #556
• feat(spd): support resource portrait plugin by @tibberzh in #589
• support share cores with numa binding by @nightmeng in #554
• feat(lifecycle): emitter cluster health rate by @zzzzhhb in #621
• fix(qrm): make ramp up cpus not overlapping with cpus in NUMAs containing NUMA_binding pods by @csfldf in #631

New Contributors

• @flpanbin made their first contribution in #600
• @tibberzh made their first contribution in #589

Full Changelog: v0.5.3...v0.5.4

v0.5.3

What's Changed

• feat(advisor): skip the share pool without cpu request by @nightmeng in #591
• remote reporter plugin and topology adapter support context timeout by @luomingmeng in #567
• refine(util): refine getting reserved by @csfldf in #587
• fix: convert memory by quantity.Value in overcommit controller by @WangZzzhe in #588

Full Changelog: v0.5.2...v0.5.3

v0.5.2

What's Changed

• feat(*): move all dependency for kubelet configurations to util by @waynepeking348 in #582
• fix: ignore not existed cgroup paths by @cheney-lin in #585
• bugfix: fix resource leakage caused by the duplicate sample of rodan … by @WangZzzhe in #586
• refine vpa apply strategy by @luomingmeng in #590

Full Changelog: v0.5.1...v0.5.2

v0.5.1

What's Changed

• feat: Add TransparentMemoryOffloading plugin to trigger memory offloading by @cheney-lin in #549
• feat: adding wbt switch function for nvme devices by @lubinszARM in #564
• feat(qrm): use async worker to set extra cg mem limit to avoid thread stucking by @csfldf in #563
• fix sibling numa set by @luomingmeng in #544
• feat(qrm&advisor): add flag allow_shared_cores_overlap_reclaimed_cores to ListAndWatchResponse from cpu advisor by @csfldf in #569
• use token when uploading codecov results by @waynepeking348 in #571
• add healthz metrics by @zzzzhhb in #570
• kubelet plugin run add fatal error by @luomingmeng in #573
• change gofmt to gofumpt for strict formatting by @waynepeking348 in #574
• refine(proto): refine make generate in makefile by @csfldf in #575
• feat: support dynamic config for reclaimed_cores overlaping shared_cores by @cheney-lin in #576
• feat(sysadvisor): update default enabled plugins by @nightmeng in #580
• refine metric-fetcher to support specified collecting intervals by @waynepeking348 in #581

Full Changelog: v0.5.0...v0.5.1

v0.5.0

What's Changed

• fix(qrm): fix panic when handleAdvisorCPUSetMems got nil allocationInfo, add robustness by @csfldf in #440
• add logs by @cheney-lin in #441
• remove orm from /agent/resourcemanager into /agent by @WangZzzhe in #442
• fix reclaim pool entry of non binding numas in calculation result by @sun-yuliang in #439
• bugfix: request summary from podsEndpoint when enable kubelet secure port by @WangZzzhe in #438
• refactor(sysadvisor): consider per numa memory reserved by @cheney-lin in #443
• refine the implementation for qos configurations by @waynepeking348 in #444
• fix(metric): numa memory stats by @cheney-lin in #446
• feat(ORM): outofband topology manager by @WangZzzhe in #435
• refactor(eviction): support to config eviciton tolerance duration by @cheney-lin in #450
• bugfix(metric): fix metric expire by @nightmeng in #454
• fix eviction manager grpc connection leak by @luomingmeng in #455
• feat(async_work): add async work duration and panic metrics by @csfldf in #458
• feat(eviction): add qos tag by @nightmeng in #457
• fix qos enhancement by @luomingmeng in #459
• fix(metaserver): disalbe metrics expire check by default by @nightmeng in #461
• feat: adding wbt switch function by @lubinszARM in #449
• feat(sysadvisor): fix sysadvisor lw stream leak by @nightmeng in #464
• refactor(sysadvisor): add metric and tag by @cheney-lin in #463
• feat(eviction): add rootfs eviction options by @nightmeng in #465
• refactor(sysadvisor): refine boundType updating logic by @cheney-lin in #462
• fix(eviction): fix rootfs eviction protection by @nightmeng in #467
• refactor(sysadvisor): add idle status for region by @cheney-lin in #468
• feat(eviction): optimize rootfs eviction logs and metrics by @nightmeng in #471
• chore(agent): monitor agent start by @zzzzhhb in #475
• refine code owners for sysadvisor by @waynepeking348 in #485
• fix(vpa webhook): fix vpa webhook could not update request for pod without setting limit by @ZhangsongLee in #484
• feat(cgroup): setting memory.low/min in ApplyMemory() by @lubinszARM in #476
• spd api support extended indicator by @luomingmeng in #486
• feat(sysadvisor): support numa memory balance by @zzzzhhb in #429
• feat: support node metrics reporter by @cheney-lin in #474
• refine(qrm): calculate pool size with much accuracy by @csfldf in #445
• refine implementation for metaserver configurations by @waynepeking348 in #479
• add cgroup metric fetcher in metaserver by @waynepeking348 in #478
• support creating new pools by @nightmeng in #491
• feat(qrm): use move_pages syscall to migrate numa memory by @yadzhang in #477
• fix(qrm): fix advisor drop cache handler missing nbytes parameter by @csfldf in #500
• kubelet reporter plugin support shared_cores with numa_binding pod by @luomingmeng in #473
• feat(meatserver): use multiple subsystems to detect cgroup path by @zzzzhhb in #497
• chore(*): add logs by @cheney-lin in #501
• feat(qrm): do evict and balance at the same time by @zzzzhhb in #503
• customized plugin manager to skipping running if no handles and registered by @waynepeking348 in #490
• refine(qrm): add metrics for plugin start failed by @csfldf in #505
• feat(qrm): refine healthz by @zzzzhhb in #482
• support dedicated image fs disk by @nightmeng in #502
• fix(sysadvisor): numa balancer miss eviction by @zzzzhhb in #506
• add cnr reporter update cnr cost metrics and log by @luomingmeng in #507
• feat(resource-recommend): add resource recommend controller by @ZhangsongLee in #483
• spd support extended indicator sdk and extended baseline by @luomingmeng in #495
• cnr reporter support merge siblings and machine info provide sibling numa info by @luomingmeng in #511
• renaming transformed logic for pod informer by @waynepeking348 in #512
• fix(sysadvisor): get wrong max latency numa by @zzzzhhb in #513
• refactor(sysadvisor): make target indicator configurable by @cheney-lin in #452
• feat(sysadvisor): optimize advisor qosaware update logic by @nightmeng in #514
• fix invalid type transformation for BorweinInferenceResults by @justadogistaken in #508
• Malachite adapt to network metrics by @waynepeking348 in #510
• chore(*): add metrics and logs by @cheney-lin in #517
• feat(sysadvisor): fix the concurrent read-write issue in metacache by @nightmeng in #509
• bugfix: marshal checkpoint by json by @WangZzzhe in #466
• fix ServiceExtendedIndicator return nil object by @luomingmeng in #515
• chore(*): add metric by @cheney-lin in #521
• feat(eviction): add eviction pod metric labels by @nightmeng in #523
• spd: get qos Level by pod and its annotations by @cheney-lin in #522
• feat(advisor): fix cpu advisor calc checking logic by @nightmeng in #526
• skip memory estimation for reclaimed cores pod by @luomingmeng in #524
• refactor(sysadvisor): refine memory guard by @cheney-lin in #499
• feat(kcmas): enable grouped aggregate metric by @zzzzhhb in #530
• reduce the frequency of spd manager try to get unused spd from remote api-server by @luomingmeng in #532
• ignore error if coverage uploading failed by @waynepeking348 in #534
• feat(sysadvisor): add healthz check by @zzzzhhb in #520
• Revert "refactor(sysadvisor): refine memory guard" by @zzzzhhb in #535
• feat: adding support to io.cost setting by @lubinszARM in #496
• report numa sibling and memory bandwidth by @luomingmeng in #533
• feat: adding support to io.weight setting by @lubinszARM in #494
• fix(sysadvisor): fix metacache concurrent read/write by @nightmeng in #538
• io: adding fake functions for other platforms by @lubinszARM in #540
• chore(sysadvisor): disable some healthcheck rule by @zzzzhhb in #539
• chore(sysadvisor): refine some healthcheck rule by @zzzzhhb in #543
• util(cgroup): relax constraints by @cheney-lin in #545
• [wip] refine testing cases to make sure all of them can be tested parallelly by @waynepeking348 in #537
• refactor(sysadvisor): support dynamic enable memory guard by @cheney-lin in #536
• add codecov uploading secret tokens by @waynepeking348 in #552
• feat(qrm): add ResourceAnnotationKeyNICNetNSName by @csfldf ...

v0.4.2

What's Changed

• util(cgroup): relax constraints by @cheney-lin in #545
• [wip] refine testing cases to make sure all of them can be tested parallelly by @waynepeking348 in #537
• refactor(sysadvisor): support dynamic enable memory guard by @cheney-lin in #536
• add codecov uploading secret tokens by @waynepeking348 in #552
• feat(qrm): add ResourceAnnotationKeyNICNetNSName by @csfldf in #546
• spd support cnc cache by @luomingmeng in #542
• feat(reporter): validate specific resources in pod resources response by @caohe in #541
• feat(ORM): add orm podResources server by @WangZzzhe in #453
• fix MergeAttributes sort zone attribute by @luomingmeng in #557
• doc(*): add proposal for enhance orm by nri by @Airren in #525

New Contributors

• @Airren made their first contribution in #525

Full Changelog: v0.4.1...v0.4.2

v0.3.5

What's Changed

• cherry pick of #541: feat(reporter): validate specific resources in pod resources response by @caohe in #555

Full Changelog: v0.3.4...v0.3.5

v0.4.1

What's Changed

• fix(qrm): fix panic when handleAdvisorCPUSetMems got nil allocationInfo, add robustness by @csfldf in #440
• add logs by @cheney-lin in #441
• remove orm from /agent/resourcemanager into /agent by @WangZzzhe in #442
• fix reclaim pool entry of non binding numas in calculation result by @sun-yuliang in #439
• bugfix: request summary from podsEndpoint when enable kubelet secure port by @WangZzzhe in #438
• refactor(sysadvisor): consider per numa memory reserved by @cheney-lin in #443
• refine the implementation for qos configurations by @waynepeking348 in #444
• fix(metric): numa memory stats by @cheney-lin in #446
• feat(ORM): outofband topology manager by @WangZzzhe in #435
• refactor(eviction): support to config eviciton tolerance duration by @cheney-lin in #450
• bugfix(metric): fix metric expire by @nightmeng in #454
• fix eviction manager grpc connection leak by @luomingmeng in #455
• feat(async_work): add async work duration and panic metrics by @csfldf in #458
• feat(eviction): add qos tag by @nightmeng in #457
• fix qos enhancement by @luomingmeng in #459
• fix(metaserver): disalbe metrics expire check by default by @nightmeng in #461
• feat: adding wbt switch function by @lubinszARM in #449
• feat(sysadvisor): fix sysadvisor lw stream leak by @nightmeng in #464
• refactor(sysadvisor): add metric and tag by @cheney-lin in #463
• feat(eviction): add rootfs eviction options by @nightmeng in #465
• refactor(sysadvisor): refine boundType updating logic by @cheney-lin in #462
• fix(eviction): fix rootfs eviction protection by @nightmeng in #467
• refactor(sysadvisor): add idle status for region by @cheney-lin in #468
• feat(eviction): optimize rootfs eviction logs and metrics by @nightmeng in #471
• chore(agent): monitor agent start by @zzzzhhb in #475
• refine code owners for sysadvisor by @waynepeking348 in #485
• fix(vpa webhook): fix vpa webhook could not update request for pod without setting limit by @ZhangsongLee in #484
• feat(cgroup): setting memory.low/min in ApplyMemory() by @lubinszARM in #476
• spd api support extended indicator by @luomingmeng in #486
• feat(sysadvisor): support numa memory balance by @zzzzhhb in #429
• feat: support node metrics reporter by @cheney-lin in #474
• refine(qrm): calculate pool size with much accuracy by @csfldf in #445
• refine implementation for metaserver configurations by @waynepeking348 in #479
• add cgroup metric fetcher in metaserver by @waynepeking348 in #478
• support creating new pools by @nightmeng in #491
• feat(qrm): use move_pages syscall to migrate numa memory by @yadzhang in #477
• fix(qrm): fix advisor drop cache handler missing nbytes parameter by @csfldf in #500
• kubelet reporter plugin support shared_cores with numa_binding pod by @luomingmeng in #473
• feat(meatserver): use multiple subsystems to detect cgroup path by @zzzzhhb in #497
• chore(*): add logs by @cheney-lin in #501
• feat(qrm): do evict and balance at the same time by @zzzzhhb in #503
• customized plugin manager to skipping running if no handles and registered by @waynepeking348 in #490
• refine(qrm): add metrics for plugin start failed by @csfldf in #505
• feat(qrm): refine healthz by @zzzzhhb in #482
• support dedicated image fs disk by @nightmeng in #502
• fix(sysadvisor): numa balancer miss eviction by @zzzzhhb in #506
• add cnr reporter update cnr cost metrics and log by @luomingmeng in #507
• feat(resource-recommend): add resource recommend controller by @ZhangsongLee in #483
• spd support extended indicator sdk and extended baseline by @luomingmeng in #495
• cnr reporter support merge siblings and machine info provide sibling numa info by @luomingmeng in #511
• renaming transformed logic for pod informer by @waynepeking348 in #512
• fix(sysadvisor): get wrong max latency numa by @zzzzhhb in #513
• refactor(sysadvisor): make target indicator configurable by @cheney-lin in #452
• feat(sysadvisor): optimize advisor qosaware update logic by @nightmeng in #514
• fix invalid type transformation for BorweinInferenceResults by @justadogistaken in #508
• Malachite adapt to network metrics by @waynepeking348 in #510
• chore(*): add metrics and logs by @cheney-lin in #517
• feat(sysadvisor): fix the concurrent read-write issue in metacache by @nightmeng in #509
• bugfix: marshal checkpoint by json by @WangZzzhe in #466
• fix ServiceExtendedIndicator return nil object by @luomingmeng in #515
• chore(*): add metric by @cheney-lin in #521
• feat(eviction): add eviction pod metric labels by @nightmeng in #523
• spd: get qos Level by pod and its annotations by @cheney-lin in #522
• feat(advisor): fix cpu advisor calc checking logic by @nightmeng in #526
• skip memory estimation for reclaimed cores pod by @luomingmeng in #524
• refactor(sysadvisor): refine memory guard by @cheney-lin in #499
• feat(kcmas): enable grouped aggregate metric by @zzzzhhb in #530
• reduce the frequency of spd manager try to get unused spd from remote api-server by @luomingmeng in #532
• ignore error if coverage uploading failed by @waynepeking348 in #534
• feat(sysadvisor): add healthz check by @zzzzhhb in #520
• Revert "refactor(sysadvisor): refine memory guard" by @zzzzhhb in #535
• feat: adding support to io.cost setting by @lubinszARM in #496
• report numa sibling and memory bandwidth by @luomingmeng in #533
• feat: adding support to io.weight setting by @lubinszARM in #494
• fix(sysadvisor): fix metacache concurrent read/write by @nightmeng in #538
• io: adding fake functions for other platforms by @lubinszARM in #540
• chore(sysadvisor): disable some healthcheck rule by @zzzzhhb in #539
• chore(sysadvisor): refine some healthcheck rule by @zzzzhhb in #543

New Contributors

• @ZhangsongLee made their first contribution in #484
• @yadzhang made their first contribution in #477

Full Changelog: v0.4.0...v0.4.1

v0.4.0

What's Changed

• feat(sysadvisor): add indicator cpu usage ratio in rama policy for share region by @sun-yuliang in #200
• feat(spd): make spd indicator plugins configurable by @cheney-lin in #204
• adding support to periodicalhandler manager in memory plugin by @lubinszARM in #203
• fix(sysadvisor): wait for metrics fetcher synced once when starting up by @cheney-lin in #201
• refine malachite metrics to unified units by @waynepeking348 in #206
• chores(eviction): rename metric by @zzzzhhb in #207
• fix(sysadvisor): replace memory avaiable metric by @cheney-lin in #208
• refactor(sysadvisor): set min reserved memory limit for reclaim group by @cheney-lin in #211
• enhance cpu advisor robustness by @sun-yuliang in #209
• refactor spd fetcher to async get spd from remote by @luomingmeng in #210
• feat(qrm): judge shoudSharedCoresRampUp to avoid vibrating in takeover procedure by @csfldf in #225
• fix advisor issues by @sun-yuliang in #226
• refine kcmas to support multiple host ips by @waynepeking348 in #218
• fix(qrm): still try to ramp up if specified pool entry doesn't exist by @csfldf in #228
• fix lifecycle agent pod ready check by @luomingmeng in #231
• fix bugs for justifying isolation region names by @waynepeking348 in #230
• fix bugs for isolation ownerpool setting and support multiple containers to share region by @waynepeking348 in #232
• some rand functions are deprecated by @zhy76 in #233
• fix(qrm): mem plugin pushes containers to mem adviosr at starting by @csfldf in #235
• Fix reporter plugin re-register data loss by @luomingmeng in #234
• fix: fix the bug caused by invalid numanode value on VMs by @smart2003 in #227
• fix invalid network reservation for network plugin by @waynepeking348 in #236
• fallback to canonical when reclaim disabled; add buffer ratio for cpu estimation by @sun-yuliang in #238
• chore(doc): update enhanced k8s installation guide for head by @pendoragon in #244
• fix and complement region tags by @sun-yuliang in #245
• refactor(sysadvisor): refine log format by @cheney-lin in #243
• fix kcmas use label selector failed by @luomingmeng in #248
• add more metrics for kcmas and refine performance for read requests by @waynepeking348 in #239
• fix cnc update by @luomingmeng in #247
• fix(util): return nil when samples pool is not filled by @cheney-lin in #249
• fix memory leak by close all response body by @waynepeking348 in #251
• getZoneAllocations filter out failed and succeeded pod by @luomingmeng in #254
• feat(eviction): support container killer by @zzzzhhb in #257
• fix(sysadvisor): assemble duplicate block by @cheney-lin in #259
• implement io qrm plugin by @csfldf in #258
• fix(sysadvisor): fix invalid cpulimit by @cheney-lin in #262
• GetTopologyZones return error if list pod resources empty by @luomingmeng in #261
• resource eviction ThresholdGetter support threshold not set by @luomingmeng in #264
• fix owner pool name for isolated container by @waynepeking348 in #263
• feat(qrm): support option which prefers to use kubelet reserved resource configuration to katalyst reserved resource configuration by @csfldf in #260
• fix(controller): fix cnr residue when node deletion by @chenxi-seu in #250
• fix set-pool-regions in sysadvisor: always use origin owner pool name by @waynepeking348 in #267
• always set isolated container name as region name by @waynepeking348 in #268
• fix KatalystCustomConfigTargetHandler to avoid cnc being cleared when cache no synced in restart by @luomingmeng in #265
• fix(sysadvisor): set share region owner pool as container origin owne… by @cheney-lin in #269
• refine isolation threshold judgement logic by @waynepeking348 in #271
• fix(sysadvisor): avoid gc living containers' original owner pools by @sun-yuliang in #272
• fix(qrm): clean pool concerning owner pool name instead of specified pool by @csfldf in #274
• refactor: refine memory log format by @cheney-lin in #273
• hint got from extra state file must be subset of available NUMAs by @csfldf in #276
• refactor(sysadvisor): bind reclaim_cores contianers to non-exclusive … by @cheney-lin in #275
• feat(sysadvisor): list containers before updating by @cheney-lin in #270
• always set the min-value for isolation bound as 1 core by @waynepeking348 in #277
• feat(sysadvisor): support numa aware memory headroom policy by @cheney-lin in #278
• PluginRegistrationWrapper inherit GenericPlugin directly by @luomingmeng in #280
• fix(sysadvisor): consider pod reclaim disabled by @cheney-lin in #282
• add metrics for failed events by @cheney-lin in #281
• Add device io stat metrics by @justadogistaken in #283
• enhancement(eviction): system load configuration by @zzzzhhb in #287
• fix(sysadvisor): check pod number for dedicated region by @cheney-lin in #288
• fix region info clone no clone region status by @luomingmeng in #289
• fix node_advisor_knob_status tag by @luomingmeng in #290
• feat(sysadvisor): set a relatively large threshold for early cache re… by @cheney-lin in #291
• feat(auth): auth framework by @zzzzhhb in #279
• fix kcmas precise when some metric with small value by @luomingmeng in #292
• feat(sysadvisor): bind memset to no pressure NUMAs by @cheney-lin in #293
• feat(sysadvisor): add metrics for headroom by @cheney-lin in #294
• enhancement(eviction): numa pressure support victim minimum usage thr… by @zzzzhhb in #296
• feat(sysadvisor): modify numa aware memory headroom policy by @LuyaoZhong in #302
• fix: code comment linked address by @wanglei4687 in #304
• Updating cpu_request of containerInfo in periodicWork of MetaCachePlugin by @justadogistaken in #286
• chore: ignore test coverage profile files by @wanglei4687 in #308
• move locking file metrics name to const and renaming by @waynepeking348 in #310
• feat: monitoring accuracy and latency of reported information in KCNR by @zhy76 in #237
• fix: character type by @wanglei4687 in #309
• Fix calculation of rate metrics by @justadogistaken in #306
• wip: support to select service-profiling workload by selector by @waynepeking348 in #311
• enhancement(sysadvisor): Prevents NAN from being generated due to a division by zero error by @zhy76 in #301
• enhancement(auth): basic auth use kcc as backend by @zzzzhhb in https://github.com/kubewharf/katalyst-core...

v0.3.4

What's Changed

• support report numa count to kcnr by @luomingmeng in #413

Full Changelog: v0.3.3...v0.3.4