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 #305
• increase the log level of getting object failed when check internal metric by @luomingmeng in #313
• refine kcmas and support aggregating logic in local cache by @waynepeking348 in #312
• enhancement(kcmas): emit metric latency by @zzzzhhb in #315
• enhancement(sysadvisor): skip the containers if cache usage less then… by @cheney-lin in #297
• fix(eviction): parse container id when stop container by @zzzzhhb in #316
• decode metric list according to metric name whether is aggregator metric by @luomingmeng in #317
• fix window seconds by @luomingmeng in #319
• feat(sysadvisor): implement borwein algorithm by @csfldf in #307
• feature(kcmas): scrape with basic auth by @zzzzhhb in #318
• check coreCount within AggregateCoreMetric by @justadogistaken in #320
• feat(advisor): support disable registered model result fetcher by @csfldf in #321
• Fix problem of divided by 0 in memory bandwidth calculation by @justadogistaken in #322
• feat(http): non-strict authentication by @zzzzhhb in #323
• feat(advisor): borwein fix & enhancement by @csfldf in #326
• kcams add collector scrape out of date metrics by @luomingmeng in #328
• feat(overcommit): add node overcommit webhook by @WangZzzhe in #229
• feat(overcommit): add nodeovercommit controller by @WangZzzhe in #246
• fix(eviction): panic when get evict pods by @zzzzhhb in #333
• add pod_cpu_cpi in custom metric by @justadogistaken in #332
• add standard framework for service discovery by @waynepeking348 in #325
• fix: update the ResourceName of reclaim pod in the example directory by @Lan-ce-lot in #330
• feat(eviction): system free memory threshold minimum by @zzzzhhb in #327
• perf(kcmas): performance optimization by @zzzzhhb in #329
• support to disable isolation when exceeding total capacities by @waynepeking348 in #334
• chore(generic): add kube-aggregator client by @zzzzhhb in #337
• feat(metaserver): add node cpu usage ratio metric by @cheney-lin in #339
• misc(doc): Add community meeting & group by @pendoragon in #340
• [WIP]feat(kcmas):mock collector by @zzzzhhb in #335
• spd support baseline ratio by @luomingmeng in #344
• performance(kcmas): lock optimize by @zzzzhhb in #343
• optimize(metric): optimize calculation of counter rate by @justadogistaken in #342
• fix(kcmas): deadlock by @zzzzhhb in #346
• enhancement(sysadvisor): export region reclaimable info as non-reclaim when dedicated_cores can not be co-located with reclaimed_cores by @cheney-lin in #345
• fix PodEnableReclaim pod spd not found by @luomingmeng in #350
• enhancement(kcmas): object index by @zzzzhhb in #349
• add more eviction messages about threshold and observed values by @waynepeking348 in #352
• Print out NIC info upon qrm network plugin initialization by @smart2003 in #354
• enhancement(kcmas): bucket object store by @zzzzhhb in #353
• fix(kcmas): nil pointer by @zzzzhhb in #357
• doc: some fixes to examples and proposals for the reclaimed core pod cannot be scheduled by @Lan-ce-lot in #351
• perf(kcmas): remove high frequency monitor by @zzzzhhb in #360
• perf(kcmas): don't purge useless metric meta by @zzzzhhb in #361
• refine numa-aware memory headroom policy by @LuyaoZhong in #359
• enhancement(sysadvisor): update region info atomically by @cheney-lin in #358
• fix bugs for creating isolation region multiple times by @waynepeking348 in #363
• perf(eviction): collector optimize by @zzzzhhb in #364
• feat(reporter-plugin): support report rdma topology by @fjding in #314
• enhancement(spd): get spd by workload's name by @cheney-lin in #369
• misc: fix meeting link by @pendoragon in #372
• adapt to the latest version of katalyst api by @waynepeking348 in #373
• feat(sockmem-plugin): unified solution for TCP memory limitation #365 by @lubinszARM in #366
• updating counterDelta func by @justadogistaken in #374
• feat(qrm): support register handler after group started by @csfldf in #376
• enhancement(spd): set default baselint ratio by qos level by @cheney-lin in #375
• refine some comments and coding styles by @waynepeking348 in #377
• fix(spd): remove interface by @cheney-lin in #378
• node updater support patch node status by @luomingmeng in #379
• fix unit test race in sockmem plugin by @lubinszARM in #380
• add metrics for eviction and handlers by @waynepeking348 in #382
• fix(sysadvisor): always set regionInfo by @cheney-lin in #384
• fix bugs for not updating bound-type for sysadvisor regions by @waynepeking348 in #385
• fix(spd): parseBaselineCoefficient returns error if BaselineCoefficient is empty by @cheney-lin in #388
• feat(kcmas): multiple index keys by @zzzzhhb in #386
• feat(sysadvisor): emit bound type code for knob status by @zzzzhhb in #391
• fix(kcmas): missing store instance in same node by @zzzzhhb in #392
• feat(qrm): Support for OOM priority as a QoS enhancement by @y-ykcir in #371
• fix(spd): baseline checking by @cheney-lin in #393
• feat(kcmas): multiple metric labels by @zzzzhhb in #390
• feat(qrm): adapt drop cache operation with cgv2 by @csfldf in #394
• refactor cpu pressure eviction framework by @luomingmeng in #395
• support custom GetCPUQuantity and GetMemoryQuantity func by @luomingmeng in #396
• suppression eviction add PodUniqKeyCmpFunc to make sure the sort result is stable by @luomingmeng in #397
• Updating malachite metric for supporting model inference by @justadogistaken in #368
• feat(qos): support enhancement updater to decorate enhancement kvs according to pod annotations by @csfldf in #400
• feat(qrm): support getting specified pool name in QoS conf by @csfldf in #401
• chore(eviction): set default minimum system memory watermark to zero by @zzzzhhb in #402
• feat(qrm): refine enhancement & qosLevel updater by @csfldf in #403
• enhancement(eviction): set skip pools via conf by @zzzzhhb in #404
• feat(util): add interface Empty() to determine whether samples should… by @cheney-lin in #398
• Removing duplicated logging by @justadogistaken in #399
• fix(qrm): use phase pending instead of active to judge if need to ramp up by @csfldf in #407
• fix(spd): fix deadlock by @cheney-lin in #408
• feat(advisor): enable borwein result fetcher independently by @csfldf in #409
• support report numa count to kcnr by @luomingmeng in #410
• Fix/deadlock by @cheney-lin in #411
• feat(advisor): add some metrics for borwein model result fetcher by @csfldf in #412
• isolation support non-exclusive by @luomingmeng in #405
• add timestamp to get feature function by @justadogistaken in #416
• Add owners for katalyst by @waynepeking348 in #417
• feat(tide): support tide nodepool by @xjh1996 in #387
• refine coding styles for katalyst-agent cmd and eviction plugins by @waynepeking348 in #418
• modify numa-aware memory headroom policy to consider about inactive pages by @LuyaoZhong in #383
• enhancement(sysadvisor): introduce minCriticalWatermark to determin m… by @cheney-lin in #415
• refine cpu provision framework and share region support restrict poli… by @luomingmeng in #419
• feat(sysadvisor): Refined NUMA-granularity Memory Management and Control Framework by @Lan-ce-lot in #381
• report ratio of overload container by @justadogistaken in #424
• feat: node resource topology scheduler plugin by @WangZzzhe in #298
• add none policy for headroom and provision by @waynepeking348 in #421
• feat(advisor): add indicator offset metric for borwein controller by @csfldf in #425
• fix race conditions for malachite by @waynepeking348 in #426
• add locking framework to detect potential deadlocks by @waynepeking348 in #427
• fix container metrics by @justadogistaken in #428
• feat(orm): support outofband-resource-manager by @WangZzzhe in #406
• add model ouput to kcmas by @justadogistaken in #423
• feat(eviction): support rootfs eviction by @nightmeng in #422
• fix(tide): fix fields typo by @xjh1996 in #433
• fix(tide): fix fields typo by @xjh1996 in #434
• misc(doc): update enhanced k8s version for v0.4.0 by @pendoragon in #431
• chore(*): release v0.4.0 by @caohe in #437

New Contributors

• @lubinszARM made their first contribution in #203
• @justadogistaken made their first contribution in #283
• @LuyaoZhong made their first contribution in #302
• @wanglei4687 made their first contribution in #304
• @WangZzzhe made their first contribution in #229
• @Lan-ce-lot made their first contribution in #330
• @fjding made their first contribution in #314
• @y-ykcir made their first contribution in #371
• @xjh1996 made their first contribution in #387
• @nightmeng made their first contribution in #422

Full Changelog: v0.3.1...v0.4.0