First release candidate of the Coherent Update Strategy feature.

What's Changed

• Refactor/e2e waiter package by @oleg-kushniriov in #526
• Add Annotations field to PodCliqueScalingGroupConfig with MNNVL propa… by @shmuel-runai in #541
• Refactor/e2e unify k8s clients by @oleg-kushniriov in #529
• Upgrade to kindest/node:v1.34.3 by @renormalize in #544
• fix: initc panic on pod update after termination by @steved in #547
• Refactor/e2e centralize data by @oleg-kushniriov in #546
• feat: add /grove-grep AI agent skill for authoring GREPs by @shayasoolin in #522
• refactor(mnnvl): add group-aware ComputeDomain naming and lifecycle m… by @shmuel-runai in #551
• perf: reduce reconcile CPU on the 1000-pod scale test by @danbar2 in #545
• fix: avoid Docker Hub rate limits using GHCR registry image by @danbar2 in #558
• feat(mnnvl): add group-aware MNNVL injection into PodSpec by @shmuel-runai in #557
• add finalizer when build resource to reduce one patch call to api server by @kangclzjc in #477
• docs: use --version flag for helm install/upgrade by @yankay in #563
• docs: document immutable operator ConfigMap by @yankay in #564
• refactor(mnnvl): remove auto-MNNVL mutating webhook, switch to opt-in by @shmuel-runai in #559
• Update code owner list by @sanjaychatterjee in #569
• multi topology implementation by @enoodle in #496
• docs: add helm template / GitOps installation guidance by @yankay in #572
• refactor(mnnvl): replace two-annotation model with single mnnvl-group annotation by @shmuel-runai in #574
• fix: cascade-delete PCS/PCSG/PCLQ via Kubernetes GC by @danbar2 in #556
• test(mnnvl): rewrite E2E tests for single-annotation model by @shmuel-runai in #583
• Add repo-managed git hooks for local validation by @enoodle in #555
• feat: add grove-user-guide AI agent skill for authoring user docs by @shmuel-runai in #589
• fix: propagate PodCliqueSet annotations to PodGang by @danbar2 in #573
• bound UpdateProgress status payload by @oleg-kushniriov in #576
• make linear complexity instead of embedded loops by @oleg-kushniriov in #579
• Fix decode test error for operator config defaults by @xianlubird in #581
• Scale Test CI by @shayasoolin in #575
• Fix gang termination when scheduled replicas regress below MinAvailable by @danbar2 in #561
• Fix KWOK stage manifest installation by @shayasoolin in #611
• Retry KWOK Stage CRD wait by @shayasoolin in #612
• Replace scheduler backend global mutable state with well defined interface and dependency injection by @kangclzjc in #512
• fix: PodCliqueSet webhook crashes on unknown scheduler name by @weizhoublue in #613
• docs(mnnvl): update GREP-417 to single-annotation model by @shmuel-runai in #568
• fix: tolerate missing ResourceClaim API during PCLQ reconcile and delete by @SAY-5 in #610
• Refine scale test delete measurements by @shayasoolin in #621
• use lodash func by @oleg-kushniriov in #618
• Use embedded etcd for scale test cluster by @shayasoolin in #624
• Rename ClusterTopology CRD to ClusterTopologyBinding by @enoodle in #616
• change TopologyConstraint.TopologyName to optional by @enoodle in #601

New Contributors

• @steved made their first contribution in #547
• @shayasoolin made their first contribution in #522
• @yankay made their first contribution in #563
• @xianlubird made their first contribution in #581
• @weizhoublue made their first contribution in #613
• @SAY-5 made their first contribution in #610

Full Changelog: v0.1.0-alpha.8...v0.1.0-alpha.9

What's Changed

• GREP 369 - support multiple ClusterTopology by @enoodle in #413
• Crd upgrader impl by @danbar2 in #497
• [GREP-291] Introduce end-to-end tests for the OnDelete update strategy by @renormalize in #469
• feat: Expose pod index as a label to allow user-defined env vars to reference it via Downward API by @julienmancuso in #505
• Add Scheduler Backend framework by @kangclzjc in #293
• fix: use server-side apply in RU e2e tests by @enoodle in #504
• refactor: Use the scheduler backend to implement topology scheduling by @enoodle in #515
• Refactor/e2e reuse k8s clients by @oleg-kushniriov in #502
• [GREP] hierarchical resource sharing by @julienmancuso in #501
• remove utils package. move code to dedicated packages by @oleg-kushniriov in #521
• feat(e2e): scale test ergonomics with pprof and timeline tracking by @Ronkahn21 in #528
• chore: update kai version to non rc v0.14 by @enoodle in #467
• crd installer off by default by @danbar2 in #527
• add cache options for pods managed by grove only by @oleg-kushniriov in #533
• Add GREP-417: Selective MNNVL proposal by @shmuel-runai in #492
• Add mnnvl-group annotation validation and extend webhook to PodClique… by @shmuel-runai in #535
• feat: add hierarchical resource sharing by @julienmancuso in #507

New Contributors

• @enoodle made their first contribution in #413
• @oleg-kushniriov made their first contribution in #502

Full Changelog: v0.1.0-alpha.7...v0.1.0-alpha.8

What's Changed

• Fix e2e diagnostic log artifact upload path by @gflarity in #392
• test: remaining topology e2e tests by @Ronkahn21 in #383
• Upgrade golang version and k8s deps patch version by @unmarshall in #412
• Split e2e cluster creation by @danbar2 in #371
• add pr vaidation - issue must exist by @danbar2 in #410
• unskip rolling update test 18 by @danbar2 in #414
• Release ci automation by @danbar2 in #418
• Correct API documentation in PodCliqueSet by @renormalize in #432
• bug fix: prevent auto-mnnvl annotation injection on update by @shmuel-runai in #420
• docs: add TAS version requirement for KAI-scheduler by @Ronkahn21 in #440
• GREP-375 add scheduler backend framework by @kangclzjc in #372
• Run make tidy by @sanjaychatterjee in #443
• fix: handle startup-probe-phase pods in rolling update categorization by @gflarity in #435
• Auto-MNNVL: Add autoMNNVL e2e tests and cluster setup scripts by @shmuel-runai in #421
• GREP-291: OnDelete update strategy for PodCliqueSet by @renormalize in #403
• Fix: let operator self-create webhook TLS secret to avoid GitOps overwrites by @gflarity in #454
• Record Delete in CreateExpectation by @kangclzjc in #458
• [e2e] [mnnvl] resolve flaky e2e test in Config2_UnsupportedButEnabled by @shmuel-runai in #451
• Addition of a deletionTimestamp on a Pod should trigger a PodClique reconciliation by @xulinfei1996 in #433
• fix: add podCliqueScalingGroup concurrentSyncs to Helm chart by @Ronkahn21 in #472
• fix: validate container env var names in PodCliqueSet webhook by @shmuel-runai in #470
• feat: add configurable pprof bind address by @Ronkahn21 in #464
• replace test image to busybox by @kangclzjc in #476
• Add NVIDIA PLC-OSS issue templates by @itayvallach in #481
• [GREP-291] OnDelete implementation for standalone PodCliques and PodCliqueScalingGroups by @renormalize in #438
• refactor: add infra-manager CLI for cluster lifecycle by @Ronkahn21 in #465
• replace deprecated runner by @ranrubin in #447
• chore: migrate e2e cluster to KWOK nodes for faster CI by @Ronkahn21 in #489
• feat(e2e): add scale test measurement infrastructure by @Ronkahn21 in #484
• ci: setting up DCO validation for external contributors by @saturley-hall in #473
• Add stale issues workflow by @danbar2 in #494
• Remove duplicate and unused issue templates according to updated guidelines by @itayvallach in #486
• Add crd upgrade readme file by @danbar2 in #495
• Add auto MNNVL user guide documentation by @shmuel-runai in #437

New Contributors

• @itayvallach made their first contribution in #481
• @ranrubin made their first contribution in #447
• @saturley-hall made their first contribution in #473

Full Changelog: v0.1.0-alpha.6...v0.1.0-alpha.7

What's Changed

• fix startup ordering checks to match implementation by @gflarity in #323
• Auto-MNNVL: add MNNVL configuration and startup validation by @shmuel-runai in #346
• only warn about MTU/PMTU issues by @gflarity in #360
• add tests for updateObservedGeneration functions by @gflarity in #354
• add command for creating k3d debug cluster identical to one used in E2E by @gflarity in #361
• retry cluster creation when it fails by @gflarity in #327
• fix typo in SO1 and SO2 tests by @gflarity in #365
• Introduce GREP template and a refactored TAS GREP by @unmarshall in #362
• Danbar/e2e rolling update 10 by @danbar2 in #356
• test: add TAS simple level and constraint tests by @Ronkahn21 in #349
• Auto-MNNVL: Update PCS's webhook support auto-mnnvl by @shmuel-runai in #370
• Auto-MNNVL: add ComputeDomain component for PCS controller by @shmuel-runai in #363
• doc: Add pod-naming and environment-variables docs by @nvrohanv in #355
• Fix tas label by @kangclzjc in #380
• run e2e only if code changed and build pass by @danbar2 in #381
• fix: use patch for topology label repplay on replaced nodes by @Ronkahn21 in #384
• Auto-MNNVL: add PodSpec injection for MNNVL resourceClaims by @shmuel-runai in #385
• Gather all operator logs on e2e test failure by @gflarity in #358
• Enable ai-dynamo copy-pr-bot for Grove by @sanjaychatterjee in #382
• Auto-MNNVL: use correct ComputeDomain CRD field paths by @shmuel-runai in #391
• Support external certificate management for webhooks by @gflarity in #344
• Auto-MNNVL: validate annotation values and sync design doc by @shmuel-runai in #386

Full Changelog: v0.1.0-alpha.5...v0.1.0-alpha.6

What's Changed

• skip patch ObservedGeneration if no change by @xulinfei1996 in #337
• handle clean up failures better by @gflarity in #326
• fix: add PCS topology constraints to scaled PodGangs by @Ronkahn21 in #347
• fix: correct PCSG topology constraint handling for scaled PodGangs by @Ronkahn21 in #357
• test: add TAS e2e test infrastructure and basic tests by @Ronkahn21 in #348

New Contributors

• @xulinfei1996 made their first contribution in #337

Full Changelog: v0.1.0-alpha.4...v0.1.0-alpha.5

What's Changed

• document internal/utils by @gflarity in #211
• document internal/logger and internal/utils by @gflarity in #208
• document internal/controller by @gflarity in #205
• Changes for migration from @NVIDIA to @ai-dynamo by @renormalize in #225
• Introduce badges in README.md. by @renormalize in #227
• Remove Ask DeepWiki badge from README and add Go report badge by @unmarshall in #234
• bump CRD_REF_DOCS_VERSION by @gflarity in #232
• E2E Test Foundations by @gflarity in #207
• Grove proposal/topology by @Ronkahn21 in #224
• api: add Topology aware support by @Ronkahn21 in #235
• Bump github.com/docker/docker from 28.2.2+incompatible to 28.3.3+incompatible in /operator by @dependabot[bot] in #237
• added missed missed PR feedback by @gflarity in #238
• check for existing cluster and delete if it already exists by @gflarity in #239
• test coverage for internal/logger by @gflarity in #229
• Add Core Concepts Tutorial by @nvrohanv in #217
• Update Grove discord link with a permanent link. by @renormalize in #249
• Bump github.com/containerd/containerd from 1.7.28 to 1.7.29 in /operator by @dependabot[bot] in #253
• Fixed documentation links and formatting in README and installation by @sanjaychatterjee in #250
• test coverage for internal/webhooks by @gflarity in #230
• test coverage for internal/utils by @gflarity in #231
• Disallow reducing PodCliqueSetTemplateSpec.PodCliqueScalingGroupConfig.Replicas to 0. by @renormalize in #256
• add support for prepulling images to speed up tests on slow networks by @gflarity in #241
• Fix indentation in docs/designs/topology.md. by @renormalize in #257
• Feat/Topology Configuration Infrastructure by @Ronkahn21 in #247
• Add validation webhook for ClusterTopology resource by @shmuel-runai in #251
• Dependency version upgrades and fixes by @unmarshall in #263
• fix(charts): correct webhook configuration scope and metadata by @shmuel-runai in #258
• docs: update topology configuration naming in design doc by @Ronkahn21 in #266
• Bump golang.org/x/crypto from 0.44.0 to 0.45.0 in /operator by @dependabot[bot] in #268
• e2e tests gang scheduling by @gflarity in #242
• prepend a g to the github has in package version to avoid semver issues by @gflarity in #272
• Update Go image in Dockerfile to 1.25.3, and other tools. by @renormalize in #254
• E2E tests for startup ordering by @gflarity in #269
• ci: add e2e tests to GitHub Actions by @shmuel-runai in #282
• E2E: Fix flaky Helm installation failures due to "cannot re-use a nam… by @shmuel-runai in #284
• improve test coverage for internal/controller by @gflarity in #252
• Stablize E2E Tests by @gflarity in #287
• New TAS Design by @Ronkahn21 in #288
• feat: chart add ns by @ls-2018 in #290
• Cleaned and updated indirect go mod deps by @unmarshall in #295
• feat: remove useless branch condition by @ls-2018 in #289
• Feat/create cluster topology and KAI topology by @Ronkahn21 in #298
• Api/add name to topology constraint group by @Ronkahn21 in #299
• add mnnvl requirements GREP file by @danbar2 in #296
• Remove YEAR in generated files, adhering to the community convention. by @renormalize in #307
• Added more code owners for Grove by @sanjaychatterjee in #306
• E2e tests rolling updates by @gflarity in #280
• Added new code owner for Grove by @sanjaychatterjee in #308
• E2E stability fixes by @gflarity in #312
• Reconcile PodClique TopologyConstraints by @unmarshall in #302
• Introduce validations for TopologyConstraints in PodCliqueSet by @unmarshall in #317
• MNNVL support design doc by @shmuel-runai in #297
• cancel stale E2E runs by @gflarity in #322
• Fix get selector labels for pod by @gflarity in #318
• E2E Failure Diagnostics by @gflarity in #314
• Fixes for topology aware scheduling validation webhook by @unmarshall in #324
• Fixes TopologyConstraints for scaled PodGangs by @unmarshall in #340

Full Changelog: https://github.com/ai-dynamo/grove/commits/v0.1.0-alpha.4

What's Changed

• Gflarity/allow kai by @gflarity in #219
• Allow system:kube-controller-manager to update init container secret by @unmarshall in #221
• fixed the owner name to be lower case nvidia by @unmarshall in #222
• Disable Authorizer webhook by default. by @renormalize in #223

Full Changelog: v0.1.0-alpha.2...v0.1.0-alpha.3

What's Changed

• Add a attribution file for all the licenses used in Grove by @sanjaychatterjee in #189
• Remove LastOperation from CRDs and restructure component operators by @unmarshall in #192
• Remove scheduler development doc by @sanjaychatterjee in #197
• Remove validation which prevents setting NodeSelector on PodSpec by @unmarshall in #203
• Increase Default Value of TerminationDelay by @nvrohanv in #199
• Add unit-tests for initc and improve in-line doc strings by @gflarity in #204
• remove redundancy in initial grove readme paragraphs by @nvrohanv in #213
• Remove deadlock when deploying PCS with ComputeDomain by @unmarshall in #215
• document in internal/webooks by @gflarity in #210
• Introduce the Authroizer Webhook. by @renormalize in #214
• Rename leftover *podgangset*\.go to *podcliqueset*\.go from #186. by @renormalize in #216

New Contributors

• @gflarity made their first contribution in #204

Full Changelog: v0.1.0-alpha.1...v0.1.0-alpha.2

What's Changed

• Skeleton code and scripts for grove operator by @unmarshall in #5
• Adding the operator config api which got overwritten by @unmarshall in #6
• update license files and headers by @dmitsh in #7
• adapted license header to include The Grove Authors by @unmarshall in #8
• Added Dockerfile, Skaffold, Helm Charts and other misc changes by @unmarshall in #13
• allow setting object meta in PodClique; fix typos in types.go by @dmitsh in #14
• Removed PodGang CRD by @unmarshall in #15
• implement podgangset validating webhook by @dmitsh in #9
• Bump github.com/opencontainers/runc from 1.1.13 to 1.1.14 in /scheduler-plugins by @dependabot[bot] in #3
• Bump golang.org/x/crypto from 0.24.0 to 0.31.0 in /scheduler-plugins by @dependabot[bot] in #16
• Small fixes to hack scripts directory by @unmarshall in #20
• API changes and changes to validating webhook by @unmarshall in #23
• simplify PodCliqueSpec by @dmitsh in #24
• Introduced scheduler-api, modified PodGangSet API, re-generated code by @unmarshall in #25
• Bump golang.org/x/net from 0.26.0 to 0.33.0 in /scheduler-plugins by @dependabot[bot] in #26
• update podgangset crd by @dmitsh in #28
• Add Mutating Webhooks for PodGangSet by @ritikasrivastava in #17
• Configuration and deployment of webhooks by @unmarshall in #29
• Sample NIM LLM deployment specs using LWS and Grove by @sanjaychatterjee in #30
• update API by @dmitsh in #31
• implement validation for update operation by @dmitsh in #27
• Adds skeleton reconciler code and minor modifications to API by @dmitsh in #21
• Fixes for defaulting and validating webhooks by @unmarshall in #32
• fixed typos by @dmitsh in #33
• Add Default webhook unit test by @ritikasrivastava in #35
• Introduces miscellaneous changes by @unmarshall in #38
• Fixed helm charts and added default for podclique reconciler by @unmarshall in #39
• Fixes API, controller-runtime manager scheme and charts by @unmarshall in #42
• Bump k8s.io/kubernetes from 1.31.1 to 1.31.6 in /scheduler-plugins by @dependabot[bot] in #34
• implement basic reconciliation loop by @dmitsh in #40
• Bump golang.org/x/net from 0.33.0 to 0.36.0 in /scheduler-plugins by @dependabot[bot] in #46
• Refactor PodGangSet reconciler by @unmarshall in #51
• fixed roles for events and fixed test by @unmarshall in #52
• implement pclq status update by @dmitsh in #48
• move pclq status update to reconciler by @dmitsh in #53
• Added validation for pclq metadata by @unmarshall in #54
• Added changes to the PodGang API spec by @unmarshall in #55
• Update scheduler API by @unmarshall in #56
• Added TerminationDelay to PodGangTemplateSpec by @unmarshall in #58
• Ritika/headlessservice by @ritikasrivastava in #50
• Bump golang.org/x/net from 0.34.0 to 0.36.0 in /operator by @dependabot[bot] in #47
• Renamed PodClique to PodGroup in scheduler-api by @sanjaychatterjee in #61
• Refactoring operator by @unmarshall in #62
• Added ServiceAccount, Role and RoleBinding components by @unmarshall in #63
• Bump golang.org/x/net from 0.34.0 to 0.36.0 in /scheduler-api by @dependabot[bot] in #59
• Add stub functions for Grove scheduler plugin by @sanjaychatterjee in #60
• Fix broken test, upgrade to golangci-lint@v2.1.1 and fix numerous lint errors, upgrade tool versions, remove hack/tools.go, etc. by @renormalize in #64
• Replaced types.NamespacedName with own NamespacedName by @unmarshall in #65
• Corrections to API by @unmarshall in #67
• Bump golang.org/x/net from 0.35.0 to 0.38.0 in /operator by @dependabot[bot] in #66
• Introduced scheduling policy configuration in PodGangTemplateSpec by @unmarshall in #69
• Introduces scheduler policy config and other changes in operator and scheduler API by @unmarshall in #71
• Fix broken build targets, ld-flags during docker builds, work-tree state during docker builds, etc. by @renormalize in #70
• Added test utilities and other misc changes by @unmarshall in #74
• Reorg scheduler-plugins dir by @sanjaychatterjee in #75
• Upgraded k8s dependencies by @unmarshall in #76
• Updated operator and scheduler-api by @unmarshall in #77
• Introduced PodCliqueScalingGroup and refactored API modules by @unmarshall in #78
• PodCliqueScalingGroupConfig enhancement and validations by @unmarshall in #79
• Introduce make targets and charts for the development, and deployment of grove-kube-scheduler. by @renormalize in #80
• Introduce the reset-scheduler target which resets the kube-scheduler running in kind with the default. by @renormalize in #82
• introducing generated scheduler client by @unmarshall in #81
• Corrections in PGS components by @unmarshall in #83
• MinReplicas moves out of AutoScalingConfig to PodCliqueSpec by @unmarshall in #84
• Enhancements to API and reconcilers by @unmarshall in #86
• Allow usage of hyphen in the pgs and pclq names by @unmarshall in #87
• Specify defaults using annotations, use validating functions exposed by apimachinery, etc. by @renormalize in #89
• Misc fixes and partial implementation of Pod component by @unmarshall in #90
• Add validation to ensure all PodSpecs specify the same schedulerName. by @renormalize in #91
• Implement the init container. by @renormalize in #88
• Fix scale-in issues when PodGangSet replicas are changed. by @renormalize in #94
• Update Readme by @nvrohanv in #93
• Enhance PGS and PCLQ reconcilers to support PodGang lifecycle management by @unmarshall in #95
• Docs and API updates by @unmarshall in #97
• Updated diagrams and docs by @unmarshall in #99
• Introduce docs/getting-started.md. by @renormalize in #100
• Refactor PodGangSet and PodGang APIs by @unmarshall in #101
• Add test and coverage targets to Makefile by @Ronkahn21 in #102
• Modify PodCliqueScalingGroup behavior to create new PodCliques for each replica by @unmarshall in #103
• Fix HPA selector labels for PodCliques created by PodCliqueScalingGroups. by @renormalize in #104
• Enable GitHub Actions by @renormalize in #106
• API changes for Gang termination by @unmarshall in #107
• Pod name validation pgs by @Ronkahn21 in #105
• fix: fix example by @julienmancuso in #108
• Correct example in docs/getting-started.md. by @renormalize in #110
• Bump k8s.io/kubernetes from 1.33.1 to 1.33.2 in /scheduler by @dependabot[bot] in #92
• Add validation target to Makefile and update build-and-test.yaml by @Ronkahn21 in #111
• Gang Termination by @unmarshall in #114
• Pod discovery env vars by @Ronkahn21 in #119
• update release schedule to reflect dynamo alignment by @nvrohanv in #120
• feat: Add replicas and minAvailable fields for PodCliquesScalingGroups by @julienmancuso in #116
• Integrate `grove-initc...