odh-v3.1

What's Changed

• bumping release tag to odh-v3.1 in push jobs by @brettmthompson in #954
• RHOAIENG-34916: CA Cert signing and validation by @pierDipi in #957
• RHOAIENG-37595: Change TLS certs mount path to not conflict with CA bundles by @pierDipi in #963
• chore: change default deployment mode to RawDeployment by @KillianGolds in #961
• choir: update kserve version by @Jooho in #965
• Add support for protocol in s3 endpoint by @brettmthompson in #964
• Improve ca bundle management for llmisvcs by @brettmthompson in #967
• Update samples to use the correct latest TLS cert directory by @pierDipi in #970
• Configure prefix-cache-scorer for P/D by @pierDipi in #969

Full Changelog: odh-v3.0...odh-v3.1

odh-v3.0

What's Changed

• Odh v2.35 release by @mholder6 in #910
• fix(webhook): relax HTTPRoute/Gateway validation by @bartoszmajsak in #911
• [llmisvc] docs/samples: Add READMEs and improve LLM inference examples by @pierDipi in #915
• Switch to official LWS operator (stable-v1.0) for OpenShift by @KillianGolds in #918
• Use flexible regex patterns to match kserve images from any registry by @Jooho in #927
• fix: Simplify image substitution by updating params.env before Kustomize by @Jooho in #929
• Add P/D and DP+EP docs regarding networking by @pierDipi in #925
• Add a guide on how to setup LLM Inference Service dependencies on OCP 4.18 by @pierDipi in #898
• fix: Skip validation for empty cpuModelcar and memoryModelcar fields by @KillianGolds in #930
• bump up llmd images to 0.3 by @Jooho in #916
• RHOAIENG-32956: Update kv-cache routing example by @pierDipi in #933
• Add metadata propagation for Kueue configurations to both Deployment and LeaderWorkerSet workloads by @hdefazio in #928
• Add listener hostname fallback with wildcard support for URL discovery by @pierDipi in #934
• fix: no disk space error when multiple arch image build. by @Jooho in #937
• Fix e2e tests llmisvc by @pierDipi in #935
• Only Completions Requests Go Through IGW by @brettmthompson in #938
• Add a guide for LLM Inference Service for GA and simple quickstart/Authentication tutorial by @Jooho in #904
• RHCL is now optional and status is reported accordingly by @pierDipi in #939
• set correct schema in .status.address.url when auth is disabled by @VedantMahabaleshwarkar in #942
• [release-v0.15] Support for Configurable oauth-proxy Timeout by @openshift-cherrypick-robot in #946
• [cherry-pick][RHOAIENG-32520] - Migrate oauth-proxy to kube-rbac-proxy image by @spolti in #951
• fix keda raw e2e tests for odh by @VedantMahabaleshwarkar in #952
• odh-v3.0 release prep by @brettmthompson in #953

Full Changelog: odh-v2.35...odh-v3.0

odh-v2.35

What's Changed

• [stable-2.x] Add DP+EP and no-scheduler e2e tests by @openshift-cherrypick-robot in #868
• Update authz part with the latest openshift cluster(4.19.9) by @Jooho in #869
• no longer prematurely exiting llmisvc route reconciliation by @brettmthompson in #877
• [RHOAIENG-32568] - Model serving builds for ODH community 2.34 by @spolti in #875
• modify keda autoscaling e2e test to work with ODH by @VedantMahabaleshwarkar in #874
• Propagate network configuration annotations and fix data-parallel-size argument for multi node by @pierDipi in #871
• Fix Potential HF Token Leak by @brettmthompson in #879
• add install scripts(kuadrant/gateway api) and setup for llm by @Jooho in #876
• Single node workload test cases by @brettmthompson in #870
• Avoid service account errors by @brettmthompson in #880
• Add more examples and fix discovery HCA list by @pierDipi in #885
• Resolve LWS leader address as ZMQ doesn't always resolve it by @pierDipi in #888
• Add Konflux ODH tag bump automation to our existing release workflow. by @KillianGolds in #892
• injecting ca bundle for s3 storage on llmisvc reconciliation by @brettmthompson in #894
• P/D over RoCE and kv-cache aware routing example by @pierDipi in #891
• ci(manifests): validates kustomize manifests by @bartoszmajsak in #890
• Add Support for Configuring S3 Storage via Secret Data by @brettmthompson in #900
• configuring hf downloads to lower memory usage by @brettmthompson in #895
• Retry DNS resolution and add more DP+EP with various all2all backends examples by @pierDipi in #897
• fix(router): readiness logic for multiple ctrl statuses by @bartoszmajsak in #908

New Contributors

• @KillianGolds made their first contribution in #892

Full Changelog: odh-v2.34...odh-v2.35

odh-v2.34

What's Changed

• fix(gomega): uses default timeouts when ctx is passed by @bartoszmajsak in #834
• chore(e2e): improves pytest setup by @bartoszmajsak in #832
• chore(ci): bumps default ginkgo timeout to 30s by @bartoszmajsak in #838
• chore: simplifies external CRD fetching by @bartoszmajsak in #831
• Update DP<>EP multi node deployment for llm-d 0.2 by @pierDipi in #802
• feat(e2e): installs llm-d specific infra stack by @bartoszmajsak in #833
• [RHOAIENG-30228] add-http-route-conditions by @andresllh in #827
• Odh v2.33 release by @mholder6 in #837
• Stop and resume an inference graph [Raw Deployment] by @hdefazio in #836
• [RHOAIENG-30228]-follow up by @andresllh in #841
• refactor: uses kustomize overlays for separation of concerns by @bartoszmajsak in #815
• update DEV.md to deploy kserve through opendatahub-operator by @Jooho in #846
• llmisvc: Fix conditions propagation by @pierDipi in #840
• Add basic llmisvc multi-node env tests by @pierDipi in #797
• Guard IsObjectNamespaced in Delete with NoMatchError by @pierDipi in #853
• Allow using HTTPRoute without scheduler and external InferencePool Ref by @pierDipi in #839
• chore: renames controlled-by owner assertion by @bartoszmajsak in #852
• fixing bug caused by misspelled BeControlledBy method by @brettmthompson in #856
• Add InferencePoolsReady as a subcondition to RouterReady by @hdefazio in #857
• Create DestinationRule with service hostname when there is no scheduler by @pierDipi in #862
• RHOAIENG-31916: adding support for credentials with s3 and hf storage for llmisvcs by @brettmthompson in #842
• feat(monitoring): add monitoring resource management for LLMInferenceService by @israel-hdez in #847
• feat(validation): ensures references are valid by @bartoszmajsak in #757
• Update LoRA types in llmisvc to use LLMModelSpec by @pierDipi in #861
• Always create workload DestinationRule when router is enabled by @pierDipi in #864
• Add DP+EP and no-scheduler e2e tests by @pierDipi in #855
• update LWS operator to use brew image by @Jooho in #865

Full Changelog: odh-v2.33...odh-v2.34

odh-v2.33

What's Changed

• [CI fix] update auto generated code by make precommit by @Jooho in #672
• feat: add Makefile target and scripts for LLMInferenceService dev by @Jooho in #671
• Various reconcile fixes by @bartoszmajsak in #674
• feat: improves status and events handling by @bartoszmajsak in #675
• Add llm-d inference service base configurations by @pierDipi in #676
• Enable LLMInferenceService controller by @pierDipi in #679
• feat(envtest): loads bundled presets automatically by @bartoszmajsak in #681
• Add well-know preset for single node by @pierDipi in #683
• Propagate Workloads conditions based on child deployments status by @pierDipi in #680
• fix(presets): refers to existing scheduler preset by @bartoszmajsak in #686
• Add LLMInferenceService CRD printer columns by @pierDipi in #682
• Deploy GIE EPP and associated InferencePool by @pierDipi in #687
• fix: follow-up to #687 by @bartoszmajsak in #688
• fix(reconcile): panic when copying InferencePool spec by @bartoszmajsak in #691
• fix(deployment): retries for status update by @bartoszmajsak in #692
• feat(test): custom matchers by @bartoszmajsak in #694
• feat: generic resource CRUD handling by @bartoszmajsak in #697
• feat(reconcile): managed http routes by @bartoszmajsak in #696
• test: ensures presets are valid by @bartoszmajsak in #698
• follow up deployment script for llmd by @Jooho in #699
• feat(presets): additional context in templating by @bartoszmajsak in #700
• feat(gw): uses default ingress gateway by @bartoszmajsak in #701
• chore(config): injects llmisvc.Config struct in the templates by @bartoszmajsak in #708
• Fix panic for invalid templates and add additional tests for MergeSpecs and ReplaceVariables by @pierDipi in #707
• fix(config): reworks config to be immutable by @bartoszmajsak in #711
• Path-based routing for model services by @pierDipi in #712
• feat(test): cleans up test fixtures by @bartoszmajsak in #715
• docs: initial dev guide by @bartoszmajsak in #718
• feat(router): resolves model URLs by @bartoszmajsak in #719
• Support LWS-based multi node with data parallelism of MoE models by @pierDipi in #716
• Fix typo in LLMD dev md. by @Jooho in #720
• chore(presets): simplifies template defaulting by @bartoszmajsak in #721
• chore: moves sample LLMInferenceService to its own func by @bartoszmajsak in #724
• chore(kustomize): consolidates webhook patching by @bartoszmajsak in #725
• chore(test): splits existing/non-existing well known configs by @bartoszmajsak in #726
• Add instructions for dev setup of LLMIsvc on minikube by @israel-hdez in #728
• feat(webhook): introduces validation for configs by @bartoszmajsak in #723
• update istio experimental image by @Jooho in #722
• fix(scheduler): creates missing endpoint picker service by @bartoszmajsak in #729
• fix(webhook): adds missing patch for kustomize by @bartoszmajsak in #731
• chore: follow-up to #729 by @bartoszmajsak in #733
• chore: remaining Reconcile re-use by @bartoszmajsak in #734
• feat(storage): add PVC support for model storage to LLMIsvc by @israel-hdez in #730
• chore: adds unit tests for webhook warnings by @bartoszmajsak in #735
• feat(validation): cross-reference field checks by @bartoszmajsak in #732
• fix(reconcile): populates routing status for user-def routes by @bartoszmajsak in #745
• Improve enqueue logic for Gateway and well-known presets by @pierDipi in #746
• fix(webhook): adds new webhooks to self-signed-cert script by @bartoszmajsak in #747
• Configure scheduler to FailOpen and fallback to random loadbalancing by @pierDipi in #749
• feat(storage): add OCI storage support via modelcars by @israel-hdez in #738
• update owners by @VedantMahabaleshwarkar in #751
• fix(test): uses local httptest server urls for downloader tests by @bartoszmajsak in #753
• feat(webhook): ensures proper paralellism settings by @bartoszmajsak in #754
• Add an example for using PVC in LLMIsvc by @israel-hdez in #750
• Support TLS with self-signed certificates for LLMInferenceService by @pierDipi in #742
• chore: enriches error for storage initializer by @bartoszmajsak in #759
• fix(test): adds required storage initializer settings to test configmap by @bartoszmajsak in #758
• RHOAIENG-30217: Enable SSRF protection for sidecar by @pierDipi in #748
• fix(deps): adds missing LLMInferenceService types for python by @bartoszmajsak in #762
• chore: replaces gatewayclass in-memory by @bartoszmajsak in #763
• feat(ci): adds status check for system namespace by @bartoszmajsak in #764
• feat(dev): handles GIE mode for dev/ci setup by @bartoszmajsak in #765
• RHOAIENG-30166: Upgrade to llm-d 0.2 by @pierDipi in #761
• fix(presets): respects baseRefs defaults by @bartoszmajsak in #772
• feat(storage): Add support to LLMIsvc for downloading from HF and S3 by @israel-hdez in #755
• Odh v2.32 release by @mholder6 in #773
• Merge feature branch to release 0.15 branch by @Jooho in #776
• add llmd scheduler/routing sidecar image to params.env by @Jooho in #777
• update istiod yaml file to increase resources by @Jooho in #781
• add GIE CRD by @Jooho in #778
• fix: fixes misleading print for parallelism when $# > 2 by @bartoszmajsak in #784
• chore: updates router sidecar image by @bartoszmajsak in #783
• Prefetch GIE CRDs by @pierDipi in #786
• Fix multi node prefill template and collect routes by @pierDipi in #787
• feat: stub for llm-inference-service test runner on oc/prow by @bartoszmajsak in #788
• follow-up to #788 by @bartoszmajsak in #789
• fix: defaults GITHUB_SHA for graph images by @bartoszmajsak in #785
• chore: simplifies url filtering by @bartoszmajsak in #790
• feat: Enable LLM disaggregated serving on OpenShift by @Jooho in #782
• fix: add HuggingFace environment variable(HF_HOME) for storage-initializer by @Jooho in #780
• test(e2e): introduces e2e test skeleton by @bartoszmajsak in #774
• Restrict cache for secrets with label selector by @pierDipi in #791
• Remove storage spec from LLMIsvc by @israel-hdez in #775
• fix(envtest): ensures consistent env test setup by @bartoszmajsak in #798
• docs: minor fixes in DEV.md by @bartoszmajsak in https://github.com/opendatahub-io/kserve/...

odh-v2.32

What's Changed

• [release-v0.15] Align support secure access to prometheus in Keda autoscaling to upstream API by @pierDipi in #739
• Updating hf-xet package version to >= 1.1.2 by @brettmthompson in #743
• Update Tekton files to version odh-v2.32 by @odh-devops-app[bot] in #744
• RHOAIENG-30193-fixed-constant-looping cherry pick by @andresllh in #770

New Contributors

• @odh-devops-app[bot] made their first contribution in #744

Full Changelog: odh-v2.31...odh-v2.32

odh-v2.31

What's Changed

• [Cherry pick] Python test fixes from upstream by @hdefazio in #660
• [Cherry pick][Release 0.15] Stop and resume a model [Raw Deployment] by @hdefazio in #661
• [Cherry-Pick] update pip version by @Jooho in #685
• [release-v0.15] Resolve inference endpoint using runtime protocol when applicable by @openshift-cherrypick-robot in #690
• [Cherry-pick] Stop and Resume a transformer by @hdefazio in #704
• Stop and resume an explainer by @hdefazio in #706

Full Changelog: odh-v2.30...odh-v2.31

odh-v2.30

What's Changed

• Allow OCI for multi-node/multi-gpu [v0.15] by @israel-hdez in #620
• Fix e2e tests by @spolti in #634
• [release-v0.15] Sync with upstream tag v0.15.2 by @israel-hdez in #632
• Fix issue with OCI support for multi-node/multi-gpu: missing MODEL_DIR envvar by @israel-hdez in #627
• [release-v0.15] Add pierDipi and bartoszmajsak to OWNERS by @openshift-cherrypick-robot in #638
• [Cherry-pick] RHOAIENG-26179: Deploy CMA / KEDA during E2E tests setup by @openshift-cherrypick-robot in #635
• [release-v0.15] Disallow Propagation of haproxy.router.openshift.io/timeout Annotation by @openshift-cherrypick-robot in #639
• [release-v0.15] RHOAIENG-24134: KServe Support secure access to the metrics server with TriggerAuthentication for KEDA autoscaling by @openshift-cherrypick-robot in #644
• [release-v0.15] Back-porting fixes for InferenceGraph by @openshift-cherrypick-robot in #648
• Fix references to ODH images in params.env by @israel-hdez in #649
• Restoring zero initial scale behavior by @brettmthompson in #646
• Fix the ODH release workflow for v0.15 branch by @hdefazio in #653

New Contributors

• @openshift-cherrypick-robot made their first contribution in #638

Full Changelog: odh-v2.29...odh-v2.30

odh-v2.29

What's Changed

• add storage-initializer uid handling for OpenShift with istio-cni by @ReToCode in #18
• Adapt GH-workflows to correctly push to ODH container repositories by @israel-hdez in #28
• Update OWNERS files by @israel-hdez in #38
• Openshift-ci onboarding by @israel-hdez in #39
• [Sync] kserve/kserve-master to master branch by @vaibhavjainwiz in #70
• Upstream master sync by @israel-hdez in #83
• automate addition of new isues into ODH board by @heyselbi in #97
• Code sync up to upstream commit for v0.11.1 by @israel-hdez in #108
• add spoltin in the OWNERS file by @spolti in #110
• [RHODS-12555] - CVE-2023-44487 by @spolti in #109
• [RHODS-12555] - CVE-2023-44487 - qpext by @spolti in #114
• [master] Preparation for odh-opeartor v2 by @israel-hdez in #111
• Enable Istio Virtual Host by @israel-hdez in #115
• remove webhookClientConfig from inferenceservice crd of kserve by @VedantMahabaleshwarkar in #120
• Increase kserve controller limit resources(cpu/memory) by @Jooho in #125
• Code sync with upstream by @israel-hdez in #133
• [pull] master from kserve:master by @pull in #88
• [pull] master from kserve:master by @pull in #136
• [pull] master from kserve:master by @pull in #142
• Use ThirdParty tokens for CI, for compatibility with ROSA-Hosted clusters by @israel-hdez in #146
• Revert and re-adapt test/e2e/common/utils.py by @israel-hdez in #147
• Sync by @spolti in #159
• [pull] master from kserve:master by @pull in #161
• [pull] master from kserve:master by @pull in #162
• Bump golang.org/x/crypto from 0.1.0 to 0.17.0 in /docs/samples/graph/bgtest/bgtest by @dependabot in #149
• Reverts of unneeded commits; and upstream code sync by @israel-hdez in #179
• Update OWNERS in main branch by @heyselbi in #204
• Code sync with upstream 2024-02-15 by @israel-hdez in #206
• [pull] master from kserve:master by @pull in #194
• [pull] master from kserve:master by @pull in #219
• set disableIngressCreation to true in odh overlay by @VedantMahabaleshwarkar in #251
• add .tekton to gitignore to avoid conflict due to Konflux by @Jooho in #254
• fix syntax error in inferenceservice-config patch by @VedantMahabaleshwarkar in #257
• [RHOAIENG-3551] - fastapi - Regular Expression Denial of Service (ReDoS) by @spolti in #256
• Increase memory limit of kserve-controller pod by @israel-hdez in #261
• Synch by @spolti in #264
• Fix CI: Serverless removed domain-mapping deployment by @israel-hdez in #277
• Sync by @spolti in #275
• [pull] master from kserve:master by @pull in #276
• Add capability to run RawDeployment E2Es in OpenShift-ci by @israel-hdez in #220
• [RHOAIENG-5073] - Routing and Headless Service Support in KServe Raw … by @spolti in #280
• Sync2 by @spolti in #289
• Create Workflow for release and tag with Changelog by @rpancham in #287
• remove unnecessary operators by @Jooho in #316
• [RHOAIENG-4617] - follow up - remove hardcoded fastapi from Dockerfile by @spolti in #292
• Fix storage-initializer-docker-publisher workflow by @israel-hdez in #340
• Sync kserve/master with odh/master by @spolti in #356
• [pull] master from kserve:master by @pull in #357
• [pull] master from kserve:master by @pull in #358
• replace upstream Dockerfiles with ubi dockerfiles. by @VedantMahabaleshwarkar in #362
• Docs for authorization feature by @israel-hdez in #253
• enable DirectPvcVolumeMount by @VedantMahabaleshwarkar in #367
• [pull] master from kserve:master by @pull in #361
• [pull] master from kserve:master by @pull in #374
• [pull] master from kserve:master by @pull in #379
• Add workflow to Trigger build on PR add LGTM and Create Tag and Release with Changelog and push image to quay by @rpancham in #366
• update local gateway information by @Jooho in #372
• CWE-362 - anyio Race Condition by @spolti in #382
• set protocol https for .Status.Address.URL by @Jooho in #386
• explicitly enable ubi-8 baseos repo by @riprasad in #395
• Removed arm/v7 to make the jobs working again for now by @npanpaliya in #391
• [RHOAIENG-10904] Updated istio versions based on go version 1.21 by @mholder6 in #396
• Fix apiGroups in aggregate roles on manifests by @israel-hdez in #401
• [RHOAIENG-12577] - enable modelCar by default by @spolti in #407
• [RHOAIENG-12260] - [konflux] hermetic build for storage-initializer i… by @spolti in #411
• Documentation on using OCI images for model storage (modelcars) by @israel-hdez in #415
• [RHOAIENG-14595] Updated the OWNERS file by @mholder6 in #420
• Code sync for upstream v0.14.0 by @israel-hdez in #421
• Revert 'Increase memory limit of kserve-controller pod' by @hdefazio in #422
• [20241101] ODH Master Sync by @Jooho in #425
• Fix issues after 0.14.0 sync by @Jooho in #424
• [Cherry-pick] To fix e2e test, cherry pick commits by @Jooho in #428
• Revert "[Cherry-pick] To fix e2e test, cherry pick commits" by @Jooho in #429
• [RHOAIENG-15409] add knativeLocalGatewayService to fix VS issue by @Jooho in #431
• Upstream sync 20241106 by @Jooho in #433
• upgrade golang from 1.21 to 1.22 by @Jooho in #435
• [Revert] b7a868f commit for 1.22 upgrade by @Jooho in #437
• [RHOAIENG-11034] - Disabling Authorino Token Authorization forces a Model Pod to rollout by @spolti in #441
• Trust OpenShift service CA in kserve-router by @israel-hdez in #426
• Use odh-model-controller incubating branch by @hdefazio in #443
• Seprate creating dsc/dsci objects part from kustomization by @Jooho in #445
• add oauth-proxy to rawdeployments if odh auth label is present by @VedantMahabaleshwarkar in #419
• Fix old tag grab by @mholder6 in #452
• Sync with upstream main by @hdefazio in #456
• Add GenAI example for using OCI images for model storage by @israel-hdez in #447
• Sync kserve/master into ODH/master by @spolti in #459
• [RHOAIENG-16851] - Rawdeployment bug fixes by @VedantMahabaleshwarkar in #462
• Remove starlette pinned version because it is already in the requirem… by @hdefazio in #473
• [master] Enable forward...

odh-v2.28

What's Changed

• [Cherry Pick] add flag to control tls behavior for kserve router by @israel-hdez in #566
• [RHOAIENG-23169] - StorageInitializer fails do download models from h… by @spolti in #565
• [Cherry-pick][RHOAIENG-7607]Enhance InjectAgent to Handle Only HTTPGet, TCP Readiness Probes (kserve#4012) by @spolti in #570
• Fix kserve-controller creating two InferenceGraph pods by @israel-hdez in #575
• disable autoMountServiceAccountToken: true globally (#553) by @spolti in #576
• Prevent External Route Changes To ISVCs From Creating New Revisions (… by @spolti in #578
• [RHOAIENG-19717] Old pull secret persists when updating isvcs (#522) by @spolti in #577
• Cherry Pick for 0.14 Release by @brettmthompson in #579

Full Changelog: odh-v2.27...odh-v2.28