All pull requests

Pull requests

• [Router][Feat] Add /reset_prefix_cache router passthrough

  #1106·AyushKashyapII opened Sep 25, 2026·

  ·
  1

• [Bugfix][Router] Only buffer the proxied response body when it is consumed

  #1104·lfsun02 opened Sep 25, 2026Contributor·

  ·
  4

• [Router][Bugfix] Keep the KV-aware /tokenize fallback off the event loop

  #1101·David-Wu1119 opened Sep 24, 2026·

  ·
  3

• [Feat] Add global image registry override to Helm chart

  #1100·rohan-patnaik opened Sep 23, 2026·

  ·
  1

  2

• feat(helm): optional runtime AI inventory via k8s-aibom (default off)

  #1099·glenmessenger opened Sep 22, 2026·

  ·
  4

• [Router] Add --timeout-keep-alive to configure uvicorn's idle keep-alive timeout

  #1098·mrembalski opened Sep 22, 2026·

  ·
  1

• [Bugfix][Router] Log request headers as a lazy argument so they can be redacted

  #1097·chrikrah opened Sep 18, 2026Contributor·

  ·
  3

• [Bugfix][Router] dispatch PriorityRouter in proxy_multipart_request

  #1095·chrikrah opened Sep 18, 2026Contributor·

  ·
  2

• [Bugfix][Router] carry the three label fields through DynamicRouterConfig.from_args

  #1094·chrikrah opened Sep 17, 2026Contributor·

  ·
  6

• [Bugfix][Router] follow the batch module through the rename that moved it

  #1093·chrikrah opened Sep 17, 2026Contributor·

  ·
  6

• [Bugfix][Router] Reconcile the engine list when the K8s watch reconnects

  #1092·jStrider opened Sep 16, 2026·

  ·
  1

  6

• [Bugfix] Keep multi-round benchmark summaries consistent

  #1090·1fanwang opened Sep 14, 2026·

  ·
  1

  4

• [Bugfix][Router] Use least-inflight for PrefixAware below-threshold fallback

  #1089·dundysm opened Sep 13, 2026Contributor·

  ·
  1

  1

• [Feat] Opt-in DRA allocation mode per model

  #1086·jdubs opened Sep 13, 2026·

  ·
  1

  11

• [CI/Build] Use mock backends for static-discovery e2e (part of #531)

  #1085·ighutake-debug opened Sep 13, 2026Contributor·

  ·
  3

• [Bugfix][Router] Validate static service discovery list lengths

  #1084·Alex-Wengg opened Sep 11, 2026·

  ·
  1

  11

• [Router] Make backend request timeouts configurable via environment variables

  #1082·vitush93 opened Sep 10, 2026·

  ·
  3

• [Helm][Feature] Add vllm-stack-operator chart: operator, CRDs and CRs in one helm install

  #1081·jackiesre721 opened Sep 10, 2026·

  ·
  21

• [Bugfix][Router] Use supplied timestamp for avg_latency and drop completed request maps

  #1080·nisaral opened Sep 7, 2026·

  ·
  5

• [Operator] Add separate resource requests and limits

  #1079·banlor opened Sep 3, 2026Contributor·

  ·
  1

  4

• [Router][Bugfix] Log a warning when an engine scrape matches zero known metric names

  #1076·pujitha24 opened Sep 2, 2026·

  ·
  1

  8

• [Router][Bugfix] Decouple the engine-stats read timeout from the scrape interval

  #1075·sethforprivacy opened Sep 2, 2026·

  ·
  5

• [Feat] Support command and args on servingEngineSpec.modelSpec

  #1071·dundysm opened Sep 1, 2026Contributor·

  ·
  1

  4

• [Bugfix][Router] Run KvawareRouter /tokenize fallback in executor to avoid blocking event loop

  #1069·Asthenia0412 opened Aug 29, 2026·

  ·
  3

• [Bugfix][Helm] Only pass --k8s-service-discovery-type when explicitly set to non-default

  #1068·Asthenia0412 opened Aug 29, 2026·

  ·
  1

  2