All pull requests

Pull requests

• refactor(metrics): share disaggregation path label values

  area/coordinatorarea/devarea/docsarea/epparea/schedulingarea/sidecararea/telemetrykind/cleanupsize/L
  #3113·satyamg1620 opened Oct 1, 2026Contributor·

  ·
  1

• Remove deprecated pd-sidecar vllm-port configuration

  area/sidecarkind/cleanupkind/deprecationsize/L
  #3112·0z5a opened Oct 1, 2026Contributor·

  ·
  1

  3

• feat(capacityledger): add 3-axis capacity attribute and account model

  area/datalayerarea/epparea/schedulingsize/XL
  #3111·LukeAVanDrie opened Oct 1, 2026Contributor·

  ·
  1

• fix: canonicalize inference API paths and document routing policy

  area/coordinatorarea/docsarea/sidecarkind/featuresize/XL
  #3109·satyamg1620 opened Oct 1, 2026Contributor·

  ·

• feat(metrics): scrape vLLM NIXL failure counters per endpoint

  area/datalayerarea/docsarea/eppkind/featuresize/L
  #3108·satyamg1620 opened Sep 30, 2026Contributor·

  ·

• docs: Update AGENTS.md: require value check before constant declaration

  area/docskind/documentationsize/XS
  #3107·roytman opened Sep 30, 2026Contributor·

  ·
  5

• Make predicted latency failures observable

  area/docsarea/epparea/schedulingkind/bugsize/L
  #3106·Re1lya opened Sep 30, 2026·

  ·
  1

  2

• coordinator: share the connector and decode proxy code of the steps

  area/coordinatorkind/cleanupsize/L
  #3101·revit13 opened Sep 30, 2026Contributor·

  ·

• feat(sidecar): warn at startup when SSRF protection is disabled

  area/docsarea/sidecarkind/featuresize/L
  #3100·satyamg1620 opened Sep 30, 2026Contributor·

  ·
  1

• epp: add served-model-filter to route only to pods listing model

  area/datalayerarea/docsarea/epparea/schedulingkind/featuresize/XXL
  #3096·bartoszmajsak opened Sep 29, 2026·

  ·

• feat(sidecar): record stage metrics on the p2p and shared-storage paths

  area/sidecarkind/featuresize/L
  #3094·satyamg1620 opened Sep 29, 2026Contributor·

  ·
  1

• fix(sidecar): record stage metrics in the P2P connector

  area/sidecarkind/bugsize/M
  #3093·hmdsefi opened Sep 29, 2026·

  ·
  1

• fix(disagg): do not count rejected prefill requests as decode-only

  area/docsarea/epparea/schedulingkind/bugsize/M
  #3092·satyamg1620 opened Sep 29, 2026Contributor·

  ·
  1

• fix(epp): strip client-supplied routing headers before forwarding

  area/eppkind/bugsize/L
  #3090·satyamg1620 opened Sep 29, 2026Contributor·

  ·
  1

• deps: bump Gateway API/GAIE to v1.6.2, drop legacy CRD group

  area/devarea/eppkind/deprecationsize/L
  #3086·elevran opened Sep 29, 2026Collaborator·

  ·
  1

  3

  

• fix(datalayer): read queue, running requests and KV use across a pod's engines

  area/datalayerarea/eppkind/bugsize/L
  #3085·mtparet opened Sep 29, 2026Contributor·

  ·
  1

  5

  

• feat(epp): simplified thunder-agent per-pod admission gate in fairnessPolicy

  area/eppkind/featuresize/XXL
  #3076·zetxqx opened Sep 29, 2026Contributor·

  ·

• perf: kvcache-align and fuse prefix matching for ~2x speedup

  area/epparea/kvcachekind/performancesize/XXL
  #3073·mauriciopoppe opened Sep 28, 2026Contributor·

  ·
  1

• Record prefix-cache prediction for the prefill endpoint under P/D

  area/docsarea/epparea/schedulingkind/featuresize/L
  #3070·satyamg1620 opened Sep 28, 2026Contributor·

  ·
  1

  1

• perf: Add EPP tracing overhead performance study

  area/devkind/documentationsize/XXL
  #3068·albertoperdomo2 opened Sep 28, 2026Contributor·

  ·
  1

• Count the request's own tokens in the concurrency filter

  area/epparea/flowcontrolkind/featuresize/L
  #3066·vMaroon opened Sep 28, 2026Member·

  ·
  2

• Count dispatched requests in saturation before PreRequest

  area/epparea/flowcontrolarea/schedulingkind/bugsize/L
  #3065·vMaroon opened Sep 28, 2026Member·

  ·
  5

• Add EPP flags for the leader-election lease name and timings

  area/eppkind/featuresize/L
  #3064·vMaroon opened Sep 28, 2026Member·

  ·
  3

  

• fix(kvcache): count admissions/evictions only on successful index writes

  area/kvcachekind/bugsize/M
  #3062·ptimizeroracle opened Sep 28, 2026Contributor·

  ·
  1

  7

• test(helm): add HTTPRoute rendering validation

  area/devkind/testsize/XS
  #3060·learner0810 opened Sep 28, 2026Member·

  ·