All pull requests

Pull requests

• WIP: kvcache-aware: bind memory-mode KV event sources to their pods

  do-not-merge/work-in-progresskind/enhancementsize/XL
  #1867·YaoZengzeng opened Oct 10, 2026Member·

  ·
  4

• Update AutoscalingPolicy examples to the current vLLM KV-cache series.

  kind/documentationsize/S
  #1866·widerErwarten opened Oct 10, 2026·

  ·
  4

• router: fix healthy pods not picked when no score plugin scores them

  kind/bugsize/M
  #1865·Abhayanthk opened Oct 10, 2026Contributor·

  ·
  1

  9

• Support model express bootstrap acceleration

  kind/enhancementsize/XXL
  #1862·hzxuzhonghu opened Oct 9, 2026Member·

  ·
  9

• fix(router): reject a score plugin weight that is not positive

  kind/bugsize/M
  #1859·iamtanishqjain opened Oct 8, 2026Contributor·

  ·
  1

  5

• router: accept the stream terminator without the optional space

  kind/bugsize/M
  #1855·iamtanishqjain opened Oct 3, 2026Contributor·

  ·
  1

  10

• router: wake queues holding requests after each metrics scrape

  kind/bugsize/L
  #1853·btxu-db opened Oct 3, 2026Contributor·

  ·
  2

  1

• fix(router): wake session boost queue on backend updates

  kind/bugsize/XXL
  #1852·Piyush-13090 opened Oct 2, 2026·

  ·
  1

  4

• test: add KV-cache runtime E2E fixture

  kind/enhancementsize/L
  #1851·avinxshKD opened Oct 2, 2026Member·

  ·
  25

• Added proposal to support model express startup accelerate

  kind/enhancementlgtmsize/XL
  #1849·hzxuzhonghu opened Sep 30, 2026Member·

  ·
  48

  

• autoscaler: fix pod metric sum before recommendation logic

  kind/bugsize/L
  #1843·Abhayanthk opened Sep 29, 2026Contributor·

  ·
  1

  13

• router: decrement on-flight count on the PodInfo counted at dispatch

  kind/bugsize/L
  #1837·btxu-db opened Sep 29, 2026Contributor·

  ·
  1

  3

• feat(router): enable lora-affinity filter plugin

  kind/enhancementsize/XS
  #1836·Psykii22 opened Sep 28, 2026·

  ·
  1

  7

• Proposal: Support In-Place Rolling Update

  kind/documentationsize/XL
  #1829·kube-gopher opened Sep 25, 2026Member·

  ·
  14

• router: bound the path and upstream_model metric labels

  kind/bugsize/XL
  #1821·bolubo opened Sep 24, 2026·

  ·
  3

• controller: exit on unexpected leader election loss

  kind/bugsize/M
  #1820·pujitha24 opened Sep 23, 2026Contributor·

  ·
  13

  

• router: make the session sticky memory store safe to close concurrently

  kind/buglgtmsize/M
  #1817·aeron-gh opened Sep 23, 2026Contributor·

  ·
  7

  

• controller: exit process on leader election loss

  kind/bugsize/M
  #1814·AdeshDeshmukh opened Sep 22, 2026Contributor·

  ·
  1

  3

• docs: add the OpenAI-compatible batch API proposal

  kind/documentationsize/XL
  #1812·aeron-gh opened Sep 22, 2026Contributor·

  ·
  15

• feat(model-serving): lossless upgrade for ModelServing rolling updates

  kind/featuresize/XXL
  #1807·chenhuiluo opened Sep 22, 2026Contributor·

  ·
  1

  12

• feat: make ModelServing Role names immutable

  kind/enhancementsize/L
  #1805·haoeeeee opened Sep 22, 2026·

  ·
  1

  2

• feat: manage headless services through ModelServing plugin hooks

  kind/featuresize/XXL
  #1801·VanderChen opened Sep 22, 2026Member·

  ·
  1

  8

• feat: support per-pod ports for host-network workloads

  kind/featuresize/XXL
  #1800·VanderChen opened Sep 22, 2026Member·

  ·
  13

• fix(modelserving): preserve semantic revisions across upgrades

  kind/bugsize/XXL
  #1799·VanderChen opened Sep 21, 2026Member·

  ·
  1

  3

• fix: preserve Role configuration in outdated ServingGroups

  kind/bugsize/L
  #1796·haoeeeee opened Sep 21, 2026·

  ·
  1

  1