All pull requests

Pull requests

• feat(controller): add TrainJob TTL cleanup and runtime defaults

  do-not-merge/work-in-progresssize/XXL
  #4297·XploY04 opened Oct 22, 2026Contributor·1 task done·

  ·
  1

• feat(api): implement MPI node groups (KEP-4064)

  do-not-merge/work-in-progresssize/XXL
  #4296·matbun opened Oct 22, 2026Contributor·1 task done·

  ·
  1

• fix(webhooks): reject TrainJob env names and resources that Pods cannot accept

  size/L
  #4293·harshrajdebug opened Oct 21, 2026·1 task·

  ·
  1

  1

• fix(api): reject OptimizationJob trainer env that trial TrainJobs cannot accept

  size/M
  #4286·rshdhere opened Oct 21, 2026Contributor·1 task·

  ·
  1

  1

• fix(framework): apply runtime numNodes to MPI worker nodes

  size/L
  #4284·sakethalladaaa opened Oct 20, 2026·1 task·

  ·
  1

  2

• chore(test): add Open MPI 4 e2e for the MPI runtime

  ok-to-testsize/L
  #4282·matbun opened Oct 20, 2026Contributor·1 task·

  ·
  1

  2

• fix(cache): reuse one Flight connection per worker in the data cache writer

  ok-to-testsize/L
  #4281·DEVJHAWAR11 opened Oct 20, 2026·1 of 2 tasks·

  ·
  1

  5

  

• chore(deps): bump github/codeql-action/upload-sarif from 4.38.2 to 4.39.0

  dependenciesgithub_actionssize/XS
  #4277·dependabot[bot] opened Oct 19, 2026Bot·

  ·
  2

• chore(deps): bump mlx-lm from 0.31.3 to 0.32.2 in /cmd/runtimes/mlx

  dependenciespythonsize/XS
  #4276·dependabot[bot] opened Oct 19, 2026Bot·

  ·
  2

• chore(deps): bump mlx from 0.32.3 to 0.32.4 in /cmd/runtimes/mlx

  dependenciespythonsize/XS
  #4274·dependabot[bot] opened Oct 19, 2026Bot·

  ·
  2

• chore(deps): bump transformers from 5.18.0 to 5.20.1 in /cmd/runtimes/deepspeed

  dependenciespythonsize/XS
  #4275·dependabot[bot] opened Oct 19, 2026Bot·

  ·
  2

• chore(deps): bump tokio from 1.53.1 to 1.53.3 in /pkg/data_cache

  dependenciesrustsize/XS
  #4273·dependabot[bot] opened Oct 19, 2026Bot·

  ·
  2

• chore(deps): bump tokio from 1.53.1 to 1.53.3 in /pkg/data_cache/test

  dependenciesrustsize/XS
  #4272·dependabot[bot] opened Oct 19, 2026Bot·

  ·
  2

• feat(runtime): resolve the GPU count from DRA ResourceClaimTemplates

  size/XXL
  #4269·Sridhar1030 opened Oct 19, 2026Member·1 task done·

  ·
  1

  2

• feat(docs): add OptimizationJob user guide and runnable example

  size/L
  #4268·aniket2405 opened Oct 19, 2026Member·1 task done·

  ·
  5

• feat(api): allow per-TrainJob schedulerName overrides

  size/L
  #4265·sfeng1996 opened Oct 19, 2026·1 task done·

  ·
  1

• fix(cache): stop discarding failures in the data cache assignment handshake

  ok-to-testsize/L
  #4262·tanmay442 opened Oct 18, 2026·1 task·

  ·
  1

  4

  

• fix(cache): cut quadratic work from data cache startup and query path

  size/L
  #4259·tanmay442 opened Oct 17, 2026·1 task·

  ·
  1

  3

• fix(runtimes): include list index in RuntimePatch manager validation path

  size/S
  #4257·nikhil3495 opened Oct 17, 2026Contributor·1 task·

  ·
  1

  1

• fix(runtimes): validate inherited TorchTune node counts

  size/M
  #4255·spicysharkco opened Oct 17, 2026·1 task done·

  ·
  2

• [WIP] feat: Workload-Aware Scheduling for TrainJob – Phase 1 (KEP-3015)

  do-not-merge/work-in-progresssize/XXL
  #4254·andreyvelich opened Oct 17, 2026Member·1 of 3 tasks·

  ·
  1

  28

  

• fix(controller): surface runtime resource creation errors

  size/L
  #4253·Prat260104 opened Oct 16, 2026Contributor·

  ·
  1

  3

• feat(controller): Support Job Replicas for multi-slice TPU

  size/XL
  #4250·andreyvelich opened Oct 16, 2026Member·

  ·
  7

  

• docs: update DeepSpeed and PyTorch user guides

  size/L
  #4243·Nesar976 opened Oct 15, 2026·

  ·
  1

  1

• fix(runtimes): do not overwrite existing runtime snapshot

  do-not-merge/holdsize/L
  #4239·vanshika2720 opened Oct 14, 2026·1 task·

  ·
  1

  3