All pull requests

Pull requests

• Fix flaky broker proxy port persistence tests

  #187·kjlubick opened Oct 29, 2026Collaborator·

  ·

  · 1.0.x

• Add ModelScope as a llama.cpp model download source

  #186·NV-sschneider opened Oct 29, 2026Collaborator·5 tasks done·

  ·

• Simplify pairing code and reject out-of-order messages

  #184·kjlubick opened Oct 28, 2026Collaborator·

  ·

• Pin the Fern CLI at 5.168.0

  #183·Noah-Tervalon-Nvidia opened Oct 28, 2026Collaborator·6 of 7 tasks·

  ·

• feat: add vLLM engine with on-request model switching

  #181·nvtushark opened Oct 28, 2026Collaborator·7 tasks done·

  ·
  25

• Keep slow engine status checks from delaying engine controls

  #180·kjlubick opened Oct 27, 2026Collaborator·

  ·

  · 1.0.x
  1

• Authenticate peer inventory before using it for service decisions

  #179·kjlubick opened Oct 26, 2026Collaborator·

  ·

  · 2.0.x

• Broker concurrent teardown fix

  #173·Noah-Tervalon-Nvidia opened Oct 14, 2026Collaborator·7 tasks done·

  ·
  6

• Forget a peer's manual entry when the TUI removes it from the cluster

  #172·Noah-Tervalon-Nvidia opened Oct 14, 2026Collaborator·7 tasks done·

  ·
  7

• review and add of Llama.cpp in 3 places

  #171·megnvidia opened Oct 14, 2026Collaborator·2 of 7 tasks·

  ·

• Limit JSON reads for node metadata and peer messages

  #167·kjlubick opened Oct 12, 2026Collaborator·

  ·

  · 1.0.x

• Check previously discarded errors and simplify Go tests

  #161·kjlubick opened Oct 9, 2026Collaborator·7 tasks done·

  ·

  · 1.0.x

• feat(proxy): order a stalled candidate behind its peers for one deadline

  #144·canja006 opened Oct 2, 2026·

  ·
  11

• Show a job as running once its engine starts it

  #143·ckelseynv opened Oct 2, 2026Collaborator·7 tasks done·

  ·

• Keep same-numbered jobs apart and apply each job's events in order

  #142·ckelseynv opened Oct 1, 2026Collaborator·7 tasks done·

  ·

• test(proxy): bound TCP buffers in flush deadline regression

  #139·RecursiveIntell opened Sep 30, 2026Contributor·6 tasks done·

  ·
  1

• fix(engine-manager): sweep engines once per shutdown

  #137·Noah-Tervalon-Nvidia opened Sep 29, 2026Collaborator·7 tasks done·

  ·
  12

• fix(engine-manager): Stopping a leftover Ollama engine works on macOS

  #136·Noah-Tervalon-Nvidia opened Sep 29, 2026Collaborator·7 tasks done·

  ·

• docs(nvpair-ui-broker): add normative service specification

  #135·Noah-Tervalon-Nvidia opened Sep 28, 2026Collaborator·6 of 7 tasks·

  ·
  3

• docs(nvpair-proxy): specify candidate resolution, aggregation, and ingress codes

  #130·Noah-Tervalon-Nvidia opened Sep 28, 2026Collaborator·6 of 7 tasks·

  ·
  2

• docs(nvpair-errors): add normative service specification

  #129·Noah-Tervalon-Nvidia opened Sep 28, 2026Collaborator·6 of 7 tasks·

  ·
  2

• Add managed vLLM with multi-node serving over direct and ring fabrics, and device setup

  #125·nv-pgoode opened Sep 28, 2026Collaborator·7 tasks done·

  ·

• Publish installed engine CLIs on the user's PATH

  #113·ckelseynv opened Sep 22, 2026Collaborator·

  ·
  32

• Add model download cancellation and progress

  #111·ckelseynv opened Sep 22, 2026Collaborator·7 tasks done·

  ·

  · 1.0.x
  20

• node-info: keep every GPU at boot when the DirectX registry is stale, and recover an empty Windows inventory

  #98·dmmdea opened Sep 20, 2026·7 tasks done·

  ·