Issues

• refactor(frontend): extract the HTTP server core and separate chat inference

  dynamo-runtimeenhancementfrontendlanguage::rustsig-frontend
  ai-dynamo/dynamo#17177 · Filter by author sachalmalick opened on Oct 23, 2026

• test(operator): capture existing upgrade fixtures from the v1.6.0 release

  deployment::k8sdynamo-deployenhancementoperator
  ai-dynamo/dynamo#17175 · Filter by author xianlubird opened on Oct 23, 2026

• Worker-selection policy for uncached-compute pool split (taint-declared tiers, overload-classified rejection)

  enhancementlanguage::rustperformanceroutersig-router
  ai-dynamo/dynamo#17173 · Filter by author jokerhaos opened on Oct 23, 2026

• [BUG]: v1.5-to-v1.6 upgrade rejects previously accepted strict tool schemas

  bugfrontendlanguage::rusttool-calling
  ai-dynamo/dynamo#17164 · Filter by author cmdy opened on Oct 23, 2026

• Deprecate the operator-generated Ray launch for multinode vLLM

  backend::vllmdeployment::k8sdynamo-deployenhancementoperator
  ai-dynamo/dynamo#17158 · Filter by author tmonty12 opened on Oct 23, 2026

• Automate the main version bump after a release branch cut

  Automationci
  ai-dynamo/dynamo#17151 · Filter by author pvijayakrish opened on Oct 22, 2026

• docs: add TP2 GPU fractions reproduction guide

  documentation
  ai-dynamo/dynamo#17148 · Filter by author scheckerNV opened on Oct 22, 2026

• [BUG]: Operator counts nvidia.com/gpu.shared as 0 GPUs, unlike nvidia.com/gpu and MIG .shared resources

  bugdeployment::k8soperator
  ai-dynamo/dynamo#17138 · Filter by author richajoy opened on Oct 22, 2026

• DEP: Deadline-aware decode regulation for interactive requests under agent load

  agentsbackend::vllmenhancementperformancepython
  ai-dynamo/dynamo#17137 · Filter by author anjalibshah opened on Oct 22, 2026

• [BUG]: Hybrid-model KV events quarantine the KVCR cache owner

  backend::vllmbugdynamo-llmlanguage::rustroutersig-router
  ai-dynamo/dynamo#17124 · Filter by author karya0 opened on Oct 22, 2026

• Add an experimental aggregated BF16 Flint recipe for vLLM

  backend::vllmtool-calling
  ai-dynamo/dynamo#17119 · Filter by author elizabetht opened on Oct 22, 2026

• fix(multimodal): bound custom tensor expansion and async worker preparation

  buglanguage::pythonmultimodalperformance
  ai-dynamo/dynamo#17113 · Filter by author pstjohn opened on Oct 22, 2026

• DEP: Separate parser capture history from frontend-crates until production database rollout

  dep:draftfrontendsig-frontend
  Feature

  ai-dynamo/dynamo#17112 · Filter by author keivenchang opened on Oct 22, 2026
  

• Update bundled dependencies and image bases

  buildcherry-pick:approveddependencieslanguage::rust
  ai-dynamo/dynamo#17106 · Filter by author sara4dev opened on Oct 22, 2026

• [CI] Recipe tests / Kimi K2.5 disaggregated (nightly): workers can't resolve nvidia/Kimi-K2.5-NVFP4 from shared-model-cache offline, readiness times out

  backend::trtllmbugci
  ai-dynamo/dynamo#17102 · Filter by author devin-ai-integration opened on Oct 22, 2026

• fix(frontend): close HTTP/1.1 connections rejected during drain

  deployment::k8sfault tolerancefrontendlanguage::rust
  ai-dynamo/dynamo#17100 · Filter by author xianlubird opened on Oct 22, 2026

• Serve LPX engines through a Dynamo OpenEngine sidecar

  deployment::k8sdynamo-runtimeenhancementoperatorruntime
  ai-dynamo/dynamo#17098 · Filter by author andrewpaprotsky opened on Oct 22, 2026

• Global Router: route pools by client-declared uncached-token estimate (nvext.router.uncached_tokens)

  enhancementlanguage::rustperformanceroutersig-router
  ai-dynamo/dynamo#17094 · Filter by author jokerhaos opened on Oct 22, 2026

• [FEATURE]: Expose Mooncake soft-pin commands to custom KV router policies

  enhancementkv-memoryrouter
  ai-dynamo/dynamo#17092 · Filter by author magicYang1573 opened on Oct 22, 2026

• Planner diagnostics report: zooming one subplot does not zoom the others

  enhancementplannerpython
  ai-dynamo/dynamo#17089 · Filter by author potsju opened on Oct 22, 2026

• fix(vllm): preserve one-token prompt logprobs and native token labels

  backend::vllmfrontendlanguage::rust
  ai-dynamo/dynamo#17088 · Filter by author Aphoh opened on Oct 22, 2026

• [FEATURE]: Support prefix-aware prefill latency in mocker NPZ profiles

  dynamo-llmFeaturelanguage::rustperformance
  ai-dynamo/dynamo#17085 · Filter by author zzulihrs opened on Oct 22, 2026

• DEP: Runtime-coordinated worker admission for rolling updates

  dep:draftdeployment::k8sdynamo-runtimeenhancementfault-tolerancefrontendrouterruntime
  ai-dynamo/dynamo#17083 · Filter by author xianlubird opened on Oct 22, 2026

• DEP: KV-preserving migration of active decode requests during scale-down

  enhancementfault tolerancefault-tolerancerouterruntime
  ai-dynamo/dynamo#17073 · Filter by author qianlihuang opened on Oct 22, 2026

• [BUG]: Muse Spark Open — Rust vLLM sidecar reasoning-gate metadata blocks model registration

  bugdynamo-llmlanguage::rusttool-calling
  ai-dynamo/dynamo#17071 · Filter by author indrajit96 opened on Oct 22, 2026