Issues

• [Bug]: Streaming SGLang errors are relayed with an incorrect SSE content type

  bugtriage
  LLM360/smg#74 · Filter by author mvillmow opened on Sep 1, 2026

• [Bug]: least-load cache credit can override load balance

  bugtriage
  LLM360/smg#71 · Filter by author DavidBellamy opened on Aug 12, 2026

• [Performance]: prevent cache affinity from creating hot engines

  performancetriage
  LLM360/smg#68 · Filter by author DavidBellamy opened on Aug 12, 2026

• fix(router): commit K3 distribution seeds after streaming completion

  LLM360/smg#65 · Filter by author DavidBellamy opened on Aug 11, 2026

• fix(scheduler): arm held capacity credits at dispatch

  LLM360/smg#64 · Filter by author DavidBellamy opened on Aug 11, 2026

• fix(admission): seed clean peers under cache-owner pressure

  LLM360/smg#61 · Filter by author DavidBellamy opened on Aug 11, 2026

• fix(scheduler): make external fair-share credits work-conserving

  LLM360/smg#60 · Filter by author DavidBellamy opened on Aug 11, 2026

• feat(admission): wire scheduler capacity credits

  LLM360/smg#58 · Filter by author DavidBellamy opened on Aug 8, 2026

• Worker activation can leave aggregate capacity stale

  LLM360/smg#56 · Filter by author DavidBellamy opened on Aug 7, 2026

• feat(scheduler): add single-use capacity credit contract

  LLM360/smg#53 · Filter by author DavidBellamy opened on Aug 7, 2026

• fix(scheduler): revert fair-share admission gating after streaming regression

  LLM360/smg#51 · Filter by author DavidBellamy opened on Aug 6, 2026

• fix(grpc): classify adaptive admission rejections accurately

  LLM360/smg#49 · Filter by author DavidBellamy opened on Aug 6, 2026

• fix(scheduler): keep tenant queue caps opt-in

  LLM360/smg#48 · Filter by author DavidBellamy opened on Aug 6, 2026

• fix(admission): dispatch fair-share queues at engine-feedback capacity

  LLM360/smg#45 · Filter by author DavidBellamy opened on Aug 6, 2026

• KV event gaps trap cache-aware workers in a replay reconnect loop

  LLM360/smg#42 · Filter by author DavidBellamy opened on Aug 5, 2026

• feat(admission): add bounded prediction samples

  LLM360/smg#35 · Filter by author DavidBellamy opened on Aug 4, 2026

• feature: adaptive model admission from predicted token work and engine pressure

  LLM360/smg#30 · Filter by author DavidBellamy opened on Aug 3, 2026

• Make priority admission queue work-conserving and add Retry-After estimates

  LLM360/smg#23 · Filter by author DavidBellamy opened on Aug 2, 2026

• Expose priority scheduler settings through Python router entrypoint

  LLM360/smg#17 · Filter by author DavidBellamy opened on Jul 31, 2026

• [Feature]: make cache-aware routing engine-pressure aware

  enhancementtriage
  LLM360/smg#11 · Filter by author DavidBellamy opened on Jul 28, 2026
  1