Commit Verlauf

Autor SHA1 Nachricht Datum
  Alexander Korolev 1b4f49aaa6 feat: bump to vllm-gb10 v0.24.0-gb10.5 base (vLLM 0.24.1 + b12x) vor 3 Monaten
  Alexander Korolev 421d123a31 feat: GLM-5.2 NF3-Hybrid image builds successfully (22GB, vLLM 0.23.1 + b12x) vor 3 Monaten
  Alexander Korolev f5e65e3338 feat: vllm-gb10 based Dockerfile for GLM-5.2 NF3-Hybrid vor 3 Monaten
  Alexander Korolev fc2593bb0b feat: rewrite GLM-5.2 NF3-Hybrid Dockerfile (single-stage, local repos, real overlay kernels) vor 3 Monaten
  Alexander Korolev 15698ba6b7 feat: add GLM-5.2 NF3-Hybrid recipes (context/fast lanes) + build system vor 3 Monaten
  Alexander Korolev 31aaf141e5 refactor: split ends (HA/SSH) from control logic with traits and orchestrator vor 3 Monaten
  Alexander Korolev 56bcdf4bf5 feat(recipe): update deepseek-v4-flash-dspark from NVIDIA reference vor 3 Monaten
  Alexander Korolev 66b9879240 fix: auto-power-on now powers on per-node entities, not just cluster extras vor 3 Monaten
  Alexander Korolev 05f923dfaf fix: add served-model-name with two names to dspark recipe vor 3 Monaten
  Alexander Korolev 732cef4a04 fix: remove served-model-name and custom port from dspark recipe vor 3 Monaten
  Alexander Korolev d6cd9196d4 feat: add deepseek-v4-flash-dspark recipe vor 3 Monaten
  Alexander Korolev 7158e63649 refactor: separate cluster state from SSH probing for testability vor 3 Monaten
  Alexander Korolev f54e192ef7 fix: mark unreachable nodes as unavailable in cluster selection vor 3 Monaten
  Alexander Korolev ff23a80a81 feat: extend NcclConfig with per-cluster NCCL overrides vor 3 Monaten
  Alexander Korolev ae1285dfdf fix: reduce gpu-memory-utilization to 0.75 for deepseek-v4-flash vor 3 Monaten
  Alexander Korolev 04e04b4e8c fix: add initial boot delay and retry logic to auto-power-on vor 3 Monaten
  Alexander Korolev 9123df86cd fix: use original cluster name for solo instances instead of 'solo/<node>' vor 4 Monaten
  Alexander Korolev 27c02331d8 refactor: rename --power-on-if-off to --auto-power-on vor 4 Monaten
  Alexander Korolev 811b8b0848 feat: add --power-on-if-off and --power-off flags vor 4 Monaten
  Alexander Korolev 176946c99f feat: power off command stops containers first with timeout vor 4 Monaten
  Alexander Korolev 98abe64703 fix: address soundness issues from audit vor 4 Monaten
  Alexander Korolev 0ae454c259 feat: add --node parameter to logs command vor 4 Monaten
  Alexander Korolev 4b08863c18 chore: update qwen-agentworld-35b recipe with tool call args and model name vor 4 Monaten
  Alexander Korolev 81f43858be feat: health checks NCCL socket interface status per node vor 4 Monaten
  Alexander Korolev 0c19edf62d feat: health checks HA cluster entities (switch) to detect switched-off clusters vor 4 Monaten
  Alexander Korolev 604f61491b feat: improved health command with cluster-scoped container view vor 4 Monaten
  Alexander Korolev cefdfa2368 feat: deployment discovery + stop ambiguity resolution vor 4 Monaten
  Alexander Korolev 3774e765e5 feat: cluster selection with occupancy awareness vor 4 Monaten
  Alexander Korolev ea8fb4e287 fix: select_cluster should use >= not == for required_nodes matching vor 4 Monaten
  Alexander Korolev 2513ac6f85 feat: add Qwen-AgentWorld-35B-A3B recipe (2× DGX Spark, TP=2) vor 4 Monaten