ワンクリックで
dgx-spark-bare-metal
dgx-spark-bare-metal には urbanspr1nter から収集した 3 個の skills があり、リポジトリ単位の職業カバレッジとサイト内 skill 詳細ページを表示します。
このリポジトリの skills
Install or reinstall vLLM across the entire 4× DGX Spark cluster. Builds vLLM from source on all nodes, using the setup/install_vllm.sh script. Handles rsync-ing the repo, kicking off builds via tmux, and verifying the binary exists on every node afterward.
Start the Ray cluster across all 4 DGX Spark nodes. Copies the per-node start-ray scripts, starts spark-01 as head, then joins spark-02/03/04 with 10-second pauses between each. Verifies cluster health with ray status afterward.
Run diagnostics on the 4x DGX Spark cluster. Checks node reachability via SSH, verifies model directories exist on all nodes, and validates that paths referenced in model run scripts actually exist on each node. Use before starting Ray or running models, or when troubleshooting cluster issues.