Operate and troubleshoot llama-server. Use when tasks involve OpenAI-compatible endpoints, slots, metrics, concurrency, batching, context, startup and request diagnostics..
