init

docs/source/serving/metrics.rst

@@ -0,0 +1,13 @@
+Production Metrics
+==================
+
+vLLM exposes a number of metrics that can be used to monitor the health of the
+system. These metrics are exposed via the `/metrics` endpoint on the vLLM
+OpenAI compatible API server.
+
+The following metrics are exposed:
+
+.. literalinclude:: ../../../vllm/engine/metrics.py
+    :language: python
+    :start-after: begin-metrics-definitions
+    :end-before: end-metrics-definitions