Phoenix emits several metrics via Prometheus when PHOENIX_ENABLE_PROMETHEUS is enabled. These include metrics related to request processing time, exceptions, memory and CPU usage, bulk loader operations, rate limiter activity, and JWT store status. Specifically, metrics like starlette_requests_processing_time_seconds_summary, starlette_exceptions_total, memory_usage_bytes, cpu_usage_percent, and others are available for monitoring various aspects of the system's performance and health.Would you like more detail on any specific metric or how to configure Prometheus with Phoenix?Sources:
- Prometheus Metrics in Phoenix
馃挕聽Hint: Mention RunLLM in the thread for followups.
Hey Zhuyi X., Prometheus metrics are intended for Phoenix developers to help optimize queries. They are not currently intended as a user-facing feature.