what does aec_warmup_duration do?

Thanks for reaching out, Utkarsh. I'm an AI assistant and I'm here to help. Give me a moment while I research this topic.

how to add prometheus port to the worker

Which worker do you mean — the LiveKit Server, an Agent worker, or the Egress worker? Also tell me how you’re running it (self-hosted Docker, k8s Helm chart, or from source) so I can give exact steps to expose a Prometheus

/metrics

port. Sources: Server options | LiveKit Documentation, with interactive elements

an agent worker

Thanks — responding to both items:

aec_warmup_duration

and how to expose Prometheus metrics on an agent worker. • `aec_warmup_duration`: This sets how long (in seconds) the agent will ignore user audio interruptions after the agent starts speaking to avoid being interrupted by echo while AEC warms up. Default is 3.0 seconds; setting it to

None

disables the warmup. • Prometheus on agent worker: The agent worker does not expose a built-in metrics HTTP endpoint. To expose

/metrics

you must run a Prometheus HTTP server in the main process (use

prometheus_client.start_http_server(port)

) and ensure child-process metrics are handled correctly (call

_update_child_proc_count()

periodically because

set_function()

isn’t supported in multiprocess mode). The worker’s existing `host`/`port` options only apply to the health check endpoint, not metrics, so this custom step is required. If you want, tell me: • which port you’d like to use, and • how you run the agent (Docker, k8s, or from source) and I’ll give exact code/snippet and deployment steps. , with interactive elements