Thanks — I see the traceback and you’re hitting an

AssignmentTimeoutError

when the worker calls

accept()

on the job request. I’ll walk through what that means and give focused troubleshooting steps you can try. What this error means • The agent attempted to accept a job but the assignment to a room did not complete before the worker’s assignment timeout, so

_on_accept

raised

AssignmentTimeoutError

. • This is typically an availability/latency issue (agent server not responding, overloaded, or connectivity problems) rather than a bug in your job logic. Quick checklist to triage and fix 1. Confirm the agent process is healthy ◦ Ensure the agent worker process is running and responsive on the host. ◦ Check agent logs around the timestamp of the error for any prior errors, slow startup, or crashes. 1. Check resource and load issues ◦ Verify CPU, memory, and network usage on the agent host. High load can delay handling of assignments. ◦ If you’re running multiple agents, confirm you're not exhausting available agent capacity. 1. Verify connectivity to LiveKit ◦ Ensure the agent can reach the LiveKit control plane (no firewall, DNS, or routing issues). ◦ If you use TLS or tokens, verify certificates and auth tokens are valid and not expired. 1. Confirm deployment / scaling configuration ◦ If you use auto-scaling or a limited pool of agents, ensure there are enough agents to accept incoming jobs. ◦ Increase agent instances or adjust timeouts if assignments can legitimately take longer during peak load. 1. Review timeout settings and retry logic ◦ If your environment is expected to have variable latency, consider increasing the assignment timeout if configurable. ◦ Add or tune retry/backoff logic around accept if the worker library supports it. 1. Look for entrypoint or initialization delays ◦ If the agent performs heavy work on startup (large imports, DB connections, warming caches), that can delay readiness. Move long initialization out of the critical path if possible. Targeted diagnostics to run now • Tail the worker logs while reproducing the failure to capture the full sequence leading up to the timeout. • From the agent host, run a connectivity check (ping/traceroute and a simple request) to the LiveKit endpoint used by your agents. • Check system metrics (CPU/memory) over the timeframe of the error and compare to normal operation. If you can share any of the following, I can be more specific: • Are you running agents on LiveKit Cloud or self-hosted? (This affects how you check scaling/health.) • Recent agent logs around the error (a few lines before and after the traceback). • Whether this started after a deployment or traffic spike. I’m here to help — share the logs or answer the above questions and I’ll suggest the next concrete steps. Sources: Deployment reliability on LiveKit Cloud | LiveKit | Python Agents: Handling exceptions in your entrypoint | LiveKit | Self-hosted deployments | LiveKit Documentation