rapid-painting-51133
12/22/2025, 10:23 PMtall-airline-16756
12/22/2025, 10:23 PMtall-airline-16756
12/22/2025, 10:23 PMdelete_room API, all participants (including agent sessions) are disconnected, which starts the session shutdown sequence; any shutdown hooks registered with ctx.add_shutdown_callback() run as part of that teardown. This applies whether you delete the room before or after calling session.shutdown().
Sources: Job lifecycle | LiveKit docsrapid-painting-51133
12/22/2025, 10:33 PMtall-airline-16756
12/22/2025, 10:33 PMdelete_room API, an SDK call, or some other process?
2. Do you call session.shutdown() anywhere before/after deletion, or do you rely on deletion to trigger the session shutdown?
3. Is the behavior reproducible in certain environments (local, staging, production) or with certain room sizes/participant counts?
4. Do you see any errors or logs from the agent/service around the time of deletion (disconnects, timeouts, process exits)? If so, please share relevant log snippets and timestamps.
5. What LiveKit version and SDK/runtime are you running for the agent that registers ctx.add_shutdown_callback()?
With those details I’ll suggest targeted checks and next steps (common causes include abrupt process termination, race conditions where the process exits before callbacks run, or deletion via a path that forces immediate disconnects).