able-jackal-84140
12/23/2025, 7:38 AM07:37:44.479 INFO livekit.agents process exiting
{"reason": "", "pid":
2976389, "job_id": "AJ_deUA58GMMhCv", "room_id":
"RM_i7k8kRrqZBGM"}tall-airline-16756
12/23/2025, 7:38 AMtall-airline-16756
12/23/2025, 7:38 AMAJ_deUA58GMMhCv, room RM_i7k8kRrqZBGM. I can help diagnose — I just need a bit more context.
Could you share the following so I can investigate further?
• Logs from ~1–2 minutes before and after 073744.479 (any WARN/ERROR lines).
• Whether you were doing a deployment, rolling update, or manually shutting down agents at that time.
• Host-level events: were there OOM kills, reboots, or SIGTERM/SIGINT delivered to the process?
• How your agents are deployed (Kubernetes, VM, autoscaling group) and whether draining was enabled.
• If possible, the agent PID (you included 2976389) and the node/host name.
If you prefer, paste the nearby log excerpt here and I’ll scan it for likely causes and next steps.
Sources: Job lifecycle | LiveKit docs | Distributed Setup | LiveKit docs | Deploying to production | LiveKit Docsable-jackal-84140
12/23/2025, 8:14 AMINFO livekit.…echStream Starting message
processing
{"session_id":
139223513363920, "ws_closed": false, "pid": 2976389, "job_id":
"AJ_deUA58GMMhCv", "room_id": "RM_i7k8kRrqZBGM"}
07:37:08.535 DEBUG livekit.agents using audio io:
`RoomIO` ->
`AgentSession` ->
`TranscriptSynchroni…
-> `RoomIO`
{"pid": 2976389,
"job_id": "AJ_deUA58GMMhCv", "room_id": "RM_i7k8kRrqZBGM"}
DEBUG livekit.agents using transcript io:
`AgentSession` ->
`TranscriptSynchroni…
-> `RoomIO`
{"pid": 2976389,
"job_id": "AJ_deUA58GMMhCv", "room_id": "RM_i7k8kRrqZBGM"}
07:37:42.463 DEBUG livekit.agents input stream detached
{"participant":
"user-3fe62c4d", "source": "SOURCE_UNKNOWN", "accepted_sources":
["SOURCE_MICROPHONE"], "pid": 2976389, "job_id":
"AJ_deUA58GMMhCv", "room_id": "RM_i7k8kRrqZBGM"}
07:37:44.468 DEBUG livekit.…echStream Starting stream
cleanup
{"session_id":
139223513363920, "pid": 2976389, "job_id": "AJ_deUA58GMMhCv",
"room_id": "RM_i7k8kRrqZBGM"}
Transcript for room-3fe62c4d saved to artifacts/transcripts/transcript_room-3fe62c4d_20251223_073744.json
DEBUG livekit.…echStream Audio processing
completed, sent 0
chunks
{"session_id":
139223513363920, "pid": 2976389, "job_id": "AJ_deUA58GMMhCv",
"room_id": "RM_i7k8kRrqZBGM"}
07:37:44.469 DEBUG livekit.…echStream WebSocket closed
{"session_id":
139223513363920, "pid": 2976389, "job_id": "AJ_deUA58GMMhCv",
"room_id": "RM_i7k8kRrqZBGM"}
07:37:44.471 DEBUG livekit.…echStream HTTP session closed
{"session_id":
139223513363920, "pid": 2976389, "job_id": "AJ_deUA58GMMhCv",
"room_id": "RM_i7k8kRrqZBGM"}
07:37:44.473 DEBUG livekit.agents session closed
{"reason":
"user_initiated", "error": null, "pid": 2976389, "job_id":
"AJ_deUA58GMMhCv", "room_id": "RM_i7k8kRrqZBGM"}
07:37:44.477 DEBUG livekit.agents shutting down job
task
{"reason": "",
"user_initiated": false, "pid": 2976389, "job_id":
"AJ_deUA58GMMhCv", "room_id": "RM_i7k8kRrqZBGM"}
07:37:44.478 DEBUG livekit.agents http_session():
closing the
httpclient ctx
{"pid": 2976389,
"job_id": "AJ_deUA58GMMhCv", "room_id": "RM_i7k8kRrqZBGM"}
07:37:44.479 INFO livekit.agents process exiting
{"reason": "", "pid":
2976389, "job_id": "AJ_deUA58GMMhCv", "room_id":tall-airline-16756
12/23/2025, 8:14 AMAJ_deUA58GMMhCv / room RM_i7k8kRrqZBGM.
Short answer: this looks like a normal, graceful shutdown. The session closed with "reason": "user_initiated" and "error": null, audio processing finished and the transcript was saved, then the job shutdown ran and the process exited with an empty reason (which in this flow simply means there was no error to report).
If you intended this agent to keep running, a few likely fixes:
• Run agents as a persistent long‑running worker (e.g., use a Deployment/StatefulSet or start with --persistent) instead of per‑job processes so they don’t exit when a session ends.
• Ensure clients aren’t closing the WebSocket or calling session.shutdown() prematurely; the logs show the WebSocket and HTTP session closed just before the session closed.
If you did not expect the session to end, please share one of the following so I can dig deeper:
• Client-side events around 073742–073745 (disconnects or shutdown calls).
• Whether you deploy agents as jobs or persistent workers (Kubernetes Job vs Deployment/VM).
• Any earlier WARN/ERROR lines in the host or agent logs.
Tell me which scenario fits and I’ll provide the exact configuration or debugging steps.