I have a question about cleaning-up a phantom session (shows up on dashboard, not in lk room list) caused due to room name reuse. What are my options to get support on this?

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

Give me the instructions to collect exact logs and timestamps

Thanks — I’ll walk you through exact, actionable steps to collect the logs and timestamps that will help us diagnose a phantom session caused by room name reuse. I’ll keep each step short and include what to capture and why it’s important. 1. Capture server-side API call timestamps • What to do: Record the exact timestamps (include milliseconds) for your

DeleteRoom

and

CreateRoom

API calls on the server that manages rooms. • Why it matters: Room ID reuse can happen within a short window (roughly 30–60 seconds). Precise timestamps show whether create/delete calls overlapped. 1. Gather the dashboard link and room identifiers • What to do: Copy the dashboard URL that shows the phantom session and note the Room ID if visible (example format:

RM_nnnnn

) and the room name you reused. • Why it matters: The dashboard link and Room ID let engineers find matching entries in logs quickly. 1. Record timezone and exact event times • What to do: For each relevant event, provide the full timestamp and timezone (e.g., 2026-03-04T143212.345-07:00). Include: ◦ When the original room was deleted (server time). ◦ When the new room was created (server time). ◦ When the phantom session first appeared on the dashboard (approximate time). • Why it matters: Timezone-aligned timestamps remove ambiguity and speed correlation across systems. 1. Stream or export agent/runtime logs • If using LiveKit Cloud or agents: ◦ Use the LiveKit CLI to stream agent logs:

lk agent logs --id <agent-id>

(first run

lk agent list

to find the agent id). ◦ Or download the Agent insights/session logs from the Dashboard (Agent insights tab) in JSON if available. • What to capture: The agent logs covering the time window from a minute before the delete to a minute after the recreate. • Why it matters: Agent logs include lifecycle events and can show whether a previous Room ID was reused or deletion was still in progress. 1. Collect application & SDK logs (client and server) • What to do: From your app servers and any client SDKs involved, collect logs around connection, join, and disconnect events for the same timeframe. Include logs that show: ◦ Room name used ◦ Any reported Room ID (if available) ◦ WebSocket/RTC connect or disconnect messages • Why it matters: Initial connection logs may not include room context until after the WebSocket handshake, so both sides help correlate. 1. Add contextual fields to logs • What to do: Ensure logs include structured fields:

room_name

,

room_id

(if available),

request_id

or

trace_id

, and

user_id

(if applicable). If you can, add a custom correlation id per operation. • Why it matters: Correlation ids let engineers follow a request across services and match entries precisely. 1. If you forward logs to external monitoring • What to do: If you already push logs to Datadog, CloudWatch, Sentry, or similar, export the relevant spans/log entries for the window and include links or exported JSON. • Why it matters: These systems make it easy to search and filter by timestamp, room name, or trace id. 1. Reproduce (if intermittent) while logging is enabled • What to do: Try recreating the sequence (delete then create with same name) while you have verbose logging enabled. Keep the same time-window capture (1 minute before to 1 minute after). • Why it matters: Reproduction while logging gives the clearest evidence of the race condition. 1. Package and share what you collected • What to include: ◦ Dashboard link and Room ID / room name. ◦ All timestamps with timezone and millisecond precision. ◦ Agent logs (raw or JSON) covering the event window. ◦ Server and client SDK logs for the same window. ◦ Any correlation/request IDs or trace IDs. • Why it matters: This package lets the LiveKit team find the exact traces quickly and confirm whether Room ID reuse occurred. If you’d like, I can: • Provide a short checklist you can paste into your logging commands or runbook to ensure you collect everything, or • Give an example of the exact log lines/fields to include (for your language of choice). Which would help you most right now? Sources: Ensuring Unique Room IDs with Dynamic Room Names | LiveKit | Where to find your Agent Logs | LiveKit | How to Send Logs from the JavaScript SDK to External Providers | LiveKit | Job lifecycle | LiveKit Documentation | Still Have Questions