RoomComposite Egress Outage

Incident Report for LiveKit

Resolved

RoomComposite egress requests began failing at 6:35AM. We received an alert at 6:37AM, as well as user reports starting at 6:52AM. The outage lasted until 8:05AM.

The outage was due to a RoomComposite request causing a Chrome failure, creating a bad state on one of our instances. This caused the instance to fail all subsequent RoomComposite requests. Restarting the service fixed the issue, and both our logging and health checking have been improved to ensure this failure loop does not occur again.

Posted Dec 06, 2022 - 06:30 PST