v0.4.0

IMPORTANT This version might require clearing or updating env vars when running from source on macOS. See docs for more.

Features:

• added peer websocket close reason - now, server closes peer websocket with a reason e.g. "Internal server error", "Peer removed" or "Room stopped"
• added support for path in JF_HOST env var
• added framerate option in the File component - defaults to 30 FPS
• added healthcheck endpoint - GET /health - requires authorization and returns: status, uptime and information about distribution (enabled/disabled, node status, number of nodes in a cluster).
• added published tracks and their metadata to the REST API e.g. GET /room will return a list of rooms with their components and peers, where each component and each peer will also list all of its published tracks
• added new server notifications: PeerMetadataUpdated, TrackAdded, TrackRemoved, TrackMetadataUpdated
• added peerless room purge option - after predefined amount of time, if a room is empty, it will be automatically closed. This behaviour is disabled by default. See here.
• added SIP component
• added basic structured logging that can be enabled with JF_STRUCTURED_LOGGING
• added python RTSP -> HLS example

Bugfixes:

• fixed duplicated peer_disconnected notification
• fixed prometheus metrics not being reset - now, after room is removed, its metrics are set to 0 - dopytać o interwał
• fixed starting logs not being displayed when JF couldn't run in a distributed mode - now, even if JF crashes because of distribution-related problems, config logs will be displayed (i.e. JF version, WebRTC config, distribution config)
• fixed removing component, which crashed from JF state (thanks to @gfodor)
• fixed removing component, which finisehd its work from JF state
• fixed authorization not required for /hls and /recording endpoints [BREAKING]
• fixed authorization information in OpenAPI spec - previously, all endpoints were marked as requiring authorization, even if they didn't actually need it