How guoyoudu/docker-jitsi-meet works
Project type and purpose
• This is a container‐based, multi-service deployment of Jitsi Meet—a fully open‐source, encrypted video‐conferencing solution—packaged via Docker and Docker Compose.
• The repository provides one Docker image per Jitsi component (web, prosody, jicofo, jvb, jigasi, jibri, transcriber) along with an integrated logging/monitoring stack (Prometheus, Loki, Grafana).
File structure
• Top‐level orchestrator:
• docker-compose.yml – defines the services and inter‐container networks.
• Service directories: each subfolder (web, prosody, jicofo, jvb, jigasi, jibri, transcriber, log-analyser, prometheus) contains:
• Dockerfile – builds the container image
• rootfs/etc/services.d – run scripts & init hooks
• rootfs/defaults – default config files
• Monitoring:
• prometheus/ and log-analyser/ folders with configs for Prometheus, Loki, Grafana provisioning
• Base images: shared “base” and “base-java” Dockerfiles for repeatable build environments.
• CI/CD workflows in .github/workflows for build, test, release pipelines.
README for additional insights
• Confirms use of Docker Compose to spin up a full Jitsi stack (web frontend, XMPP signaling, conferencing focus, media router, SIP gateway, recording, transcription).
• Mentions multi‐arch support (amd64, arm64) and linkage to Kubernetes tooling (external).
• No built‐in TURN server yet—external TURN/STUN needs to be added for NAT traversal.
a. Main components
• Client browsers/mobile apps
• “web” container: HTTPS → Nginx + Prosody WebSockets proxy, static assets, config.js
• “prosody” container: XMPP signaling, authentication, MUC (multi‐user chat) rooms
• “jicofo” container: Conference focus component, XMPP control
• “jvb” container: VideoBridge—media router handling RTP/ICE
• “jigasi” container: SIP gateway (audio calls with PSTN/SIP)
• “jibri” container: Recorder & SIP gateway for recording/streaming
• “transcriber” container: Transcription sidecar
• Monitoring stack: Prometheus (scrapes exporters/sidecars), Loki (logs), Grafana (dashboards)
• External services: TURN/STUN server, OAuth / JWT auth providers (optional)
b. Relationships and interactions
• Clients ⇄ web (HTTPS/WSS)
• web ⇄ prosody (WS XMPP over internal network)
• prosody ⇄ jicofo/jigasi/jibri (XMPP pub/sub RPC control)
• jicofo ⇄ jvb (XMPP commands to allocate channels)
• Clients ⇄ jvb (media: RTP/UDP, ICE negotiation via STUN/TURN)
• jigasi/jibri communicate via XMPP and spawn local headless processes
• All containers expose Prometheus metrics endpoints; log outputs are shipped to Loki
• Prometheus scrapes, Grafana queries Prometheus & Loki
c. Architectural patterns and principles
• Microservice per functionality
• Sidecar pattern for healthchecks, autoscaler‐sidecar, config/init hooks
• Configuration‐as‐code: defaults in rootfs/defaults, overrides via environment variables
• Orchestration via Docker Compose networking
• Observability: metrics + logs + dashboards
d. Key technologies, frameworks, libraries
• Docker, Docker Compose
• Alpine Linux-based custom rootfs with s6‐overlay init system
• Nginx, Prosody XMPP, Jicofo (Java), JVB (Java), Jigasi (Java), Jibri (Java + Chrome), optional transcription engine
• Prometheus, Grafana, Loki, OpenTelemetry Collector
Final advice
• Every container as a separate component.
• Clearly separate control plane (XMPP, APIs) from data plane (media).
• Sidecar/healthcheck containers or processes if relevant to deployment logic.
• The goal is a comprehensive, at‐a‐glance view of how all pieces interconnect under Docker Compose.