Fibril documentation
Fibril is a lightweight message broker focused on durable delivery, explicit acknowledgements, leasing, retries, and asynchronous workflow coordination. It also has partitioned queues for scale, exclusive consumer groups for ordered parallel consumption, Plexus streams for fan-out where every subscriber sees every record, and an experimental clustered mode with partition ownership, replication, and failover for both queues and streams. The broker is implemented in Rust, but the user-facing model is about durable messaging rather than a Rust-only ecosystem.
It is early-stage (0.x). The useful baseline works and is tested hard, but APIs, persistence formats, protocol details, and operational behavior can still change between minor versions. The clustering and replication paths are experimental and not yet production-ready high availability.
Start using Fibril
Section titled “Start using Fibril”- Follow the quickstart to run a broker, then use the client guide for Rust, TypeScript, Python, Go and C# examples.
- Read the core model for queues, consumer groups for ordered parallel consumption and Plexus streams for fan-out.
- Explore the admin dashboard and its embedded demos, or try a cluster with Docker.
Watch the architecture in motion
Section titled “Watch the architecture in motion”The animated stories explain decisions and ordering one step at a time. Each has playback controls, a transcript and Save frame for presentations. Their timing is illustrative.
- Explore partition placement: follow queue owners and replicas across brokers.
- Follow the message path: compare delivery modes and the work that must finish before confirmation.
- Follow a failover: watch owner loss, recovery, activation and rejoin, including a blocked attempt.
- Follow an agreed checkpoint: see how replicas establish a shared recovery starting point.
Operate and evaluate
Section titled “Operate and evaluate”- Configure TLS and users or deploy a secured cluster, then set up monitoring.
- Review delivery guarantees, retries and delays and dead lettering for behavior and limits.
- Use failure modes and recovery quarantine when troubleshooting.
- Consult benchmarks, backpressure and many idle queues for workload planning.
Follow development
Section titled “Follow development”Project status summarizes maturity. Implemented surface details supported paths and conditions. The roadmap and failover plan describe remaining work. Engineering notes and the optimization log record fixes and investigations. The changelog records changes.
Versions
Section titled “Versions”These pages track the active pre-1.0 codebase. The version picker links to historical documentation snapshots, starting with /0.2/.