What is leviathan?

Leviathan is a container orchestration platform built entirely from scratch in Rust, similar in spirit to Kubernetes but much smaller. The README is explicit that this is not a beginner tutorial project but a structured systems engineering exercise, meant to build distributed consensus, container isolation, resource scheduling, and self healing networking without relying on existing frameworks to do the hard parts. The system is organized around a control plane that holds cluster state and talks to a scheduler for placing workloads, a Raft consensus cluster for leader election and replicating logs across nodes, a write ahead log storage engine using memory mapped files, Prometheus based metrics, and a service discovery layer for internal networking. Worker nodes run the actual containers, which are isolated using Linux namespaces and cgroups v2, the same low level mechanisms real container runtimes use. Operators interact with the whole system through a command line tool built with the Clap library. The README documents the project as a seven day build roadmap, with each day covering a distinct layer: project setup and core types on day one, an async runtime using Tokio on day two, TCP networking and serialization on day three, the write ahead log storage engine on day four, Raft consensus on day five, the container runtime using Linux namespaces and cgroups on day six, and the scheduler plus observability on day seven. All seven days are marked completed. For each phase, the README also lists the specific Rust language features and libraries it exercises, such as async and await, tokio channels, TCP framing, unsafe Rust and raw system calls for container isolation, and bin packing algorithms for scheduling. The project is structured as a Cargo workspace with separate crates for the command line interface and other components, and the README includes a partial breakdown of that folder structure. It does not include install or run instructions beyond describing the workspace layout, since it appears aimed at readers studying the code rather than deploying it. The project is licensed under the MIT license, based on the badge shown at the top of the README.