Three deployment options, all covering the same zisk-coordinator and zisk-worker binaries. Pick the binaries-directly path for a developer machine or CI runner; pick the services path when the prover needs to run 24/7 under a real service manager (systemd on Linux, launchd on macOS); pick Docker for a containerised stack with Prometheus and Grafana bundled in.
Lightweight
Deploy the binaries directly
Install zisk-coordinator and zisk-worker on your PATH via ziskup or build them from source. Launch as foreground processes — no service manager involved.
Read more →Production
Deploy as services
Run the binaries under systemd (Linux) or launchd (macOS) using the upstream install scripts. Dedicated service users, /etc/zisk config, hardened unit files.
Read more →Containers
Deploy with Docker
One docker compose up brings the coordinator, a scalable worker pool, Prometheus, and Grafana online. Best for demos, CI, and single-host staging.
Read more →Which to pick?
| Scenario | Path |
|---|---|
| Developer laptop, smoke-testing the prover | Binaries directly |
| CI runner that spins a cluster up for a few minutes | Binaries directly or Docker |
| Contributor on a feature branch / custom features | Binaries directly — source build |
| Demo or single-host staging with Prometheus + Grafana attached | Docker |
| Bare-metal or VM host running the coordinator 24/7 | As services |
| Worker fleet that needs to register on boot | As services |
| You want hardened units, dedicated service users, journald routing | As services |