How it works
A push in. A release out.
A control plane holds your apps and history. One small agent on each server does the work. Nothing between the two but your own network.
01 / the pipeline
Verify, then swap.
Every deploy runs the same six stages. The version serving traffic keeps serving it until the new one passes its health check — through the real proxy, over real HTTPS.
- A failed probe rolls back and swaps nothing
- Five clean checks before traffic moves
- The old container is kept for one-click rollback
- 01 Clone Shallow clone at your ref.
- 02 Build BuildKit, layer cache shared across the fleet.
- 03 Release Your release task runs — migrations included.
- 04 Probe Readiness check on the new container.
- 05 Swap Traffic cuts over after five clean HTTPS probes.
- 06 Retire The old container stops. Rollback keeps it a click away.
02 / the fleet
Your servers, one board.
Attach a machine over SSH or have Railyard provision one in your cloud account. Each runs a single agent that heartbeats CPU, memory and disk back to the control plane.
- Cordon a server and placement skips it
- Detach one and the fleet keeps running
- Dedicated build servers, if you want them
03 / the logs
Watch it happen.
Build output and application logs stream to your browser as they're written — timestamped, tagged by stream. Open a shell on the running container without leaving the page.
Your next deploy could be on a server you own.
Attach a machine with a host, a user and an SSH key. Railyard installs the rest and watches it come online.
Free for one server, forever. No credit card.