Project-wide deployment is dependency-aware. RailDock builds a service graph and uses topological ordering so prerequisites can be dispatched before services that depend on them.

123postgres ─┐
          ├──> api ──> web
redis ────┘

A service graph makes relationships explicit and lets RailDock avoid an arbitrary loop over service records. This is especially useful when applications consume connection data from declared datastore links.

The project API includes cancellation of active project deployments, while individual deployment resources expose cancellation as well. After a failed sequence, inspect the failing service logs and state before retrying the whole graph.