Deployment statuses
The labels a deployment shows on the card, in the panel, in the CLI and in webhooks, and what each one means.
| Status | Label | Meaning |
|---|---|---|
QUEUED | Initializing | Waiting for a build host or for capacity |
BUILDING | Building | The image is being built (the sub-line shows the step) |
DEPLOYING | Deploying | Replicas are starting and being health checked |
SUCCESS (active) | Active | Serving now ("Deployed 3m ago via GitHub") |
SUCCESS (older) | Success | Succeeded earlier and was replaced |
COMPLETED | Completed | A one-off or cron run exited with code 0 |
CRASHED | Crashed | Exhausted its restarts; Restart is offered |
FAILED | Failed | Build, pre-deploy or health check failed; the diagnosis sentence says why |
REMOVING | Removing | Being stopped |
REMOVED | Removed | Stopped and replaced |
CANCELLED | Removed | Aborted by a person, superseded by a newer deployment, or timed out in the queue; the sub-line says which |
SKIPPED | Skipped | "No changed files matched the watch paths" |
The labels match Railway's. Railway has no visible cancelled state, so an aborted or superseded deployment shows Removed with the reason.
Restarts
The restart policy decides when a crashed container restarts: on_failure (default, up to 10 times), always or never. Restarts work even while the agent is down, because Docker performs them. When on_failure runs out, the deployment becomes Crashed. Out-of-memory kills say "Out of memory at X MB (limit)".