a month ago
We have three accepted Railway-built staging deployments (Memory, Hub, and relay) and approval to deploy the same frozen candidate dark in a separate production environment. We need the production services to run the exact accepted image digests, not merely rebuild the same Git commits. Production federation flags must remain off.
Is there a supported Dashboard, CLI, or public API procedure to attach/promote a private Railway-built image from a staging deployment to a NEW service in another environment without rebuilding? If so, please give the steps and how to verify the resulting deployment image digest matches the source. Does Environment Sync preserve the image, or only configuration? Can Skipped Builds be relied on for digest identity?
If exact cross-environment image promotion is unsupported, is the recommended path to publish immutable images to an external registry, accept those digests in staging, and then deploy by digest to production? Please do not alter any service or deployment; we are requesting a documented procedure only. We can provide project/deployment IDs privately if needed.
3 Replies
a month ago
We don't have a Dashboard, CLI, or API flow that attaches a Railway-built image from one deployment to a new service in another environment. Skipped Builds can't guarantee this either. It only reuses a cached image from that same service's own build history, so a brand-new production service builds independently.
Environment Sync imports services and their configuration as staged changes into the receiving environment. Applying those changes deploys the services there. It doesn't move a specific deployed image digest from the source environment.
For exact digest identity, the path that fits is the one you described. Build and publish immutable images to your own registry, deploy those images by digest to staging, and once they're accepted, deploy the same digests to the production services as image-source services. The image reference you set on each production service then names the exact digest you accepted. Federation flags stay in that environment's own variables, so you keep them off in production.
Status changed to Awaiting User Response Railway • about 1 month ago
Railway
We don't have a Dashboard, CLI, or API flow that attaches a Railway-built image from one deployment to a new service in another environment. Skipped Builds can't guarantee this either. It only reuses a cached image from that same service's own build history, so a brand-new production service builds independently. Environment Sync imports services and their configuration as staged changes into the receiving environment. Applying those changes deploys the services there. It doesn't move a specific deployed image digest from the source environment. For exact digest identity, the path that fits is the one you described. Build and publish immutable images to your own registry, deploy those images by digest to staging, and once they're accepted, deploy the same digests to the production services as image-source services. The image reference you set on each production service then names the exact digest you accepted. Federation flags stay in that environment's own variables, so you keep them off in production.
a month ago
Thank you. This is an audited release, so please have a Railway support engineer confirm that the limitation applies to a private Railway-built staging image being attached to a NEW service in another environment, and that Environment Sync and Skipped Builds cannot guarantee the source digest. Please also confirm the supported image-source procedure for pinning an externally published immutable digest in both environments, including any private-registry authentication requirement and the deployment field we should read back to verify the digest. We will not change production until the procedure is validated in staging. No Railway-side changes are requested.
Status changed to Awaiting Railway Response Railway • about 1 month ago
a month ago
That's correct on all three points. There's no Dashboard, CLI, or API flow that attaches a Railway-built image from a staging deployment to a new service in another environment. Environment Sync carries services and their configuration as staged changes, not a deployed image digest. Skipped Builds only reuses an image from that same service's own build history, so a new production service builds on its own and can't be relied on for digest identity.
The supported path is an image-source service in each environment, with the source set to the full registry path pinned by digest, for example ghcr.io/your-org/memory@sha256:<digest>. Image sources accept digest references and deploy that exact digest. For a private image, add credentials in the service's Settings, under the Source section, in Registry Credentials. For GHCR that's a personal access token (classic). Note that private registry deployments require the Pro plan. Public images need no credentials.
To verify, read meta.imageDigest on each deployment through the public GraphQL API and compare the staging and production values. Federation flags stay in each environment's own variables, so production keeps its own values.
Status changed to Awaiting User Response Railway • about 1 month ago
Status changed to Solved bbshocking • about 1 month ago