Railway Import

Railway import translates a Railway project into Forge projects and services. It is for moving workloads from Railway into your own server.

What You Need

• A Railway Personal API token.
• A running Forge server.
• Enough disk for cloned apps, Docker images, database volumes, and optional database data imports.
• Public networking enabled for Railway databases if you want Forge to import data automatically.

Railway internal hosts ending in .railway.internal are only reachable inside Railway. Forge rejects those for Postgres and Redis data import and asks for a public URL instead.

Import Flow

The import modal walks through four steps:

1. Enter Railway API token.
2. Select a Railway project.
3. Choose environment and migration options.
4. Run the import.

Forge reads Railway environments, service instances, service variables, source metadata, repo triggers, command overrides, root directory, custom domains, and detected service images when Railway exposes them.

Supported Railway Service Shapes

Forge can recreate:

• Git services.
• Docker image services.
• Database services.

Database detection is based on Railway image and service metadata. Forge recognizes PostgreSQL, TimescaleDB, MySQL/MariaDB, Redis, MongoDB, and ClickHouse shapes.

Services that are not Git services, Docker image services, or recognized databases are marked unsupported instead of being guessed.

Migration Options

Exclude RAILWAY_* variables is enabled by default. Keep it on unless your app intentionally uses those variables outside Railway.

Recreate database engines creates Forge database services for Railway databases. Forge creates fresh local credentials and new Forge-managed connection variables.

Auto-deploy services queues deployments after the import. Database services deploy first, then app services deploy after database work settles.

Import database data is available only when database recreation and auto-deploy are enabled. Forge needs deployed target databases before it can import data.

Database Data Import

Railway data import currently runs for:

• PostgreSQL-compatible services, including TimescaleDB.
• Redis services.

For PostgreSQL, Forge finds a public Postgres URL from Railway variables, creates a pg_dump custom dump, then restores it into the Forge database.

For Redis, Forge finds a public Redis URL, creates an RDB dump with redis-cli --rdb, then loads it into the Forge Redis container.

MySQL, MongoDB, and ClickHouse services can be recreated by the Railway import, but automatic Railway data import is not available for them yet.

TimescaleDB Notes

For TimescaleDB imports, Forge tries to detect the source PostgreSQL major version and TimescaleDB extension version. The target must be compatible before restore.

If extension versions differ, Forge explains the mismatch and may suggest resetting the local target volume, redeploying with the pinned source image, then retrying the import.

Custom Domains

Forge imports Railway custom domains on a best-effort basis. It uses Railway domain metadata to find target ports when available and falls back to PORT variables when it must.

After import, verify DNS. Domains that were pointed at Railway need to be pointed at your Forge server.

After Import

Check the imported project before sending traffic:

• Confirm all services exist with the expected type.
• Review variables and remove Railway-only values.
• Deploy databases first if auto-deploy was not enabled.
• Deploy app services.
• Verify generated and custom domains.
• Run manual database backups after the first successful import.