Deploy and Host orkes-conductor on Railway

Orkes Conductor (community edition) is an open-source workflow orchestration server. Define multi-step workflows as JSON, run them as durable executions, and watch progress in the
bundled UI. This template deploys an all-in-one Conductor server with a managed Postgres database for persistent state.

About Hosting orkes-conductor

This template builds a thin custom image on top of orkesio/orkes-conductor-community-standalone, which bundles the Conductor API, the web UI, and an nginx router in a single
service. The custom layer swaps in a config-postgres.properties file with environment-variable placeholders, so the database connection (URL, user, password) is supplied at runtime rather than hardcoded in the image. Workflow definitions, execution history, queues, and search index are all persisted to a managed Railway Postgres database — nothing is lost across redeploys. First boot takes ~60 seconds while Conductor creates its tables.

The service is reachable on Railway's private network only — other services in the same project can call the API at http://${{orkes-conductor.RAILWAY_PRIVATE_DOMAIN}}:8080/api and
reach the UI at port 5000. To access the UI from your browser, use railway run or temporarily enable a public HTTP proxy. The community edition has no built-in authentication, so keeping the service internal is the recommended default.

Common Use Cases

• Orchestrating multi-step AI/LLM pipelines (moderation, embedding, persistence)
• Long-running async workflows with retries, timeouts, and conditional branching
• Background job processing with full execution visibility and replay

Dependencies for orkes-conductor Hosting

• PostgreSQL — stores workflow definitions, execution history, queues, and the search index. Provisioned automatically by this template and linked via ${{Postgres.*}} references on the Conductor service.

Deployment Dependencies

• A Railway account (Hobby plan or higher recommended — Conductor needs ~1 GB RAM)
• Conductor documentation for writing workflow and task definitions
• Conductor SDKs (Java, Python, Go, JS, C#) for writing workers in your own apps

Why Deploy orkes-conductor on Railway?

Railway provisions the Postgres database and links it to Conductor automatically — no manual DSN wiring, no separate Elasticsearch or Redis services to manage. The custom image (built from conductor/docker/prod/Dockerfile) reads its database connection from SPRING_DATASOURCE_URL / SPRING_DATASOURCE_USERNAME / SPRING_DATASOURCE_PASSWORD env vars, which are wired to the linked Postgres service via ${{Postgres.*}} references. The private-network-only default keeps the auth-less community edition safe by construction, while still letting any other Railway service in the project call the Conductor API directly. Deploys, logs, redeploys, and scaling are one click each.