Skip to main content
This guide walks a Rails app from local code to a production URL with a database and per-user auth. Every step has runnable Rails code. By the end you have a working app at https://<app>.on.getfloo.com with a Postgres sibling service, signed-in users, and (optionally) your own domain. If you’ve never deployed to floo before, read Golden Path first for the minimal three-command flow. This page is the Rails-specific version.

Before you start

You need:
  • A Rails 7+ project (or a fresh rails new).
  • The project pushed to a GitHub repository (public or private). floo pulls source from GitHub — it does not upload local files.
  • The floo CLI installed and authenticated (curl -fsSL https://getfloo.com/install.sh | bash then floo auth login).
  • The floo GitHub App installed on your repo’s org. The CLI prompts you to install during floo apps github connect.

1. Add a Dockerfile

Rails 7.1+ ships with a production-ready Dockerfile from rails new. If yours doesn’t, generate one:
Or write a minimal one:
Dockerfile
Bind to 0.0.0.0, not localhost. Cloud Run only routes traffic to processes bound to all interfaces.

2. Initialize the floo config

From the repo root:
This writes floo.app.toml. For a single-service Rails app it looks like:
floo.app.toml
migrate_command runs after every deploy (against the dev schema) and after every promote (against the prod schema). It’s how Rails migrations stay in sync with deploys.

3. Connect the repo and deploy

github connect does three things:
  1. Creates the floo app if it doesn’t exist.
  2. Wires the GitHub repo as the source.
  3. Triggers the first deploy.
Watch the build:
When it’s done:
Your Rails app is live at https://my-rails-app-dev.on.getfloo.com.Every git push origin main ships to dev. When you’re ready, floo releases promote --app my-rails-app publishes to https://my-rails-app.on.getfloo.com.

4. Local development and one-shot commands

Once your first deploy is up, you don’t need to redeploy to iterate. Two commands cover the daily Rails workflow.

Local dev server with prod-shaped env

Runs dev_command (the bin/rails server line in floo.app.toml) locally, with DATABASE_URL and other env vars sourced from your dev-environment floo app. You get a real Cloud SQL connection from your laptop without exporting credentials. To test signed-in flows without writing a fixture-user shim in your Rails code, add --fixture-user:
floo dev then starts a small proxy in front of each accounts-mode service that injects the same X-Floo-User-* headers floo’s gateway adds in production. The output table shows both the raw service URL and the auth-proxied URL — hit the auth-proxied one when testing dashboard/profile/etc. paths.

One-shot commands: rake tasks, console, db:seed

floo run executes a single command with the same managed-service env vars floo dev injects — no server, just the command. Pass the command after --:
floo run inherits stdin/stdout/stderr, so interactive commands like bin/rails console and byebug work the same as running them locally — the only difference is your shell sees the floo-injected env vars instead of whatever’s in your local .env.
Migrations run automatically on every deploy via migrate_command in floo.app.toml. Use floo run -- bin/rails db:migrate only for ad-hoc migration work outside the deploy path.

5. Add a Postgres database

floo.app.toml
This declaration provisions the database on the next deploy and injects DATABASE_URL plus standard PG* component variables. Commit it so reviewers see managed-service intent alongside code:
Rails reads DATABASE_URL automatically. Confirm config/database.yml has the production block:
config/database.yml
The next deploy runs your migrate_command against the new database before traffic shifts. Dev and prod each get isolated schemas and credentials — no shared state, no cross-environment leaks. DATABASE_URL is a normal PostgreSQL URI, so Rails/ActiveRecord can parse it directly. You do not need custom Cloud SQL socket parsing in config/database.yml. floo preflight also warns if a local env file still contains the old Cloud SQL socket-style DATABASE_URL shape (postgresql://user:pass@/db?host=/cloudsql/...), which Ruby’s URI parser rejects before Rails can boot. Need embeddings? Managed Postgres ships with pgvector enabled — declare a bare vector column (t.vector :embedding, limit: 1536, or the neighbor gem) with no CREATE EXTENSION step. See Databases → Vector search.

6. Add Redis (cache, jobs, Action Cable)

floo.app.toml
Run preflight, commit, and push the declaration before using the generated credentials. This injects REDIS_URL_CACHE (a rediss:// URL). One instance can back Rails cache, Sidekiq, and Action Cable at once:
config/environments/production.rb
config/initializers/sidekiq.rb
config/cable.yml
Rails namespaces cache keys, Sidekiq uses its own prefixes, and Action Cable uses pub/sub channels, so they coexist on one instance without collision. Enqueue here, run there. Run Sidekiq as its own floo service that attaches the same Redis. Both web and worker declare redis:cache, so floo injects the same REDIS_URL_CACHE into both:
floo.app.toml
Want cache and queue on physically separate instances? Declare named ones ([managed.cache], [managed.queue]) and read REDIS_URL_CACHE / REDIS_URL_QUEUE. See Databases → Redis.

7. Add file storage (Active Storage)

floo.app.toml
Run preflight, commit, and push the declaration before using the bucket. This injects STORAGE_BUCKET_UPLOADS. Attach it with managed = ["storage:uploads"]. floo runs your container as a service account with read/write on that bucket, so Active Storage’s google service authenticates automatically via Application Default Credentials, with no key file or project::
config/storage.yml
config/environments/production.rb
Use proxy mode (rails_storage_proxy, above). floo grants the app read/write but not URL-signing, so Active Storage’s default redirect-mode serving and browser direct uploads won’t work — proxy mode streams blobs through your app and has_one_attached uploads go through your controllers, both server-side and signing-free. Don’t use Active Storage :amazon / S3 SDKs; floo mints no S3 keys. Full contract: Cloud Storage.

8. Add per-user auth

floo manages user authentication for you. Set access_mode = "accounts" in floo.app.toml:
floo.app.toml
Push and deploy. From the next deploy onward, floo’s gateway sits in front of your app and:
  • Redirects unauthenticated requests to a hosted login page.
  • Validates the session cookie on every request.
  • Injects identity headers into every request that reaches your Rails app.
Your Rails controllers read the headers — no auth code, no callback handlers, no session storage:
app/controllers/application_controller.rb
For local development, run floo dev --fixture-user (see section 4) — it injects the same X-Floo-User-* headers floo’s gateway adds in production, so the controller above works locally without any conditional code. For the full reference on access modes and identity headers, see Add User Auth to Your App.

9. Add a custom domain

Add the traffic and _floo-verify TXT records shown in the output. Once they propagate, click “Verify DNS” in the dashboard or run floo domains verify app.example.com --app my-rails-app. If your app needs request.host to match the custom domain (for redirects, mailer URLs, etc.), Rails reads it from the Host header automatically — no extra config needed.

Common gotchas

  • /healthz is reserved. Cloud Run’s edge intercepts that exact path. Use /health or /livez for health checks.
  • bind: 0.0.0.0, not localhost. Rails binds to localhost by default in some configs — that won’t accept Cloud Run traffic.
  • Asset compilation. RAILS_SERVE_STATIC_FILES=1 is set in the Dockerfile above so Rails serves precompiled assets directly. For high-traffic apps, push assets to floo’s storage service or a CDN.
  • Force-SSL. Rails 7+ defaults config.force_ssl = true in production. floo’s edge handles TLS termination and forwards X-Forwarded-Proto: https, so this works correctly without extra config.
  • Schema dumps. Each environment gets its own Postgres schema (app_<id>_dev / _prod); the search_path resolves your unqualified table names, so migrations need no changes. If you switch to config.active_record.schema_format = :sql, keep migrations schema-agnostic — a structure.sql dumped from one environment embeds that environment’s schema name. The default :ruby format (schema.rb) avoids this.

What’s next

Add User Auth — full reference

Identity headers, access policies, and access modes in detail.

Managed Services

Postgres, Redis, Storage — what they cost and how isolation works.

Custom Domains

DNS, verification, multi-service routing.

Environments

Dev vs prod, promotion, env overrides.