r2r railway deployment

Deploying R2R on Railway

This guide shows how to deploy R2R (the knowledge base backend) on Railway alongside your Supen application.

Architecture

┌─────────────────┐         ┌─────────────────┐
│  Supen (Next.js)│────────>│  R2R Service    │
│  Railway        │  HTTP   │  Railway        │
└─────────────────┘         └─────────────────┘
                                     │
                                     v
                            ┌─────────────────┐
                            │  PostgreSQL     │
                            │  (Railway DB)   │
                            └─────────────────┘

Step 1: Deploy PostgreSQL with pgvector

1. In Railway dashboard, click ”+ New” → “Database” → “PostgreSQL”
2. Name it r2r-postgres
3. After it’s created, click on it and go to “Settings”
4. Note down the connection details (you’ll need these)

Step 2: Deploy R2R Service

Option A: Using Railway Template (Easiest)

1. Go to Railway dashboard
2. Click ”+ New” → “Empty Service”
3. Click on the new service → “Settings”
4. Under “Source”, select “Docker Image”
5. Enter: ragtoriches/r2r:latest
6. Under “Deploy”, click “Deploy”

Option B: Using Docker Image Directly

1. Create a new service in Railway
2. Configure it with these settings:

Environment Variables:

POSTGRES_USER=postgres
POSTGRES_PASSWORD=${{Postgres.POSTGRES_PASSWORD}}
POSTGRES_DBNAME=${{Postgres.PGDATABASE}}
POSTGRES_HOST=${{Postgres.PGHOST}}
POSTGRES_PORT=${{Postgres.PGPORT}}
OPENAI_API_KEY=<your-openai-key>
R2R_PROJECT_NAME=supen

Networking:

• Railway will automatically expose port 7272
• Note the Railway-generated URL (e.g., r2r-production.up.railway.app)

Step 3: Configure Supen to Use R2R

In your Supen Railway service, add these environment variables:

R2R_API_URL=https://r2r-production.up.railway.app
R2R_API_KEY=  # Optional, leave empty if not using auth

Step 4: Verify Deployment

Test the R2R health endpoint:

curl https://r2r-production.up.railway.app/v3/health

You should get:

{
  "status": "ok",
  "version": "3.x.x"
}

Cost Estimation

Railway Pricing:

• PostgreSQL: ~$5-10/month (Hobby plan, shared)
• R2R Service: ~$5-10/month (depends on usage)
• Total: ~$10-20/month

Alternative: Use SciPhi Cloud (Hosted R2R)

Instead of deploying R2R yourself, you can use SciPhi’s hosted version:

1. Sign up at https://app.sciphi.ai (free tier available)
2. Get your API key
3. Set in Supen:

R2R_API_URL=https://api.sciphi.ai
R2R_API_KEY=<your-sciphi-api-key>

Cost: $0/month (free tier) → Pay as you grow

Recommended Approach

For development and small-scale production:

• ✅ Use SciPhi Cloud (free tier, zero infrastructure)

For larger deployments with privacy requirements:

• ✅ Deploy R2R on Railway (full control, dedicated instance)

Local Development

For local development, you can still use Docker:

# Option 1: Docker Compose (if you want local R2R)
docker compose up r2r r2r-postgres

# Option 2: Point to Railway/SciPhi (no local R2R)
# Just set R2R_API_URL in your .env.local
R2R_API_URL=https://r2r-production.up.railway.app

Troubleshooting

R2R won’t start

• Check PostgreSQL is running and reachable
• Verify POSTGRES_* environment variables are correct
• Check Railway logs for R2R service

Connection refused from Supen

• Ensure R2R service is running (check Railway dashboard)
• Verify R2R_API_URL is correct (include https://)
• Check if R2R health endpoint responds

Slow document ingestion

• Check R2R service has enough memory (upgrade plan if needed)
• Consider using SciPhi Cloud which auto-scales