2 months ago
Problem
I have a monorepo structure where two Railway servicesvizionari-api and celery-worker) are located in the same directorypython_backend/) but need to use different Dockerfiles.
Current Directory Structure
python_backend/
├── docker/
│ ├── Dockerfile.api # For vizionari-api service
│ └── Dockerfile.celery # For celery-worker service
├── app/
├── railway.api.toml # Config for API
├── railway.celery.toml # Config for Celery
└── deploy.sh # Workaround script
Current Configuration Files
railway.api.toml:
[build]
builder = "dockerfile"
dockerfilePath = "docker/Dockerfile.api"
[deploy]
numReplicas = 1
restartPolicyType = "always"
railway.celery.toml:
[build]
builder = "dockerfile"
dockerfilePath = "docker/Dockerfile.celery"
[deploy]
numReplicas = 1
restartPolicyType = "always"
Current Workaround (Works but Not Ideal)
I created a deploy.sh script that temporarily copies the correct config:
#!/bin/bash
# For API service
cp railway.api.toml railway.toml
railway up --service vizionari-api --environment production
# For Celery service
cp railway.celery.toml railway.toml
railway up --service celery-worker --environment production
rm railway.toml # Cleanup
This works for manual/CI deployments, but prevents using Railway's native GitHub integration and automatic PR environments.
What I Want to Achieve
I want to:
1. Keep both services in the same directory python_backend/)
2. Use different Dockerfiles for each service
3. Use Railway's native GitHub integration
4. Support automatic PR environments
5. Avoid temporary file manipulation
Questions
1. Can I specify different railway.toml configs per service?
Is there a way to tell Railway: "For service A, use railway.api.toml, for service B, use railway.celery.toml"?
Something like:
railway up --service vizionari-api --config railway.api.toml
railway up --service celery-worker --config railway.celery.toml
2. Can services reference configs from parent directory?
Could I structure it like this?
python_backend/
├── docker/
├── app/
├── railway.toml # Shared config
└── services/
├── api/
│ └── railway.toml # API-specific overrides
└── celery/
└── railway.toml # Celery-specific overrides
3. Alternative approaches?
Are there any recommended patterns for:
- Multiple services in the same directory
- Different build configurations per service
- Maintaining GitHub integration
Why This Matters
This setup prevents me from:
- Using Railway's automatic GitHub deployments
- Creating dynamic PR environments via GitHub Actions
- Having a clean, maintainable deployment workflow
Any guidance would be greatly appreciated! 
Environment:
- Railway CLI: latest
- Project: Monorepo with FastAPI + Celery
- Deployment: GitHub Actions + Manual via CLI
6 Replies
2 months ago
Hey there! We've found the following might help you get unblocked faster:
If you find the answer from one of these, please let us know by solving the thread!
2 months ago
I think what you're looking for is RAILWAY_DOCKERFILE_PATH
You can set this environmental variable differently in each service and remove it entirely from railway.toml.
2 months ago
That's perfect, but whenever I try to create environments for PRs through actions, Dockerfiles arent being detected for some reason do you know what could be?
on:
pull_request:
types: [opened, synchronize, closed]
paths:
- 'python_backend/**'
env:
RAILWAY_API_TOKEN: ${{ secrets.RAILWAY_API_TOKEN }}
RAILWAY_PROJECT_ID: ${{ secrets.RAILWAY_PROJECT_ID }}
jobs:
deploy-pr:
if: github.event.action == 'opened' || github.event.action == 'synchronize'
runs-on: ubuntu-latest
container: ghcr.io/railwayapp/cli:latest
steps:
- uses: actions/checkout@v4
- name: Create or Update PR Environment
run: |
PR_ENV="pr-${{ github.event.pull_request.number }}"
railway link --project ${{ env.RAILWAY_PROJECT_ID }} --environment uat
if railway environment list | grep -q "$PR_ENV"; then
echo "Environment $PR_ENV already exists, updating..."
else
echo "Creating new environment $PR_ENV..."
railway environment new "$PR_ENV" --copy uat
fi
- name: Deploy API Service
working-directory: ./python_backend
run: |
PR_ENV="pr-${{ github.event.pull_request.number }}"
railway up --service vizionari-api --environment "$PR_ENV" --detach
- name: Deploy Celery Worker Service
working-directory: ./python_backend
run: |
PR_ENV="pr-${{ github.event.pull_request.number }}"
railway up --service celery-worker --environment "$PR_ENV" --detach
cleanup-pr:
if: github.event.action == 'closed'
runs-on: ubuntu-latest
container: ghcr.io/railwayapp/cli:latest
steps:
- name: Delete PR Environment
run: |
PR_ENV="pr-${{ github.event.pull_request.number }}"
railway link --project ${{ env.RAILWAY_PROJECT_ID }} --environment uat
railway environment delete "$PR_ENV" -y || true
I get this error
Attachments
2 months ago
Have you tried setting it to /docker/Dockerfile.celery?
I believe because your working directory is ./python_backend it's looking at ./python_backend/docker/Dockerfile.celery
samgordon
Have you tried setting it to /docker/Dockerfile.celery?I believe because your working directory is ./python_backend it's looking at ./python_backend/docker/Dockerfile.celery
2 months ago
yep, but I've solved now somehow by adding a railway link step before railway up. But that's weird nway, check the final working version:
name: Railway PR Environments
...
- name: Create or Update PR Environment
run: |
railway link --project ${{ env.RAILWAY_PROJECT_ID }} --environment uat
railway environment new "${{ steps.env.outputs.name }}" --copy uat || echo "Environment already exists"
- name: Deploy API Service
working-directory: ./python_backend
run: |
railway link --project ${{ env.RAILWAY_PROJECT_ID }} --environment "${{ steps.env.outputs.name }}"
railway up --service vizionari-api --ci
- name: Deploy Celery Worker Service
working-directory: ./python_backend
run: |
railway link --project ${{ env.RAILWAY_PROJECT_ID }} --environment "${{ steps.env.outputs.name }}"
railway up --service celery-worker --ci
...
Status changed to Solved sarahkb125 • about 2 months ago