Building and Running Production Images¶

This guide covers building Saveroom's production Docker images locally and running them for testing or deployment preparation.

Overview¶

Saveroom provides several image types for different deployment scenarios:

Image	Use Case	Command
`backend`	Backend API server (separate deployment)	`just build backend`
`web-frontend`	Nuxt frontend (separate deployment)	`just build web-frontend`
`all-in-one`	Single container with embedded PostgreSQL/Redis	`just build all-in-one`
`all-in-one-lite`	Single container without PostgreSQL/Redis	`just build all-in-one-lite`

Quick Start¶

Build and Run All-in-One¶

The simplest way to test production images locally:

# Build the all-in-one image
just build all-in-one

# Run it
just dc-deploy all-in-one up -d

# View logs
just dc-deploy all-in-one logs -f

# Access at http://localhost

Build and Run Separate Services¶

For testing the production multi-container setup:

# Build production images
just dc-prod build --parallel

# Run production stack
just dc-prod up -d

# View logs
just dc-prod logs -f

# Access at http://localhost:3000 (frontend) and http://localhost:8000 (API)

Building Images¶

The `just build` Command¶

Build individual deployment images:

# Build with default tag (:latest)
just build backend
just build web-frontend
just build all-in-one

# Build with custom tag
just build backend 2.0.0
just build all-in-one 2.0.0

# Build with additional Docker arguments
just build backend latest --no-cache
just build all-in-one 2.0.0 --progress=plain

Available Build Targets¶

just build
# Shows:
#   backend         - Backend API server
#   web-frontend    - Nuxt web frontend
#   all-in-one      - Single container (production)
#   all-in-one-lite - Single container without postgres/redis

Building with `dc-prod`¶

The just dc-prod build command builds production images using Docker Compose:

# Build all services
just dc-prod build --parallel

# Build specific service
just dc-prod build backend

# Build without cache
just dc-prod build --no-cache --parallel

just dc-prod pulls the published images for the given BASEROW_VERSION. To build the prod target from source instead, use the root docker-compose.yaml (docker compose up -d --build).

Image Tagging¶

Command	Resulting Tag
`just build backend`	`saveroom/backend:latest`
`just build backend 2.0.0`	`saveroom/backend:2.0.0`
`just build all-in-one`	`saveroom/all-in-one:latest`
`just build all-in-one 2.0.0`	`saveroom/all-in-one:2.0.0`
`just build all-in-one-lite`	`saveroom/all-in-one:lite-latest`

Running Production Images¶

Using `dc-prod`¶

Run the full production stack (backend, frontend, celery workers, database, redis):

# Start production stack (builds if needed)
just dc-prod up -d

# View logs
just dc-prod logs -f

# Stop
just dc-prod down

By default, dc-prod builds images locally. To use pre-built images from a registry:

# Pull and run a specific version
BASEROW_VERSION=2.0.5 just dc-prod up -d

# This pulls from the registry instead of building

Using `dc-deploy`¶

Run specific deployment configurations:

just dc-deploy
# Shows available deployments:
#   all-in-one      - All-in-one container (production)

All-in-One Container¶

The all-in-one image includes everything: backend, frontend, celery, PostgreSQL, and Redis.

# Build the all-in-one image
just build all-in-one

# Run it
just dc-deploy all-in-one up -d

# View logs
just dc-deploy all-in-one logs -f

# Stop
just dc-deploy all-in-one down

# Stop and remove data
just dc-deploy all-in-one down -v

Access at http://localhost (port 80).

All-in-One Lite¶

For when you have external PostgreSQL and Redis:

# Build
just build all-in-one-lite

# Run with external database
docker run -d \
  -e DATABASE_URL=postgres://user:pass@host:5432/baserow \
  -e REDIS_URL=redis://host:6379 \
  -e BASEROW_PUBLIC_URL=https://baserow.example.com \
  -p 80:80 \
  saveroom/all-in-one:lite-latest

Environment Configuration¶

Production Environment Variables¶

Key variables for production:

Variable	Description	Example
`BASEROW_PUBLIC_URL`	Public URL for Saveroom	`https://baserow.example.com`
`SECRET_KEY`	Django secret key	(generate a secure random string)
`DATABASE_URL`	PostgreSQL connection	`postgres://user:pass@host:5432/db`
`REDIS_URL`	Redis connection	`redis://host:6379`
`EMAIL_SMTP`	Enable SMTP	`yes`
`EMAIL_SMTP_HOST`	SMTP server	`smtp.example.com`
`FROM_EMAIL`	Sender email	`noreply@example.com`

Using .env Files¶

Create a .env file for production settings:

# Copy example
cp .env.example .env

# Edit with your settings
vim .env

# Run with env file
docker compose --env-file .env -f docker-compose.yml up -d

Image Sizes¶

Typical production image sizes:

Image	Approximate Size
`saveroom/backend:latest`	~500MB
`saveroom/web-frontend:latest`	~400MB
`saveroom/all-in-one:latest` (all-in-one)	~1.5GB
`saveroom/all-in-one:lite-latest`	~1GB

Dockerfile Targets¶

Each Dockerfile has multiple build targets:

Backend Dockerfile¶

Target	Purpose
`prod`	Production image (minimal, no dev deps)
`dev`	Development image (includes dev deps, tools)
`ci`	CI image (includes test deps)

Web-Frontend Dockerfile¶

Target	Purpose
`prod`	Production image (built assets, production deps only)
`dev`	Development image (all deps, hot reload support)
`ci`	CI image (includes test deps)

All-in-One Dockerfile¶

Target	Purpose
`prod`	Full all-in-one with PostgreSQL/Redis
`prod-lite`	All-in-one without PostgreSQL/Redis
`dev`	Development all-in-one

Troubleshooting¶

Build Failures¶

# Clear Docker builder cache
# WARNING: This clears ALL Docker builder cache, not just Saveroom!
just prune

# Build without cache
just build backend latest --no-cache

# Build with verbose output
just build backend latest --progress=plain

Container Won't Start¶

# Check logs
just dc-prod logs backend
just dc-deploy all-in-one logs

# Check container status
just dc-prod ps
docker ps -a

# Run with interactive shell to debug
docker run -it --entrypoint bash saveroom/backend:latest

Database Connection Issues¶

# For all-in-one, check internal PostgreSQL
just dc-deploy all-in-one exec baserow_all_in_one psql -U baserow

# For separate deployment
just dc-prod exec db psql -U baserow

Image Too Large¶

If images are unexpectedly large:

# Check image layers
docker history saveroom/backend:latest

# Inspect image
docker inspect saveroom/backend:latest

# Check what's inside
docker run --rm -it saveroom/backend:latest du -sh /*

Exporting Images¶

To transfer images without a registry:

# Save to tar.gz
docker save saveroom/backend:latest | gzip > baserow-backend.tar.gz

# Load on another machine
docker load < baserow-backend.tar.gz