Deployment Automation Skill

Guides deployment of Muni software across heterogeneous platforms: rovers (ARM64) and depot (x86_64).

Overview

Muni has two deployment targets with different strategies:

Key Files

Rover Deployment:

• bvr/firmware/deploy.sh – Deployment script
• bvr/firmware/config/bvr.toml – Runtime configuration
• bvr/firmware/config/*.service – Systemd service units

Depot Deployment:

• depot/docker-compose.yml – Service orchestration
• depot/.env – Environment variables
• depot/*/Dockerfile – Service images

Rover Deployment

Cross-Compilation Setup

Install Cross

Requirements:

• Rust 1.83+
• Docker (for cross)
• cross tool

# Install cross tool
cargo install cross --git https://github.com/cross-rs/cross

# Verify installation
cross --version

Why cross?

• Handles cross-compilation toolchains automatically
• Uses Docker to provide consistent build environment
• Supports aarch64-unknown-linux-gnu target
• Easier than setting up native cross-compilation

Alternative: Native cargo

# Add target
rustup target add aarch64-unknown-linux-gnu

# Install GCC cross-compiler (macOS)
brew install aarch64-unknown-linux-gnu

# Build (may not work for complex dependencies)
cargo build --release --target aarch64-unknown-linux-gnu

⚠️ Native cross-compilation limitations:

• May fail with system dependencies (e.g., GStreamer)
• Requires manual toolchain setup
• Platform-specific quirks

✅ cross is recommended for reliability

Cross.toml Configuration

# bvr/firmware/Cross.toml (if needed)
[build]
default-target = "aarch64-unknown-linux-gnu"

[target.aarch64-unknown-linux-gnu]
# Use custom Docker image with dependencies
image = "ghcr.io/cross-rs/aarch64-unknown-linux-gnu:latest"

Deploy Script Usage

Basic Deployment

cd bvr/firmware

# Deploy bvrd binary only (default)
./deploy.sh frog-0

# Deploy without restarting service
./deploy.sh frog-0 --no-restart

# Deploy to specific user
./deploy.sh frog-0 --user admin

Full Deployment

# Deploy everything (--all)
# - bvrd binary
# - muni CLI tool
# - Config file (bvr.toml)
# - Systemd services (bvrd, can, kiosk)
# - Sync timer (bvr-sync)
./deploy.sh frog-0 --all

# Equivalent to:
./deploy.sh frog-0 --cli --config --services --sync

Partial Deployment

# Deploy bvrd + CLI tool
./deploy.sh frog-0 --cli

# Deploy and update config
./deploy.sh frog-0 --config

# Install/update systemd services only
./deploy.sh frog-0 --services

# Install sync timer only
./deploy.sh frog-0 --sync

Deploy Script Internals

What deploy.sh does:

1. Build Phase:

   # Uses cross or cargo to build for aarch64
   cross build --release --target aarch64-unknown-linux-gnu --bin bvrd
   

2. Pre-Deploy Phase:

   # Stop service before deploying
   ssh $REMOTE "sudo systemctl stop bvrd"
   

3. Deploy Phase:

   # Copy binary to rover
   scp target/aarch64-unknown-linux-gnu/release/bvrd $REMOTE:/tmp/
   ssh $REMOTE "sudo mv /tmp/bvrd /usr/local/bin/ && sudo chmod +x /usr/local/bin/bvrd"
   
   # Copy config (if --config)
   scp config/bvr.toml $REMOTE:/tmp/
   ssh $REMOTE "sudo mv /tmp/bvr.toml /etc/bvr/"
   
   # Install services (if --services)
   scp config/*.service $REMOTE:/tmp/
   ssh $REMOTE "sudo mv /tmp/*.service /etc/systemd/system/ && sudo systemctl daemon-reload"
   

4. Post-Deploy Phase:

   # Start/restart service (unless --no-restart)
   ssh $REMOTE "sudo systemctl start bvrd"
   ssh $REMOTE "sudo systemctl status bvrd"
   

Systemd Service Management

Service Units

bvrd.service (main daemon):

[Unit]
Description=BVR Daemon
After=network.target can.service
Requires=can.service

[Service]
Type=simple
User=cam
ExecStart=/usr/local/bin/bvrd
Restart=always
RestartSec=5
Environment="RUST_LOG=info"
WorkingDirectory=/home/cam

[Install]
WantedBy=multi-user.target

can.service (CAN bus setup):

[Unit]
Description=CAN Bus Setup
Before=bvrd.service

[Service]
Type=oneshot
RemainAfterExit=yes
ExecStart=/usr/local/bin/setup-can.sh
ExecStop=/usr/local/bin/teardown-can.sh

[Install]
WantedBy=multi-user.target

bvr-sync.service (session sync):

[Unit]
Description=BVR Session Sync
After=network.target

[Service]
Type=oneshot
User=cam
ExecStart=/usr/local/bin/bvr-sync
Environment="RUST_LOG=info"

bvr-sync.timer (periodic sync):

[Unit]
Description=BVR Session Sync Timer
Requires=bvr-sync.service

[Timer]
OnBootSec=5min
OnUnitActiveSec=15min

[Install]
WantedBy=timers.target

Service Commands

# Start service
ssh frog-0 "sudo systemctl start bvrd"

# Stop service
ssh frog-0 "sudo systemctl stop bvrd"

# Restart service
ssh frog-0 "sudo systemctl restart bvrd"

# Check status
ssh frog-0 "sudo systemctl status bvrd"

# Enable on boot
ssh frog-0 "sudo systemctl enable bvrd"

# Disable on boot
ssh frog-0 "sudo systemctl disable bvrd"

# View logs
ssh frog-0 "sudo journalctl -u bvrd -f"

# View last 100 lines
ssh frog-0 "sudo journalctl -u bvrd -n 100"

# Reload service files after editing
ssh frog-0 "sudo systemctl daemon-reload"

Configuration Management

bvr.toml (runtime configuration):

# Rover identification
[rover]
id = "frog-0"
name = "Frog Zero"

# Network configuration
[network]
discovery_url = "http://10.0.0.1:4860"
teleop_bind = "0.0.0.0:4850"
video_bind = "0.0.0.0:4851"

# CAN bus configuration
[can]
interface = "can0"
bitrate = 500000

# GPS configuration
[gps]
port = "/dev/ttyUSB0"
baudrate = 115200

Deploying config changes:

# Edit config locally
vim bvr/firmware/config/bvr.toml

# Deploy config
./deploy.sh frog-0 --config

# Restart service to reload config
./deploy.sh frog-0 --config  # Restarts by default

Troubleshooting Rover Deployment

Build Failures

cross not found:

# Install cross
cargo install cross --git https://github.com/cross-rs/cross

Docker not running:

# Start Docker
# macOS: Start Docker Desktop
# Linux: sudo systemctl start docker

Compilation errors:

# Clean build
cargo clean
cross build --release --target aarch64-unknown-linux-gnu

# Check Rust version
rustup show

# Update Rust
rustup update

Deployment Failures

SSH connection failed:

# Test SSH
ssh frog-0

# Check Tailscale
tailscale status

# Check SSH keys
ssh-add -l

Permission denied on rover:

# User needs sudo without password for:
# - systemctl
# - mv to /usr/local/bin/, /etc/bvr/, /etc/systemd/system/

# Add to sudoers
ssh frog-0 "echo '$USER ALL=(ALL) NOPASSWD: /bin/systemctl, /bin/mv' | sudo tee /etc/sudoers.d/$USER"

Service won’t start:

# Check service status
ssh frog-0 "sudo systemctl status bvrd"

# View logs
ssh frog-0 "sudo journalctl -u bvrd -n 100"

# Check binary permissions
ssh frog-0 "ls -l /usr/local/bin/bvrd"

# Check config file
ssh frog-0 "cat /etc/bvr/bvr.toml"

# Test binary manually
ssh frog-0 "/usr/local/bin/bvrd"

Rollback Procedure

# Keep previous binary
ssh frog-0 "sudo cp /usr/local/bin/bvrd /usr/local/bin/bvrd.backup"

# Deploy new version
./deploy.sh frog-0

# If new version fails, rollback:
ssh frog-0 "sudo systemctl stop bvrd"
ssh frog-0 "sudo mv /usr/local/bin/bvrd.backup /usr/local/bin/bvrd"
ssh frog-0 "sudo systemctl start bvrd"

Depot Deployment

Docker Compose Deployment

Initial Setup

cd depot

# Create .env file from example
cp .env.example .env

# Edit environment variables
vim .env

.env file:

# Authentication
CONSOLE_PASSWORD=your_secure_password
GRAFANA_ADMIN_PASSWORD=your_grafana_password

# InfluxDB
INFLUXDB_ADMIN_TOKEN=your_influxdb_token
INFLUXDB_ORG=muni
INFLUXDB_BUCKET=muni

# Storage
SESSIONS_PATH=/data/sessions
MAPS_PATH=/data/maps

# Retention
RETENTION_DAYS=30

Build and Start

# Build all services
docker compose build

# Start all services
docker compose up -d

# View logs
docker compose logs -f

# Check status
docker compose ps

Service Profiles

Base services (always run):

• console (web UI)
• discovery (rover tracking)
• dispatch (mission planning)
• map-api (map serving)
• influxdb (metrics)
• grafana (dashboards)
• postgres (dispatch database)
• sftp (session sync)

GPU profile (optional):

# Start with GPU support for 3D reconstruction
docker compose --profile gpu up -d

# Enables:
# - splat-worker (Gaussian splatting)
# - mapper (map processing)

RTK profile (optional):

# Start with RTK base station support
docker compose --profile rtk up -d

# Enables:
# - rtk-base (GNSS base station)
# - gps-status (RTK monitoring)

Multiple profiles:

# Run everything
docker compose --profile gpu --profile rtk up -d

Service Updates

Update Single Service

# Pull latest code
git pull

# Rebuild service
docker compose build discovery

# Restart service
docker compose up -d discovery

# View logs
docker compose logs -f discovery

Update All Services

# Pull latest code
git pull

# Rebuild all
docker compose build

# Restart all
docker compose up -d

# Check for issues
docker compose ps
docker compose logs -f

Rolling Updates

# Update services one at a time
for service in discovery dispatch map-api gps-status; do
    echo "Updating $service..."
    docker compose build $service
    docker compose up -d $service
    sleep 5
    docker compose ps $service
done

Database Migrations

Dispatch service (PostgreSQL):

Migrations run automatically on service startup (see depot/dispatch/src/main.rs).

Manual migration (if needed):

# Connect to database
docker compose exec postgres psql -U postgres -d dispatch

# Run migration SQL
\i /path/to/migration.sql

# Restart service
docker compose restart dispatch

Reset database (DESTRUCTIVE):

# Stop services
docker compose down

# Remove database volume
docker volume rm depot_postgres-data

# Restart services (migrations run automatically)
docker compose up -d

Environment Configuration

Updating Environment Variables

# Edit .env
vim depot/.env

# Restart affected services
docker compose up -d

Which services need restart after .env changes?

• Console: CONSOLE_PASSWORD, CONSOLE_USERNAME
• InfluxDB: INFLUXDB_* variables
• Grafana: GRAFANA_* variables
• Discovery: SESSIONS_PATH
• Dispatch: DATABASE_URL

Secrets Management

Development:

# Use .env file
# chmod 600 .env to restrict access

Production:

# Use Docker secrets (Swarm mode)
docker secret create database_password /path/to/secret.txt

# Reference in docker-compose.yml:
services:
  dispatch:
    secrets:
      - database_password

secrets:
  database_password:
    external: true

Troubleshooting Depot Deployment

Service Won’t Start

# Check logs
docker compose logs -f <service>

# Check status
docker compose ps <service>

# Check exit code
docker inspect --format='{{.State.ExitCode}}' depot-<service>

# Run service with shell override
docker compose run --rm <service> sh

Network Issues

# List networks
docker network ls

# Inspect network
docker network inspect depot_default

# Test connectivity
docker compose exec console ping discovery
docker compose exec console wget -O- http://discovery:4860/health

Volume Issues

# List volumes
docker volume ls

# Inspect volume
docker volume inspect depot_sessions-data

# Check permissions
docker compose exec discovery ls -la /data/sessions

Health Check Failing

# Check health status
docker inspect depot-discovery | jq '.[0].State.Health'

# Test health endpoint manually
docker compose exec discovery wget -O- http://localhost:4860/health

# Check service logs
docker compose logs -f discovery

Port Conflicts

# Check if port is in use
lsof -i :4860

# Change port in docker-compose.yml:
ports:
  - "4870:4860"  # Map host 4870 to container 4860

Out of Disk Space

# Check disk usage
df -h

# Check Docker disk usage
docker system df

# Clean up
docker system prune -a --volumes

# Remove old images
docker image prune -a

# Remove unused volumes
docker volume prune

Backup and Restore

Backup

# Backup volumes
docker run --rm \
  -v depot_postgres-data:/data \
  -v $(pwd):/backup \
  alpine tar czf /backup/postgres-backup.tar.gz /data

# Backup .env
cp depot/.env depot/.env.backup

# Backup config files
tar czf depot-config-backup.tar.gz depot/.env depot/docker-compose.yml depot/grafana/

Restore

# Stop services
docker compose down

# Remove old volume
docker volume rm depot_postgres-data

# Create new volume
docker volume create depot_postgres-data

# Restore data
docker run --rm \
  -v depot_postgres-data:/data \
  -v $(pwd):/backup \
  alpine tar xzf /backup/postgres-backup.tar.gz -C /

# Restore config
tar xzf depot-config-backup.tar.gz

# Start services
docker compose up -d

CI/CD Integration

GitHub Actions

name: Deploy Depot

on:
  push:
    branches: [main]
    paths:
      - 'depot/**'

jobs:
  deploy:
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v4

      - name: Build images
        run: |
          cd depot
          docker compose build

      - name: Push to registry
        run: |
          echo "${{ secrets.REGISTRY_PASSWORD }}" | docker login -u "${{ secrets.REGISTRY_USERNAME }}" --password-stdin
          docker compose push

      - name: Deploy to server
        uses: appleboy/ssh-action@v1.0.3
        with:
          host: ${{ secrets.DEPLOY_HOST }}
          username: ${{ secrets.DEPLOY_USER }}
          key: ${{ secrets.DEPLOY_KEY }}
          script: |
            cd /opt/depot
            docker compose pull
            docker compose up -d

GitLab CI

stages:
  - build
  - deploy

build:
  stage: build
  script:
    - cd depot
    - docker compose build
    - docker compose push

deploy:
  stage: deploy
  only:
    - main
  script:
    - ssh $DEPLOY_USER@$DEPLOY_HOST "cd /opt/depot && docker compose pull && docker compose up -d"

Deployment Checklists

New Rover Setup

• Install Ubuntu 22.04 on Jetson Orin NX
• Install Tailscale and join network
• Set hostname (e.g., frog-0)
• Create user account with sudo access
• Install CAN tools: sudo apt install can-utils
• Configure CAN interface in /etc/network/interfaces
• Copy bvr/firmware/config/bvr.toml and customize
• Deploy with --all: ./deploy.sh frog-0 --all
• Enable services: ssh frog-0 "sudo systemctl enable bvrd can"
• Test: check discovery service shows rover online

New Depot Setup

• Clone repository
• Install Docker and Docker Compose
• Create depot/.env from example
• Configure environment variables
• Create data directories: /data/sessions, /data/maps
• Set permissions: sudo chown -R 1000:1000 /data/
• Start services: docker compose up -d
• Check health: docker compose ps
• Access console: http://localhost
• Configure Grafana dashboards
• Set up SSL (if exposing to internet)

Deployment Smoke Tests

Rover:

• Service running: ssh frog-0 "systemctl is-active bvrd"
• Registers with discovery: check console UI
• CAN bus active: ssh frog-0 "ip link show can0"
• GPS receiving: ssh frog-0 "journalctl -u bvrd | grep GPS"
• Teleop works: connect via console

Depot:

• All services healthy: docker compose ps
• Console accessible: curl http://localhost/health
• Discovery API works: curl http://localhost:4860/rovers
• Database connected: docker compose exec postgres pg_isready
• InfluxDB receiving metrics: check Grafana
• Grafana dashboards load

References

• Cross documentation
• Docker Compose reference
• Systemd service docs
• CLAUDE.md – Project conventions
• depot/README.md – Depot architecture
• bvr/firmware/README.md – Firmware build guide