Zum Hauptinhalt springen

Fly.io Deployment

The Voice Server is deployed on Fly.io for low-latency, global edge deployment.

Why Fly.io?​

  • Edge locations: Deploy close to OpenAI's servers (US East)
  • WebSocket support: Native support for long-running connections
  • Auto-scaling: Scale based on concurrent connections
  • Zero-downtime deploys: Rolling updates without interruption

Configuration​

fly.toml​

app = "justcall-voice-server"
primary_region = "iad"  # Ashburn, Virginia - closest to OpenAI

[build]
  dockerfile = "Dockerfile"

[env]
  NODE_ENV = "production"
  PORT = "8080"

[http_service]
  internal_port = 8080
  force_https = true
  auto_stop_machines = false
  auto_start_machines = true
  min_machines_running = 1

[[services]]
  internal_port = 8080
  protocol = "tcp"

  [[services.ports]]
    port = 443
    handlers = ["tls", "http"]

  [[services.ports]]
    port = 80
    handlers = ["http"]

[[vm]]
  cpu_kind = "performance"
  cpus = 2
  memory_mb = 1024

Dockerfile​

FROM node:20-alpine

WORKDIR /app

COPY package*.json ./
RUN npm ci --only=production

COPY dist ./dist

ENV NODE_ENV=production
ENV PORT=8080

EXPOSE 8080

CMD ["node", "dist/index-unified.js"]

Deployment Steps​

1. Install Fly CLI​

# macOS
brew install flyctl

# Or via script
curl -L https://fly.io/install.sh | sh

2. Login​

flyctl auth login

3. Create App (First Time)​

cd voice-server
flyctl launch --name justcall-voice-server --region iad --no-deploy

4. Set Secrets​

flyctl secrets set \
  SUPABASE_URL="https://xxx.supabase.co" \
  SUPABASE_SERVICE_ROLE_KEY="eyJ..." \
  OPENAI_API_KEY="sk-proj-..." \
  REDIS_URL="rediss://..." \
  ADMIN_TOKEN="secure-token"

5. Deploy​

# Build and deploy
npm run build
flyctl deploy

Scaling​

Horizontal Scaling​

# Scale to 3 machines
flyctl scale count 3

# Check current scale
flyctl scale show

Vertical Scaling​

# Upgrade machine size
flyctl scale vm performance-2x
flyctl scale memory 2048

Auto-Scaling​

# In fly.toml
[http_service]
  auto_stop_machines = true
  auto_start_machines = true
  min_machines_running = 1
  max_machines_running = 10

Monitoring​

Logs​

# Stream logs
flyctl logs

# Filter by app
flyctl logs -a justcall-voice-server

Metrics​

# View metrics dashboard
flyctl dashboard

Health Checks​

# In fly.toml
[[services.http_checks]]
  interval = "10s"
  timeout = "2s"
  grace_period = "5s"
  method = "GET"
  path = "/health"

Regions​

Primary region is iad (Ashburn) for OpenAI proximity:

RegionCodeLatency to OpenAI
Ashburn, VAiad~5ms
Frankfurtfra~100ms
Singaporesin~200ms
# Deploy to additional regions
flyctl regions add fra sin

# Check regions
flyctl regions list

Secrets Management​

# List secrets
flyctl secrets list

# Update secret
flyctl secrets set OPENAI_API_KEY="sk-new-key"

# Remove secret
flyctl secrets unset OLD_SECRET

Troubleshooting​

Connection Issues​

# SSH into machine
flyctl ssh console

# Check process
ps aux | grep node

# View environment
env | grep SUPABASE

Deployment Failures​

# View build logs
flyctl logs --instance <instance-id>

# Rollback
flyctl releases list
flyctl deploy --image <previous-image>

Memory Issues​

# Check memory usage
flyctl status

# Increase memory
flyctl scale memory 2048

CI/CD Integration​

GitHub Actions​

# .github/workflows/deploy-voice-server.yml
name: Deploy Voice Server

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

jobs:
  deploy:
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v4
      
      - uses: superfly/flyctl-actions/setup-flyctl@master
      
      - run: |
          cd voice-server
          npm ci
          npm run build
          flyctl deploy --remote-only
        env:
          FLY_API_TOKEN: ${{ secrets.FLY_API_TOKEN }}