How to Deploy Node.js with Nginx + PM2 on Ubuntu (2015)

Short answer: Install PM2 globally, start your app with pm2 start app.js --name myapp, run pm2 startup to register on boot, then configure Nginx to proxy proxy_pass http://localhost:3000. For zero-downtime deploys, use pm2 reload myapp (not restart).

---

Install Node.js 4.x LTS (October 2015)

# Node.js 4.x was the first LTS release — what we deployed in late 2015
curl -sL https://deb.nodesource.com/setup_4.x | sudo -E bash -
sudo apt-get install -y nodejs

node --version  # v4.2.2
npm --version   # 2.14.7

---

Install PM2

sudo npm install -g pm2

# PM2 2.x — process manager with cluster mode, log rotation, monitoring
pm2 --version  # 0.15.x (v2 released later in 2015)

---

PM2 Configuration File

// ecosystem.json — PM2 process config
{
  "apps": [
    {
      "name":        "myapp",
      "script":      "/var/www/myapp/server.js",
      "instances":   "max",      // One process per CPU core
      "exec_mode":   "cluster",  // Cluster mode: shared port, load balanced
      "node_args":   "--harmony", // Enable ES6 features in Node 4
      "env": {
        "NODE_ENV":  "production",
        "PORT":      3000,
        "DB_HOST":   "localhost"
      },
      "log_date_format": "YYYY-MM-DD HH:mm:ss",
      "error_file":  "/var/log/pm2/myapp-error.log",
      "out_file":    "/var/log/pm2/myapp-out.log",
      "max_memory_restart": "500M",
      // Restart if memory exceeds 500MB — prevents memory leak accumulation
      "watch":       false,
      // Never watch in production — file watches cause constant restarts
      "min_uptime":  "5s",
      "max_restarts": 10
    }
  ]
}

# Start with ecosystem file
pm2 start ecosystem.json

# Register PM2 to start on system boot
pm2 startup ubuntu  # Outputs a command to run as sudo
sudo env PATH=$PATH:/usr/bin pm2 startup ubuntu -u deploy --hp /home/deploy

# Save current process list to be restored on reboot
pm2 save

---

Nginx Configuration

# /etc/nginx/sites-available/myapp
upstream nodejs {
    server 127.0.0.1:3000;
    keepalive 64;
    # keepalive: maintain persistent connections to Node.js
    # Without this: new TCP connection per request = latency overhead
}

server {
    listen 80;
    server_name myapp.com www.myapp.com;
    
    # Redirect HTTP to HTTPS
    return 301 https://$server_name$request_uri;
}

server {
    listen 443 ssl;
    server_name myapp.com www.myapp.com;

    # SSL — Let's Encrypt wasn't available until December 2015
    # In mid-2015 we still bought certificates ($10/year from NameCheap)
    ssl_certificate     /etc/ssl/myapp/myapp.crt;
    ssl_certificate_key /etc/ssl/myapp/myapp.key;
    ssl_protocols       TLSv1 TLSv1.1 TLSv1.2;
    ssl_ciphers         HIGH:!aNULL:!MD5;

    # Serve static files directly from Nginx (don't hit Node.js)
    location /static/ {
        alias /var/www/myapp/public/;
        expires 30d;
        add_header Cache-Control "public, immutable";
        gzip_static on;
    }

    # Proxy API requests to Node.js
    location / {
        proxy_pass         http://nodejs;
        proxy_http_version 1.1;
        proxy_set_header   Upgrade    $http_upgrade;
        proxy_set_header   Connection 'upgrade';
        proxy_set_header   Host       $host;
        proxy_set_header   X-Real-IP  $remote_addr;
        proxy_set_header   X-Forwarded-For $proxy_add_x_forwarded_for;
        proxy_set_header   X-Forwarded-Proto $scheme;
        proxy_cache_bypass $http_upgrade;
        proxy_read_timeout 60s;
    }
}

sudo ln -s /etc/nginx/sites-available/myapp /etc/nginx/sites-enabled/
sudo nginx -t  # Test config
sudo service nginx reload

---

Deployment Script

#!/bin/bash
# deploy.sh — zero-downtime deploy

APP_DIR="/var/www/myapp"
REPO="git@github.com:company/myapp.git"

cd $APP_DIR

echo "Pulling latest code..."
git pull origin main

echo "Installing dependencies..."
npm install --production

echo "Running migrations..."
node scripts/migrate.js

echo "Reloading app (zero downtime)..."
pm2 reload myapp
# reload = graceful restart: starts new processes before stopping old ones
# restart = hard restart: brief downtime
# In cluster mode, reload cycles workers one at a time

echo "Deploy complete"
pm2 status

---

PM2 Commands Reference

# Monitor
pm2 status           # Process list with CPU/memory
pm2 monit            # Real-time monitoring dashboard
pm2 logs myapp       # Tail logs
pm2 logs myapp --lines 200  # Last 200 lines

# Control
pm2 reload myapp     # Zero-downtime restart (cluster mode)
pm2 restart myapp    # Hard restart
pm2 stop myapp
pm2 delete myapp

# After code deploy
pm2 reload ecosystem.json --update-env  # Also reload env vars

# Check what's eating memory
pm2 describe myapp   # Detailed process info

---

WebSocket Support

# For Socket.io or ws — add to location block:
location /socket.io/ {
    proxy_pass         http://nodejs;
    proxy_http_version 1.1;
    proxy_set_header   Upgrade    $http_upgrade;
    proxy_set_header   Connection "upgrade";
    proxy_read_timeout 600s;    # Keep WS connections alive longer
}

---

Common Problem: "Cluster mode but still single process"

# Check
pm2 describe myapp | grep exec_mode
# Should say: cluster_mode

# If running in fork mode by mistake:
pm2 delete myapp
pm2 start ecosystem.json  # Re-read config file

---

Benchmark: Fork vs Cluster Mode

On a 4-core VPS with a CPU-bound route (image resize):

Mode	Requests/sec	CPU utilization
Fork (1 process)	420 req/s	25% (1 core maxed)
Cluster (4 workers)	1,580 req/s	94% (all cores used)

For I/O-bound workloads (database queries, file reads), the difference is smaller because Node's event loop handles concurrent I/O without multiple processes. But for CPU work — image processing, PDF generation, crypto — cluster mode is essential.