Deploy

& monitor

With

It's time to Deploy!

Keep Calm And wait for it...

How Can it Possibly Go Wrong?

How Can it Possibly Go Wrong?

  • Your app crashes!

  • Your server crashed and restarted, but your app didn’t…

  • Your app is under using the server resources (i.e. 8% cpu)

  • Your server is out of memory

  • You need to find out which app on
    your server is causing the memory leak

  • You want to inspect the logs

  • How will you monitor servers in real time once deployed
    to make sure all is running smoothly

Enter

PM2 is a toolbox to help you deploy
and manage your apps in Production

It ensures your app is up,
it manages processes, cluster and more.

gives you a structured workflow
by helping you with:

Resources management
and optimization (cpu, memory)

gives you a structured workflow
by helping you with:

Deployment + Environment configuration

gives you a structured workflow
by helping you with:

Apps management and monitoring

gives you a structured workflow
by helping you with:

Access your server logs

gives you a structured workflow
by helping you with:

Restarting your app on changes push

gives you a structured workflow
by helping you with:

Auto start your app on OS reboot

gives you a structured workflow
by helping you with:

Extend PM2 capabilities using Modules

is widely used, battle tested and embraced by the Node.js community!

  • +3M installations

  • +15K stars on Github

  • +2.7K commits

  • +750 tests

  • +130 contributors

  • Ranked as 84th js project on Github

npm i -g pm2

Getting started

  • Enable auto completion
     

  • Generate OS startup script to respawn pm2 on server boot 

pm2 completion install

Setup

pm2 startup

  • Simplest way to start your app

  • It auto restarts your app on crash

  • Monitor your app

  • Gives you access to your logs

pm2 start app.js

Launching your apps

process.yml:

pm2 start process.yml

Application declaration

You can also create a configuration file to manage multiple applications:

apps:
  - script   : app.js
    instances: 4
    exec_mode: cluster
  - script : worker.js
    watch  : true
    env    :
      NODE_ENV: development
    env_production:
      NODE_ENV: production

processes commands

pm2 stop    <app-name|id|'all'>

pm2 restart <app-name|id|'all'>

pm2 delete  <app-name|id|'all'>

has two modes - fork and cluster

The default is fork.

It means that behind the scenes PM2 will use child_process.fork

Which will allow you to spawn other processes, but all of them will use the same cpu resource

 

Fork vs Cluster mode

A multi-core machine running your app

Without Cluster mode enabled…

 

Cluster mode

Using the built-in Node.js cluster module
Makes it easy to use multiple cores machines

  pm2 start app.js -i <num-of-app-instances>
  pm2 start app.js -i 0

auto detecting the number of cpu’s

Manage Clustered apps

Reload with Zero downtime

  pm2 scale <app-name> <app-instances>

Scale up/down process number

  pm2 reload <app-name|’all’>

Architectural option

My suggested setup -
use nginx at the front as a load-balancer (reverse proxy).

You can still use pm2 cluster to take advantage
of your multi-core instances

Remember your app code needs to be stateless.
For sharing the same state across all instances and processes use Redis.

Inspecting Your apps

Listing all your running processes on a server

  pm2 list

Inspecting Your apps

Monitoring CPU & Memory

  pm2 monit

Inspecting Your apps

Even more info…

  pm2 show <app-name|id>

Deployment

Deployment

When you manage many micro-services
Configuring deployments manually sucks…

DEBUG="*" NODE_ENV="production" pm2 start app.js --name "API" 
-i 8 --watch --max-memory-restart="300M" -- -a 23

DEBUG="app2:*" NODE_ENV="production" pm2 start app2.js --name 
"WORKER" -i 8 --watch -- -a prod

DEBUG="app3:*" NODE_ENV="production" pm2 start app3.js --name 
"BDD" -i 8 --watch -- -a malo

Deployment

Generates a sample ecosystem.json file to store all settings about deploy

pm2 ecosystem

Deployment

Using ecosystem.json

# start
pm2 start ecosystem.json 

# restart
pm2 restart ecosystem.json

# stop
pm2 stop <app-name|app-id|’all’>

Deployment

Add deploy section to ecosystem.json

Deployment commands

# setup deployment to remote server
pm2 deploy ecosystem.json production setup


# update your app on remote server
pm2 deploy ecosystem.json production


# execute a pm2 command process in remote server
pm2 deploy ecosystem.json production <command> <options>

# example - revert to the <n> deployment
pm2 deploy ecosystem.json production revert <n>

Deploy & Monitor with PM2

By Yariv Gilad

Deploy & Monitor with PM2

  • 1,777

More from Yariv Gilad