pm2-meteor is a CLI tool, that will deploy your Meteor app (from your dev machine or from git) as Nodejs bundle and run it with PM2. (tested with Ubuntu and Freebsd hosts)
- you can deploy from a git repo
- you can deploy to almost every server construction (also freebsd jails)
- it uses PM2 to run your Meteor Apps
- reads your Meteor settings file from a path for you
- you can configure the directory where your apps will be deployed
- you can scale your App in realtime with one command
- make generic pm2-meteor settings with a
pm-meteor.js
file which module.exports settings
PM2 is a process manager, that will run/restart Nodejs apps, just like forever - but:
- PM2 has build in load balancing and scaling features
- PM2 also runs bash / python / ruby / coffee / php / perl
- We tested PM2 with some of our complex Meteor apps and it performed well (while forever crashed them without any notable reasons)
- more under pm2.keymetrics.io
$ npm i -g pm2-meteor
You should have Nodejs, npm and PM2 installed on your host machine.
pm2-meteor won't install global tools on your server.
$ mkdir ninjaApp_deployment
$ cd ninjaApp_deployment
$ pm2-meteor init
(The pm2-meteor wizzard will ask you some questions and prefill the configuration file.)
{
// the name of your app
"appName": "ninjaApp",
// where your app is located
"appLocation": {
"local": "~/Workspace/ninjaApp"
// or you can also deploy with git ;-)
// (use username:password@github.com/... for authentication)
// "git": "https://user:password@github.com/andruschka/ninjaApp",
// "branch": "production"
},
// where the meteor settings are located
"meteorSettingsLocation":"~/Workspace/ninjaApp/settings/production.json",
"meteorSettingsInRepo": false,
// or RELATIVE to app root, if settings are located in git repo (this is not a good idea...)
// "meteorSettingsLocation":"settings/production.json",
// "meteorSettingsInRepo": true,
// build flags for Meteor
"meteorBuildFlags": "--architecture os.linux.x86_64"
// runs as command in the meteor app before building
"prebuildScript": "",
// say you are still using meteorite and want to install deps before deploying:
// "prebuildScript": "mrt install",
// the env vars
// (METEOR_SETTINGS will be generated from your meteor-settings file)
"env": {
"ROOT_URL": "http://ninja.my-host.com",
"PORT": 4004,
"MONGO_URL": "mongodb://localhost:27017/ninjaApp"
},
// infos for deployment
"server": {
"host": "my-host.com",
"username": "nodejs",
"password": "trustno1",
// or auth with pem file
// "pem":"~/.ssh/id_rsa",
// optional - set port
// "port": "22",
// this dir will contain your apps
// (app will be deployed to /opt/pm2-meteor/ninjaApp)
"deploymentDir": "/opt/pm2-meteor",
// optional - will source a profile before executing tasks on the server
// "loadProfile": "",
// optional - NVM Support
// if you are using nvm - make sure to fill out bin ("~/.nvm/nvm.sh" in most cases)
// ! using multiple node versions - coming soon !
// "nvm": {
// "bin": "",
// "use": ""
// },
// exec mode for pm2
"exec_mode": "cluster_mode",
"instances": 2
// set some free ports if you want to use fork_mode with several instances
// (e.g. you want to handle loadbalancing yourself)
// "freePorts": [3001, 3002]
},
// optional - set this one if you want to undeploy your app
// "allowUndeploy": true
// optional - set this if you want to specify timestamps to the pm2 log-files
// "log_date_format": "YYYY-MM-DD HH:mm Z"
}
$ pm2-meteor deploy
If you already have deployed this app before, the old app tar-bundle will be moved to a ./backup directory.
$ pm2-meteor reconfig
Will send new pm2-env file to server and hard-restart your app.
$ pm2-meteor revert
Will unzip the old bundle.tar.gz and restart the app
$ pm2-meteor start
$ pm2-meteor stop
$ pm2-meteor status
$ pm2-meteor logs
Start 2 more instances:
$ pm2-meteor scale +2
Down/Upgrade to 4 instances
$ pm2-meteor scale 4
To delete your app from the PM2 deamon and delete all app files add "allowUndeploy":true to your pm2-meteor setting, then:
$ pm2-meteor undeploy
$ pm2-meteor generateBundle
then transfer it to your machine, unzip it and run
$ pm2 start pm2-env.json
Deploy from a private github repo and start 2 load balanced instances:
{
"appName": "todos",
"appLocation": {
"git": "https://andruschka:bestPass123@github.com/andruschka/todos.git",
"branch": "master"
},
"meteorSettingsLocation": "settings/production.json",
"prebuildScript": "mrt install",
"meteorBuildFlags": "--architecture os.linux.x86_64",
"env": {
"PORT": 3000,
"MONGO_URL": "mongodb://localhost:27017/todos",
"ROOT_URL": "http://todos.my-host.co"
},
"server": {
"host": "my-host.co",
"username": "nodejs",
"pem": "~/.ssh/id_rsa",
"deploymentDir": "/home/nodejs/",
"exec_mode": "cluster_mode",
"instances": 2
}
}
Deploy a local app and run app in fork-mode:
(using a pm2-meteor.js file and passing the MONGO_URL via bash env var
e.g.: MONGO_URL='mongodb://user:pw@host/db' pm2-meteor deploy
)
const appName = "todos"
const mongoUrl = process.env.MONGO_URL
module.exports = {
appName, // <~ just ES6 stuff for appName: appName
"appLocation": {
"local":"~/Workspace/todos"
},
"meteorSettingsLocation": "~/Workspace/todos/settings/production.json",
"prebuildScript": "",
"meteorBuildFlags": "--architecture os.linux.x86_64",
"env": {
"PORT": 3000,
"MONGO_URL": mongoUrl, // <~
"ROOT_URL": "http://todos.my-host.co"
},
"server": {
"host": "my-host.co",
"username": "nodejs",
"pem": "~/.ssh/id_rsa",
"deploymentDir": "/home/nodejs/",
"exec_mode": "fork_mode",
"instances": 1
}
}
Deploy a local app and run app in fork-mode with several instances (for managing load balancing yourself):
{
"appName": "todos",
"appLocation": {
"local":"~/Workspace/todos"
},
"meteorSettingsLocation": "~/Workspace/todos/settings/production.json",
"prebuildScript": "",
"meteorBuildFlags": "--architecture os.linux.x86_64",
"env": {
"PORT": 3000, // first fork_mode instande willl use this port ...
"MONGO_URL": "mongodb://localhost:27017/todos",
"ROOT_URL": "http://todos.my-host.co"
},
"server": {
"host": "my-host.co",
"username": "nodejs",
"pem": "~/.ssh/id_rsa",
"deploymentDir": "/home/nodejs/",
"exec_mode": "fork_mode",
"instances": 3,
"freePorts": [3002, 3003] // ... further fork_mode instances will use this ports
}
}