Here's the configuration I came up with for my PM2 API instances on my own server, hopefully it'll be useful.

ecosystem.config.js

module.exports = {
  apps : [{
    name: 'Sample API',
    exec_mode: 'cluster',
    instances: 4,
    watch: true,
    // I had my database under `./cinema` so make sure anything changing inside the project's root is ignored
    // if you have watch turned on above
    ignore_watch: ['node_modules', 'cinema'],
    script: './dist/index.js',
    // I later use these to run different instances, notice that there is an option to add env names to instance names that
    // will probably make it easier for you to deal with instances and work with them via PM2 cli. You can find it in the docs.
    env: {
      'NODE_ENV': 'test'
    },
    env_dev: {
      'NODE_ENV': 'dev'
    },
    env_prod: {
      'NODE_ENV': 'production'
    }
  }],
  deploy : {
    prod : {
      user : 'USERNAME',
      host : 'SERVER_ADDRESS/IP_ADDRESS',
      ref  : 'origin/master',
      repo : 'git@github.com:YOUR_USERNAME/REPO_NAME.git',
      // Pay attention that i'm specifing which env I'm targeting for this deployment environement. It should match your env
      path : '/home/USERNAME_ON_SERVER/{PATH_TO_PROJECT_ROOT}/prod/',
      // Here I'm telling it to use my exisiting SSH key on my local to connect to Github to fetch latest changes, You can
      // create a new one on your server and tell PM2 to use that one.
      'ssh_options': 'ForwardAgent=yes',
      'pre-deploy-local': 'echo "Started Deploying Production Build! :)"',
      // Here's where magic happens. You can gracefully reload instances of `prod` clusters you started.
      'post-deploy' : 'npm install && pm2 reload ecosystem.config.js --env prod',
      'pre-setup': 'echo "Started setting-up Deployment for Production"',
      'post-setup': 'echo "Setup Finished!"'
    },
    dev : {
      user : 'USERNAME',
      host : 'SERVER_ADDRESS/IP_ADDRESS',
      ref  : 'origin/master',
      repo : 'git@github.com:YOUR_USERNAME/REPO_NAME.git',
      path : '/home/USERNAME_ON_SERVER/{PATH_TO_PROJECT_ROOT}/dev/',
      'ssh_options': 'ForwardAgent=yes',
      'pre-deploy-local': 'echo "Started Deploying Development! :)"',
      'post-deploy' : 'yarn && yarn compile && pm2 startOrReload ecosystem.config.js --env dev',
      'pre-setup': 'echo "Started setting-up Deployment for Development"',
      'post-setup': 'echo "Setup Finished!"'
    },
    test: {
      user : 'USERNAME',
      host : 'SERVER_ADDRESS/IP_ADDRESS',
      // This line is a bit tricky, sometimes when you're testing a specific branch of your code, you might want to
      // tell github to fetch that branch on server to stage it, here, by doing so, I can later tell PM2 in command line
      // that I need the ref to be, say, `masoud/test-something` and on the server, it'd fetch that branch and run `test`
      // with the branch I told it to. Pretty handy for testing and staging.
      ref  : process.env.BRANCH || 'dev',
      repo : 'git@github.com:YOUR_USERNAME/REPO_NAME.git',
      path : '/home/USERNAME_ON_SERVER/{PATH_TO_PROJECT_ROOT}/test/',
      'ssh_options': 'ForwardAgent=yes',
      'pre-deploy-local': `echo "Started Deploying ${process.env.BRANCH || 'dev'} branch! :)"`,
      'post-deploy' : 'ls -la && yarn && yarn compile && pm2 startOrReload ecosystem.config.js --env test',
      'pre-setup': 'echo "Started setting-up Deployment"',
      'post-setup': 'echo "Setup Finished!"'
    }
  }
}