pm2
pm2 copied to clipboard
Unitech
Better support for dotenv in cluster mode ?
Im trying to setup in production a nodeJs app in cluster mode with a .env file provided. I tried doing so using 2 approaches :
1 ) Using dotenv
I can get it working only if I change my application code adding "require('dotenv').config()". I don't want to delegate this responsability to the code. So I tried using this ecosystem file :
module.exports = {
apps: [{
name: 'api-dev',
script: 'server.js',
cwd: '/var/node-app/api-dev/current',
instances: 'max',
exec_mode: 'cluster',
log: '/var/log/pm2/api-dev.log',
error: '/var/log/pm2/api-dev.error.log',
merge_logs: true,
node_args: '-r dotenv/config', // launching node with dotenv here
env: {
NODE_ENV: 'production',
DOTENV_CONFIG_PATH: '/var/node-app/api-dev/current/.env', // setting the .env path here
},
}],
};
and when I execute "pm2 deploy ecosystem.config.js develop", my apps crash instantly and silently (pm2 logs doesn't log anything which is weird in a first place) I tried to play around with apps[0].env.DOTENV_CONFIG_PATH and apps[0].node_args, and never achieved to have a working clustered env for my app. It either crash or doesn't load env variables.
2 ) Using built-in NodeJS support for .env files
With this config :
module.exports = {
apps: [{
name: 'api-dev',
script: 'index.js',
cwd: '/var/node-app/api-dev/current',
instances: 'max',
exec_mode: 'cluster',
log: '/var/log/pm2/api-dev.log',
error: '/var/log/pm2/api-dev.error.log',
merge_logs: true,
node_args: '--env-file=.env', // enabling node support for env-file
}],
};
In a first place it does work. Then I tried to add a second app clustered as well and I noticed something weird aswell : The first app that is launched will successfully load env variables, but all other apps launched after the first wont load them. I had "applicationA" and "applicationB", when I first launch "applicationB" and then "applicationA", I will have 2 nodes working for applicationB and 2 in errors for application A. If I launch application A first, its the other way around.
Im using Node 20.11.0, pm2 5.3.1 on Debian 12 Bookworm
Here is my pm2 reports
--- PM2 report ----------------------------------------------------------------
Date : Wed Feb 14 2024 18:47:33 GMT+0100 (Central European Standard Time)
===============================================================================
--- Daemon -------------------------------------------------
pm2d version : 5.3.1
node version : 20.11.0
node path : /usr/bin/pm2
argv : /usr/bin/node,/usr/lib/node_modules/pm2/lib/Daemon.js
argv0 : node
user : boss
uid : 1000
gid : 1000
uptime : 23min
===============================================================================
--- CLI ----------------------------------------------------
local pm2 : 5.3.1
node version : 20.11.0
node path : /usr/bin/pm2
argv : /usr/bin/node,/usr/bin/pm2,report
argv0 : node
user : boss
uid : 1000
gid : 1000
===============================================================================
--- System info --------------------------------------------
arch : x64
platform : linux
type : Linux
cpus : Intel(R) Xeon(R) Gold 5120 CPU @ 2.20GHz
cpus nb : 2
freemem : 1336549376
totalmem : 2026278912
home : /home/boss
===============================================================================
--- PM2 list -----------------------------------------------
┌────┬─────────────────┬─────────────┬─────────┬─────────┬──────────┬────────┬──────┬───────────┬──────────┬──────────┬──────────┬──────────┐
│ id │ name │ namespace │ version │ mode │ pid │ uptime │ ↺ │ status │ cpu │ mem │ user │ watching │
├────┼─────────────────┼─────────────┼─────────┼─────────┼──────────┼────────┼──────┼───────────┼──────────┼──────────┼──────────┼──────────┤
│ 3 │ api-dev │ default │ 2.30.0 │ cluster │ 28572 │ 22m │ 0 │ online │ 0% │ 128.8mb │ boss │ disabled │
│ 4 │ api-dev │ default │ 2.30.0 │ cluster │ 28582 │ 22m │ 0 │ online │ 0% │ 128.4mb │ boss │ disabled │
│ 1 │ auth-dev │ default │ 2.7.0 │ cluster │ 28486 │ 22m │ 0 │ online │ 0% │ 74.3mb │ boss │ disabled │
│ 2 │ auth-dev │ default │ 2.7.0 │ cluster │ 28496 │ 22m │ 0 │ online │ 0% │ 75.1mb │ boss │ disabled │
└────┴─────────────────┴─────────────┴─────────┴─────────┴──────────┴────────┴──────┴───────────┴──────────┴──────────┴──────────┴──────────┘
Module
┌────┬──────────────────────────────┬───────────────┬──────────┬──────────┬──────┬──────────┬──────────┬──────────┐
│ id │ module │ version │ pid │ status │ ↺ │ cpu │ mem │ user │
├────┼──────────────────────────────┼───────────────┼──────────┼──────────┼──────┼──────────┼──────────┼──────────┤
│ 0 │ pm2-intercom │ 1.0.0 │ 28460 │ online │ 0 │ 0% │ 52.4mb │ boss │
└────┴──────────────────────────────┴───────────────┴──────────┴──────────┴──────┴──────────┴──────────┴──────────┘
===============================================================================
--- Daemon logs --------------------------------------------
/home/boss/.pm2/pm2.log last 20 lines:
PM2 | 2024-02-14T18:24:29: PM2 log: RPC socket file : /home/boss/.pm2/rpc.sock
PM2 | 2024-02-14T18:24:29: PM2 log: BUS socket file : /home/boss/.pm2/pub.sock
PM2 | 2024-02-14T18:24:29: PM2 log: Application log path : /home/boss/.pm2/logs
PM2 | 2024-02-14T18:24:29: PM2 log: Worker Interval : 30000
PM2 | 2024-02-14T18:24:29: PM2 log: Process dump file : /home/boss/.pm2/dump.pm2
PM2 | 2024-02-14T18:24:29: PM2 log: Concurrent actions : 2
PM2 | 2024-02-14T18:24:29: PM2 log: SIGTERM timeout : 1600
PM2 | 2024-02-14T18:24:29: PM2 log: ===============================================================================
PM2 | 2024-02-14T18:24:29: PM2 log: App [pm2-intercom:0] starting in -fork mode-
PM2 | 2024-02-14T18:24:29: PM2 log: App [pm2-intercom:0] online
PM2 | 2024-02-14T18:24:35: PM2 log: App [auth-dev:1] starting in -cluster mode-
PM2 | 2024-02-14T18:24:35: PM2 log: App [auth-dev:1] online
PM2 | 2024-02-14T18:24:35: PM2 log: App [auth-dev:2] starting in -cluster mode-
PM2 | 2024-02-14T18:24:35: PM2 log: App [auth-dev:2] online
PM2 | 2024-02-14T18:25:11: PM2 log: App [api-dev:3] starting in -cluster mode-
PM2 | 2024-02-14T18:25:11: PM2 log: App [api-dev:3] online
PM2 | 2024-02-14T18:25:11: PM2 log: App [api-dev:4] starting in -cluster mode-
PM2 | 2024-02-14T18:25:11: PM2 log: App [api-dev:4] online
PM2 | {"log.level":"warn","@timestamp":"2024-02-14T17:25:12.396Z","log":{"logger":"elastic-apm-node"},"ecs":{"version":"1.6.0"},"message":"Invalid \"serverUrl\" config value, it must be a valid URL"}
PM2 | {"log.level":"warn","@timestamp":"2024-02-14T17:25:12.485Z","log":{"logger":"elastic-apm-node"},"ecs":{"version":"1.6.0"},"message":"Invalid \"serverUrl\" config value, it must be a valid URL"}
(The app "api-dev" is actually no working since it didn't load the env variables)
Did someone already achieved to launch in cluster mode, multiple nodeJS app with a .env file ? (Thats my first issue I submit, sorry if it isn't clear)
If you run in fork mode, this should work correctly.
In cluster mode, the cwd option is not being used, which there is a related issue for https://github.com/Unitech/pm2/issues/5722
node_args: '-r dotenv/config' works for me under pm2 5.3.1 cluster mode.
But it needs a clean pm2 daemon.
If I stop pm2 stop <app> and restart pm2 start <app>, there is an error says dotenv/config not found,
I have to run pm2 kill, then run pm2 start <app> in a new pm2 daemon.
