Running Node.js Projects in Production with PM2: Panel Guide

Node.js is a runtime that works with an event loop on a single thread. This architecture has tremendous performance for API servers, but it has two limits: if there is a crash, the process dies, it does not work in parallel as many CPU cores. PM2 is a process manager designed to overcome these two limits. It is one of the leading tools used for Production Node.js deployment.

In this guide, we will explain what PM2 does in production, which settings you care about and why, and how you can manage these settings from the interface with VDS Panel.

Bir bakışta

99.9%

Uptime target

4-8×

throughput with cluster

<1sec

automatic restart

Downtime reload

What does PM2 do?

PM2 “daemonises” your Node.js processes. Well:

Restarts the process when it crashes
Opens more than one process and distributes the load (cluster mode)
Writes the logs to the file and rotates them
Manages uninterrupted traffic between restarts
Monitors Memory/CPU, restarts process that exceeds limits

It is also possible to run Node.js with systemd, but you need to install the above features one by one.

Why is cluster mode important?

Node.js runs single thread by default. If you are running a Node process on a 4-core VPS, your 75% CPU capacity remains idle. PM2 cluster mode opens as many worker processes as there are CPU cores and distributes the load among them:

4-8×

more throughput

Cluster mode compared to single instance mode on 4-core VPS.

Classic installation vs via panel

Classical

Manual PM2 with terminal

npm install pm2 -g (global install)
Write ecosystem.config.js
Learn pm2 start/stop/reload commands
Log tracking with pm2 logs
Starting at boot with pm2 save + pm2 startup
SSH + manual reload for update

interface

with VDS Panel

Add project, select Node.js
Set the number of cluster instances with slider
Memory limit, auto-restart, log rotation automatic
Live log stream on the panel
Autostart on boot (already installed)
Git push → automatic zero-downtime reload

Step by step installation with panel

01

Add the Node.js project

Link or ZIP the GitHub repository. The panel detects package.json, uses 'scripts.start' or the entry point you specify.
02

Set up cluster

Slider for number of instances: 1 (fork mode) to the number of CPUs of your VPS. For most APIs, 2-4 is ideal.
03

Set memory limit

Like 256MB, 512MB, 1GB. The worker exceeding this limit will automatically restart. Memory leak is tolerated, the app continues to work.
04

Enter env variables

NODE_ENV=production, DATABASE_URL, API keys. It is not written to Git, it is stored in the panel secret vault and injected at runtime.
05

Connect domain

Custom domain name or subdomain. The panel automatically obtains the Let's Encrypt certificate and directs the nginx reverse proxy to your 3000 port.
06

Press Deploy

Your app is live in ~90 seconds. When you do a git push, zero-downtime reload is triggered.

Zero-downtime reload

One of PM2’s most powerful features is graceful reload. When you deploy new code:

PM2 starts new workers
New workers start listening
Waits for old workers to finish their current requests (graceful shutdown)
Old workers close down

The user does not feel any interruption. VDS Panel makes every deployment with this strategy.

Graceful shutdown signal

If your Node.js application is listening for the SIGINT or SIGTERM signal, when PM2 is shutting down, it gives your application 10 seconds to perform cleaning tasks such as closing open DB connections and flushing the cache. The panel sets this time by default.

Defense against memory leaks

Node.js memory leaks are common; especially when the event listener is forgotten and the global cache grows. The P0 setting of PM2 provides the solution:

In the panel interface, this value is a slider. If you specified 512M, PM2 silently restarts a worker when it exceeds 512MB. Other workers continue to handle traffic.

DEV dependencies in Production

It installs with panel P0, dev dependencies are not installed. If you import a package in P1 at runtime, the production build will fail. You will see these errors in the panel build log.

Log management

PM2 writes P0 and P1 for each worker. Over time, these files become full. The panel automatically activates log rotation:

Rotate when file size reaches 100 MB
5 old files are kept
Logs are streamed live from the panel UI (without opening a terminal)

Monitoring and alerts

For each PM2 process, the panel:

CPU and memory graph flows live
Number of restarts is monitored (too many restarts = bug signal)
Response time telemetry (if express-status-monitor or similar is added)
Automatic cpulimit daemon launch when CPU limit is exceeded (protecting neighboring projects)

Node.js frameworks

VDS Panel has been tested with and automatically detects the following frameworks:

Express, common, fast, simple
NestJS, enterprise-grade, TypeScript focused
Fastify, ultra-fast, built-in JSON-schema validation
Koa.js, the modern successor from the creators of Express
Hapi, configuration-oriented, wide plugin ecosystem
AdonisJS, Laravel-like MVC

The panel reads your P0, detects the correct framework, and selects the appropriate start script.

Conclusion

PM2 is a must for Node.js production. Cluster mode, automatic restart, memory limit, log rotation, zero-downtime reload, all come with a single daemon. VDS Panel allows you to manage these settings from the interface without memorizing the terminal command.

To get your Node.js project live, you can complete the panel setup in 8 minutes and see your first project by requesting a demo from our contact form.