How to monitor performance, bottlenecks, and points of failure on a Socket.IO server

[ThePokyWizard]

I ran into a bizarre issue a few days ago. I have a server running Socket.IO with PM2 (the one provided by Socket.IO) for almost two years, with no reboots or PM2 restarts.

One day, the server’s RAM usage started to increase gradually throughout the day, eventually hitting 100%, and I had to reboot the machine.

How can I identify bottlenecks or failures on the server to prevent this from happening again and also determine the root cause of the problem?

For example, I can’t even know, either through logs or in real time, how many sockets are connected or even how many rooms exist or have been created.

[darrachequesne]

Hi!

You can either use:

• the Admin UI: https://socket.io/docs/v4/admin-ui/
• https://www.npmjs.com/package/socket.io-prometheus

[ThePokyWizard]

Thanks @darrachequesne
GPT also suggested an improvement to my code: for all socket.on events where I make a connection to MongoDB, for example, it was recommended to use async.
Does that make sense, or is it just nonsense from it?

Example:

socket.on('disconnect', async (reason) => {