trjhtr

I know that this is an impossible task against someone determined, but I'd like to keep mid-average users from being able to open more than one copy of my program at a time.

Here's the quick code I've made:

const tmp = require('tmp'),
 fs = require('fs'),
 path = require('path');

// Randomly generated, constant in each copy of the app
const app_code = '19c2cefbfa76b4887a1d4664704a3fe0';

// Write protection here, although it doesn't help much
const lock = tmp.fileSync( mode: 0444, prefix: app_code);

const filedir = path.dirname(lock.name);
const filename = path.basename(lock.name);

// Search for already running copies
fs.readdirSync(filedir).forEach(f => 
 // If it's made by the app and it's not mine
 if (f.includes(app_code) && f !== filename) 
 // Delete my lock and close
 fs.unlink(lock.name);
 process.exit(1);
 
);

// Watch for my own lock being deleted
fs.watch(filedir, function(event, who) 
 if (event === 'rename') who.includes(app_code)) 
 // Delete my lock and close
 fs.unlink(lock.name);
 process.exit(1);
 
 
);

// Main program content, use placeholder here:
setInterval(() => console.log("Running alone"), 1000);

There are a few issues that I see right off the bat:

• Different tmp folders being used


• Locks being left in place if the program is terminated, meaning no copies could run

Anything obvious I'm missing?

I suggest you to use resource based locking. Consider the following:

• your application uses well known TCP/UDP port. Only single process can bind it.


• your application uses some configuration/database/whatever files in predefined places. flock them. Same stands for pid files under /var/run.

It is still possible to separate resources with chroot or containers. But if nothing is shared are there any benefits in forced locking?

I'd like to keep mid-average users from being able to open more than one copy of my program at a time.

I guess mid-average users won't use network namespaces, chroots or containers. If they will - I bet they have the reasones so let them do the thing.