The Problem

You're using the awesome JSDOM package to process a few hundred html files and your process crashes with an error like this:

FATAL ERROR: CALL_AND_RETRY_LAST Allocation failed - JavaScript heap out of memory

If you're sure your code it not accidentally holding on to a reference to (nodes from) prior window objects, then the most likely cause of the apparent memory leak is that you're calling JSDOM synchronously, whereas JSDOM executes garbage collection asynchronously.

The Solution

In short, you need to let the event loop execute between calls to new JSDOM(...). If you're not used to thinking in async, that may sound daunting, but it's just a matter of creatively using setImmediate to call the next iteration of your task list.

One possible solution looks like this:

const {JSDOM}=require('jsdom');

var tasks = [...];

do_tasks() // hoisting is wierd

function do_tasks(i){
    if(typeof i === "undefined") i = 0
    if(i >= tasks.length){
        console.log("All done!")
        return
    }
    console.log(`about to run task: ${i}`)
    do_one_task(/* task to run         */ task[i], 
                /* run next iteration  */ () => setImmediate(() => do_tasks(i + 1)) 
                )
}

function do_one_task(task,next){

    dom = new JSDOM( task.htmlString, {...} )

    ... do things with the dom

    dom.window.close();
    next()
}

Bonus

If calling new JSDOM(...) prints a bunch of errors to the console having to to with not being able to parse CSS files, pass a virtual console in the options to JSDOM, like so:

const {JSDOM, VirtualConsole}=require('jsdom');
const vc = new VirtualConsole();
var jsdomOptions = {virtualConsole:vc}

... then:

dom = new JSDOM( task.htmlString, jsdomOptions )
``

jdthorpe/Dealing with JSDOM Memory Usage.md

The Problem

The Solution

Bonus