Optimizing, improving, and measuring speed/performance

[icarusmiles]

Running Puppeteer v0.13.0 (In headless) on Ubuntu Server LTS (and tried Windows 10).

I am trying to get my puppeteer system as optimized as possible. I do not need the pages to look pretty for screenshots, mainly obtaining data after filling some forms and doing simple actions.

My end-goal is to have a user-requested system that goes to 4 different sites (does a few actions specific to the user request), then returns fresh data.

What I've Done

• Forced Images to Not Load
• Using an already spawned chrome/puppeteer instance (connect())
• High Resource Allocation

---

1. Does anyone know other options to speed performance in this scenario?
2. Anyway to efficiently cache JS/CSS on the clientside for future requests?
3. Is there a way to measure the timing/performance of each individual action? (Connect, goto, click, etc.)

Sorry if this began to feel like a stackoverflow question, but I've invested a good sum of time in Puppeteer and it seems great. However if I can't find a way to speed up this process I may have to find another option. In addition I am having to plan out to pool Chrome instances due to the lack of browser contexts #85. This is due to IF multiple users request at the same time.

[HanXHX]

Hey @miles-collier

1. You can use something like this. But it doesn't work on all websites (don't XHR for images/css...). Check in headfull mode, first.

const block_ressources = ['image', 'stylesheet', 'media', 'font', 'texttrack', 'object', 'beacon', 'csp_report', 'imageset'];
page.on('request', request => {
	if (
		block_ressources.indexOf(request.resourceType) > 0
		// Be careful with above
		|| request.url.includes('.jpg')
		|| request.url.includes('.jpeg')
		|| request.url.includes('.png')
		|| request.url.includes('.gif')
		|| request.url.includes('.css')
	)
		request.abort();
	else
		request.continue();
});

2. No idea
3. The easiest way: https://nodejs.org/api/console.html#console_console_time_label

console.time("pagegoto");
response = await page.goto('https://github.com');
console.timeEnd("pagegoto");

[aslushnikov]

For (2), I'd try re-using userDataDir and rely on browser own HTTP cache.
Other then that, it looks like @HanXHX gave a reasonable answer.