Error starting Chrome with puppeteer when deploying headless-chrome functions example

[mattwelke]

I was looking at the documentation for Cloud Functions (https://cloud.google.com/functions/docs/concepts/nodejs-10-runtime) and wanted to deploy one of the example functions in this repo, the headless-chrome example. My deploy command was based on what I was reading in the docs site:

gcloud functions deploy screenshot --runtime nodejs10 --trigger-http --project <redacted>

The deploy succeeds, but then in the web browser when I visit the URL, the function displays the expected message if I don't provide the URL query string parameter:

?url=https://example.com
Please provide URL as GET parameter, for example:

But it fails if I do provide a URL query string parameter, displaying the following message in the browser:

Error: could not handle the request

And the following message in the Stackdriver logs for the function:

TimeoutError: Timed out after 30000 ms while trying to connect to Chrome! The only Chrome revision guaranteed to work is r672088 at Timeout.onTimeout (/srv/functions/node_modules/puppeteer/lib/Launcher.js:351:14) at ontimeout (timers.js:436:11) at tryOnTimeout (timers.js:300:5) at listOnTimeout (timers.js:263:5) at Timer.processTimers (timers.js:223:10)

Despite the error message mentioning "30000 ms", this happens within a few seconds.

[mattwelke]

I was able to get the function to work properly (where I see an image in the response when I give it a URL), I had to change my code to the following:

'use strict';

// [START functions_headless_chrome]
const puppeteer = require('puppeteer');

exports.screenshot = async (req, res) => {
  console.log('Beginning function')

  const browser = await puppeteer.launch({
    headless: true,
    args: ['--no-sandbox'],
  });
  console.log('Puppeteer launched');

  const page = await browser.newPage();
  console.log('Puppeteer page created');

  const {url} = req.query;

  if (!url) {
    return res.send(
      'Please provide URL as GET parameter, for example: <a href="?url=https://example.com">?url=https://example.com</a>'
    );
  }

  await page.goto(url);
  const imageBuffer = await page.screenshot();
  console.log('Screenshot created');

  res.set('Content-Type', 'image/png');
  res.send(imageBuffer);
};
// [END functions_headless_chrome]

I found that I had to add the headless: true option as I launch Puppeteer (which is recommended in the Puppeteer docs), and I had to get rid of the main function since it wasn't running (confirmed by adding a console.log in it and seeing nothing in the logs). Instead, as you can see in the code above, I had to have the Puppeteer commands run each time a request comes in. I had to make both of these changes to get it to work.

[fhinkel]

@Welkie Thanks for reporting the issue. @ace-n , this is a problem with one of the functions examples, can you have a look? cc @steren

[ace-n]

This sample isn't used in the docs. @labtopia @stew-r should we fix it + add it there, or remove it from this repo?

[ace-n]

(Followed up internally - @labtopia and I decided to add this to the docs, but we aren't really sure where to put it.)

[stew-r]

I know a few customers are trying to do this (though the performance currently isn't great). Would be in favor of documenting the use case rather than throwing it out.

…

On Wed, Jul 24, 2019 at 12:49 PM Ace Nassri ***@***.***> wrote: (Followed up internally - @labtopia <https://github.com/labtopia> and I decided to add this to the docs, but we aren't really sure where to put it.) — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#1395?email_source=notifications&email_token=AHYELQDTLOWCREBZHLJSH5TQBCW3NA5CNFSM4H6RD3L2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOD2XNLTI#issuecomment-514774477>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AHYELQGHEZQNOZS7MF76AWDQBCW3NANCNFSM4H6RD3LQ> .

-- Stewart Reichling | Product Manager, Serverless | stewartr@google.com | 650.906.3381

[mattwelke]

I actually found this example really interesting. The performance isn't great, but that's exactly why I wanted to use Cloud Functions for this. I have a use case where I need to interact with a website that doesn't have an API, so I could use Puppeteer. And I could use a FaaS service so that I don't have to worry about running out of memory on a VM or App Engine etc from more than one user performing the request that runs Puppeteer at a time.

[danoc]

HI! I noticed that the examples were removed a few days ago in #1461.

Is Puppeteer still supported in Cloud Functions?

[clement128]

Same questions? I am planing to use it !!

[ace-n]

Yes, it should be supported. We removed these tutorials for a few reasons: 1. they were fairly niche 2. we don't have enough people to support/maintain the associated samples ourselves. If you'd like, I can look into adding these to the community-maintained tutorials repo (https://github.com/GoogleCloudPlatform/community) cc @steren as FYI

…

On Tue, Sep 3, 2019, 7:42 AM clement128 ***@***.***> wrote: Same questions? I am planing to use it !! — You are receiving this because you were assigned. Reply to this email directly, view it on GitHub <#1395?email_source=notifications&email_token=AARTZRL4HYG5SDWBL7CIMJTQHZZUTA5CNFSM4H6RD3L2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOD5YNWJQ#issuecomment-527489830>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AARTZRM4C2E5DMNWML4CPADQHZZUTANCNFSM4H6RD3LQ> .

[steren]

I confirm that the OS packages needed to run Headless Chrome (and thus puppeteer) will keep being present in the Cloud Functions environment.