Backstopjs hangs after 'reference' command

[rickdroce]

Once in a while, I experience a bug where backstopjs hangs in the terminal after the "reference" command is executed, without having the option to run further commands.

This wasn't a problem when I used to run tests manually, but now that I run backstopjs in a Jenkins job, it's dangerous since the job just builds forever since it doesn't receive any error.

As you can see in the screenshot the command executed successfully and then hang there for a few hours until my manual abortion of the build.

[garris]

Is it possible to migrate to the puppeteer engine? I highly recommend it if you can. The implantation is more stable.

If you can’t migrate you could try adding more memory to your container.

[rickdroce]

Right now I already set a --shm-size=2gb so I don't think it would need more.

I was considering the migration to CasperJs because I need to login to the website in order to make a couple of tests. By the way, for now, I'm not really satisfied with the results, the time needed for the full test went to 10mins from 4 mins And there are actually some layouts that break down

Gonna make some more tries

[Kenith]

Which engine you used for test?

If phantomjs, yes, the phantomjs has the hang issue... dam...

Solve the hang issue by following:

1. Set the following settings in backstop.json

"engine": "phantomjs",
  "casperFlags": [
    "--ignore-ssl-errors=true", 
    "--ssl-protocol=any",
    "--engine=phantomjs"
  ], 

2.Add following settings in casper/onBefore.js to avoid the PhantomJS hang issue. If not set this, the phantomjs has high possibilities to hang there forever.

casper.page.settings.resourceTimeout = 300000;
  casper.options.waitTimeout = 300000;
  casper.page.onResourceTimeout = function() {
    console.log("Phantomjs failed to donwload the resource in 5 minutes..."); 
    casper.bypass(1);
  };

If other engines, you could get the stable expertise here: https://github.com/garris/BackstopJS/issues/761#issuecomment-390492390

For docker use, you may add --privileged -v /dev/shm:/dev/shm to docker run

[rickdroce]

I took a look at the issue you linked and the only flag I was missing was "--hide-scrollbars" I'm using chromy I also added the "--privileged" arg to my docker run

I'll make some tests.

N.B. I also noticed that the 'hang' happens the most when the job is triggered by another build. So maybe it could really be a memory issue. Here are my args for the container: --shm-size=2gb --cap-add=SYS_ADMIN --privileged