`setdefaulttimeout` is causing issues with pyOpenSSL

[ejulio]

This issue when running selenium-wire in scrapy cloud is caused because of a call to socket.setdefaulttimeout in sh_scrapy.crawl.

The related issue in pyOpenSSL pyca/pyopenssl#168

The cause of the issue is this call
https://github.com/scrapinghub/scrapinghub-entrypoint-scrapy/blob/master/sh_scrapy/crawl.py#L27

A minimal example to reproduce the issue

Clone https://github.com/pawelmhm/quotesbot/tree/selenium

Working case

cd quotesbot
docker built -t selenium-wire-scrapy-issue .
docker run selenium-wire-scrapy-issue scrapy crawl toscrape-css

It should finish properly scraping 100 items and no errors

Now, update settings.py and add the following to the top of the file

import socket
socket.setdefaulttimeout(60)

Then rebuild and rerun the container

docker built -t selenium-wire-scrapy-issue .
docker run selenium-wire-scrapy-issue scrapy crawl toscrape-css

Now it should fail, scraping only 10 items and showing an error net::ERR_SSL_PROTOCOL_ERROR in the logs

[Foxtrod89]

I'm using selenium wire which requires certificate + key. I just followed https://docs.zyte.com/misc/ca.html for certificate creation but key is out of discussion.
According to this https://github.com/wkeeling/selenium-wire#using-your-own-certificate, selenium Wire sets it mandatory for my own certificates.
And now I'm getting 127.0.0.1:61202: Invalid certificate, closing connection. Pass --ssl-insecure to disable validation.