Closed taylor-steve closed 4 months ago
We have similar issue(cannot determine loading status from tab crashed) but with 3.39.2 version. Issue appeared between 19-23 Jan 2024. We use docker and did not any changes to launch process of Capybara. We start use '--headless=new' option some time ago and it was works.
Some time ago we was have similar issue. It was fixed by use option '--disable-dev-shm-usage'. Issue details(not my but same): https://github.com/elgalu/docker-selenium/issues/20
Looks like removing --disable-site-isolation-trials
from the default Capybara driver definition fixes our issues for this particular set of tests.
--disable-dev-shm-usage
seems like it could be helpful in some other situations I'm seeing, I will try that.
Thanks for your help!
@twalpole does it make sense to remove the --disable-site-isolation-trials
from the default driver give that this seems to be a source of problems?
I don't think so --- the fact that it fixed one users problem doesn't mean that it's unnecessary for others --- There are so many chrome options which are necessary in varying environments that if you have one of those environments, especially CI environments, you will need to customize your own driver config. What's provided by default works for most people running locally
Looks like it was memory leak in Chrome. It should be fixed in v122. At least it works for us.
Meta
Capybara Version: 3.40.0 Driver Information (and browser if relevant): selenium-webdriver (4.17.0)
After switching to Capybara 3.40.0 on https://github.com/sul-dlss/sul-embed and when using
:selenium_chrome_headless
we're getting an initial failure which causes subsequent tests to fail (locally and on CI):That's from https://github.com/sul-dlss/sul-embed/blob/main/spec/features/metrics_spec.rb
After registering a new driver, with the
'--headless=old'
option, tests pass again.I'm assuming this means this is not a Capybara problem, but I'm hoping you could:
Thanks, I appreciate it.