We have recently changed our environment to use the standalone selenium headless firefox browsers in a docker container. We are allowing up to 4 machines to connect and create a webdriver in that container at a time. Previously very stable tests are now having a variety of odd issues - hangs, clicks not seeming to be caught, etc. The problems are intermittent. We are not using selenium hub - just the standalone selenium server. After searching and reading posts, I'm not seeing anyone else doing this - most seem to be using hub. I am wondering if what we are trying is even a supported use? It seems like as long as each machine connecting is getting it's own webdriver successfully, then the processes should be able to co-exist, but in practice, it's not working well....could we be overloading the container's resources? Could the clicks be getting lost between the processes? With a one to one container to firefox, I am unable to recreate the errors. Any suggestions appreciated.
Asked
Active
Viewed 457 times
0
-
Most probably you are observing the issues because of overloading the container resources. Are you using https://hub.docker.com/r/selenium/standalone-firefox ? Also try to set the screen resolution for your test and then observe the behavior. – Rahul L Dec 12 '19 at 04:07
-
We are using our own docker container to control the levels of browser and driver and selenium. Not sure what you mean about the screen resolution - it's a headless browser, so all we can do is check the screen captures. – Amy Reitmeyer Dec 12 '19 at 14:59
-
https://stackoverflow.com/a/43542208/5324105 Try setting --window-size – Rahul L Dec 16 '19 at 04:45