5 days ago, suddenly the e2e tests started to fail with The automation client disconnected. Cannot continue running tests. which states that google chrome crashed or it’s connection to cypress got lost.
We searched for solutions, mostly they were connected to docker and /dev/shm size, disabling shm size when starting google chrome. As far as I could see, shm in travis is set to 4G anyway, so disabling shm for chrome with
--disable-dev-shm-usage flag didn’t help either.
One of the failing jobs: https://travis-ci.org/SAP/luigi/jobs/648326477
All current PRs, same error since 5 days.
Not sure if it is connected with some internal travis changes? Also I am not sure how to debug it. Maybe someone has an idea what the rootcause could be …