I have a C++/Qt project, and I have my own machine configured as a runner. In the past builds were working just fine, but for some time now, every single build fails at the step where it runs an automated test. See for example https://gitlab.com/adriyetichaves/chopshop/builds/939566
If I log in locally as the gitlab-runner user and execute the same command on the corresponding folder indicated in the link above, it works like a charm. It only fails when built remotely.
Any ideas of what the issue may be, or how to further debug it?
Update: I ended up installing a local Jenkins server, and the same happens there. I still welcome any feedback, but whatever it is it seems to be specific to my system.
Update: I though it was something to do with the CI system, but it turns out that, whatever the issue is, it was introduced in a specific commit: https://gitlab.com/adriyetichaves/chopshop/commit/69df36ad9075714de180faf146cdd61099979ea2
Solution: The issue was that I was linking my tests to libraries that require a GUI. When logging in as the gitlab-runner user, I could run the tests because I was logging in from a terminal emulator. But when the tests where run from a user connected through an actual terminal with no windowing system on, tests would crash just by being linked to Qt GUI.
