Before I open an issue on this, I thought I’d get some additional feedback here. Here’s what happened:
I created a group gitlab runner that uses a Docker executor. After that, a number pipelines that were already using shell runners started breaking because they switched over to using the new group runner. This is not what I intended. It seems to me that the existing pipelines (which had their own specific runners) should not have automatically switched to the group runner. I fixed this by disabling group runners for those projects. It also seems to me that a project’s use of group or shared runners should be opt-in, not opt-out. Thoughts?