Currently I'm thinking about how furious the developers get any time Jenkins has any kind of hiccough, even if the solution is just "re-run the workflow" - and that's just network timeouts! I don't want to imagine the tickets if the CI system started spitting out hallucinations...