automatic restart of some failed CI jobs
Some CI errors are spurious, I'm thinking of things like docker daemon errors and "unable to resolve host gitlab.torproject.org", but there might be other recurring errors.
It would be nice if the bot could automatically restart the corresponding pipelines.
Docker issues fall under the failure_reason runner_system_failure
so it shouldn't be hard to detect them, "unable to resolve host" when cloning are considered stript_failure
so catching them would require parsing job logs.