Occasionally, an install job queued after another install job is incorrectly read as having failed before it's even begun. Something seems wrong with the check for -- overt-failed: cask-install --.
In all likelihood, this applies to update and uninstall jobs as well.