When queuing unit jobs, we should rather replace existing units than
fail. This is especially important when we queued a user-shutdown and a
new login is encountered. In this case, we better raplce the shutdown
jobs. elogind takes care of everything else.