Specifically, there was a chance that `patronictl reinit --force` was overwritten by recover and we end up in a situation when Patroni was trying to start the postgres while basebackup still running.