A Raspberry Pi user described using systemd timers to check always-on services such as DNS and recover them in stages. Instead of merely checking whether a process exists, the setup tests an actual DNS response and restarts the affected service first if failures recur. Full reboots are a later resort, with logs retained to trace causes. This is a report from an individual environment, not a recommendation for an indiscriminate restart loop or a guarantee of uninterrupted operation.