[Unit] Description=Pushy liveness watchdog # systemd's own supervision only proves the PID exists. This probes the # diagnostic socket, which is answered from the runtime event loop, so a stalled # but still-running process is detected and restarted. [Service] Type=oneshot User=root ExecStart=/opt/pushy/current/deploy/systemd/health-watchdog.sh TimeoutStartSec=60 SyslogIdentifier=pushy-health-watchdog