Unattended reboot of industrial PCs — watchdog configuration

The most troublesome situation at unmanned operation sites is the state where "the screen is alive but the program is frozen." In this case, simple power recovery settings will not solve it.

a two-stage configuration is required

• Software health check — verify application responsiveness, restart the process on failure
• Hardware watchdog — forced reset when the OS itself hangs

Linux watchdog configuration

# /etc/systemd/system.conf
RuntimeWatchdogSec=30s
RebootWatchdogSec=10min

# Check the hardware watchdog device
ls /dev/watchdog*

Application health check

The simplest and most effective method is to have an HTTP health endpoint and have systemd restart the service if there is no response.

[Service]
Restart=always
RestartSec=5
WatchdogSec=60
# The app must send sd_notify(WATCHDOG=1)

If reboots are frequent, you need to look at the cause. A watchdog is a last resort, and you must be able to leave logs to trace why it stopped.

Checklist

• BIOS AC Power Loss → Power On setting
• Auto-login at boot + automatic app startup
• Record reboot history (so anyone can see it)
• Secure a remote access method — it must survive a reboot