Whenever a Linux system does not shut down cleanly (e.g., due to a power outage, a system crash, etc), it is almost always an event of interest to the system administrator. This simple script keeps track of such events by detecting the subsequent "dirty" reboot, and writing a system log entry about it.

A systemd service, started early in the boot process, writes a marker file (/var/lib/dirtyboot by default). When the service is stopped, during system shutdown, the marker file is removed.

Thus, if the marker file is already present when the service starts, it means the system wasn't shut down cleanly before rebooting. Whenever this condition occurs, the service writes a corresponding warning to the system log.

A Linux system based on systemd.

Run the command

journalctl -t dirtyboot

Put the systemd unit file in /etc/systemd/system, and the shell script to /usr/sbin/dirtyboot. Then enable the service using

systemctl enable dirtyboot

I had an unstable system which tended to hang after long periods (days) of being mostly idle. I used this simple tool to keep track of the number of hard resets I had to do to restore operation of a hung system. Since system log entries are timestamped, it also records the time interval between such hard resets.