[systemd-devel] Seeking Advice on debugging failure to reboot

[Lennart Poettering]

On Thu, 06.10.11 15:56, Barry Scott (barry.scott at onelan.co.uk) wrote:

> We can reproducably get an F15 system in a state that it fails to
> complete a reboot.
> 
> Becuase all the logging, network and getty's has been stopped all
> we can see is the kernel messages on the console.
> 
> What we need to know is why systemd is not completing the reboot.
> 
> What advice do you have on how to debug this situation please?

My usual recommendation is to boot with "systemd.log_level=debug
systemd.log_target=kmsg" on the kernel cmdline.

Then, place a script like the following in /lib/systemd/system-shutdown
and mark it +x:

#!/bin/bash
mount / -o rw,remount
dmesg > /dmesg.shutdown
mount / -o ro,remount
exit 0

Then, when you shutdown you'll have the full kernel log buffer in
/dmesg.shutdown, right before the machine powers off/reboots. If you get
that far however, it's usually a good indication that the kernel is
borked, rather than systemd.

Lennart

-- 
Lennart Poettering - Red Hat, Inc.