Chris Vogel's avatar
Chris Vogel
@me@chrichri.ween.de

Since I started to use the kernel instead of an legacy kernel on my I had a few crashes. Most of them seemed to have been related to , because they stopped after I exchanged the sata pcie with another one of the same type.

I had another afterwards and decided I should look for a to reboot the system in case of trouble.

After reading a bit about watchdogd the most simple solution I found is:


root@TEST:~# cat /etc/cron.d/watchdog 
@reboot   root wdctl -s 180 
* * * * * root echo "1" > /dev/watchdog

I'm testing it on a non-productive board and it seems to be good. It works for a forced oops echo c > /dev/sysrq-trigger and if I stop cron.

But it doesn't work in the state after a simple halt: the system tries to start and hangs after showing the first line of u-boot output.

Chris Vogel's avatar
Chris Vogel
@me@chrichri.ween.de

in reply to this object

sigmaris on the irc channel seems to have an explanation for the unexpected behaviour of the watchdog after a halt: This commits fix is not called on halt and for that reason the restart might fail...