If my application stops, the Raspberry pi4,5 should reboot so I want to have my application to ping/poke the built-in hardware watchdog timer (BCM2835-wdt).
Because I want my program to be responsible for pinging/poking the watchdog by writing to /dev/watchdog, I have disabled both the watchdog.service and wd_keepalive.service like this:Pinging the watchdog could the be a command similar to this:But the system does not reboot when this pinging stops.
The command wdctl gives:
This shows always timeleft: 15 or 14 seconds indicating that some process is pinging the /dev/watchdog0 every 1 second. Confirmed by the line: "KEEPALIVEPING Keep alive ping reply 1 0"
This explains why ceasing to does not trigger a reboot.
But the question is which (zombie) process is triggering the watchdog then if it is not watchdog.service or wd_keepalive.service ? It is not my program. There are no services which are obvious candidates.
As a final verification - this python code will trigger a reboot after 15 second:The difference to the sudo sh -c "echo 'V' > /dev/watchdog" is that the python code locks the file /dev/watchdog preventing the zombie from pinging the watchdog.
So it all adds up except - which process is running in the background pinging the watchdog?
I am unsure of the difference between /dev/watchdog and /dev/watchdog0 - so far I have not seen any difference.
Comments are welcomed.
Because I want my program to be responsible for pinging/poking the watchdog by writing to /dev/watchdog, I have disabled both the watchdog.service and wd_keepalive.service like this:
Code:
sudo systemctl disable watchdog.servicesudo systemctl disable wd_keepalive.serviceCode:
sudo sh -c "echo 'V' > /dev/watchdog"The command wdctl gives:
Code:
pi@pi119:~ $ sudo wdctlDevice: /dev/watchdog0Identity: Broadcom BCM2835 Watchdog timer [version 0]Timeout: 15 secondsTimeleft: 14 secondsPre-timeout: 0 secondsFLAG DESCRIPTION STATUS BOOT-STATUSKEEPALIVEPING Keep alive ping reply 1 0MAGICCLOSE Supports magic close char 0 0SETTIMEOUT Set timeout (in seconds) 0 0This explains why ceasing to
Code:
sudo sh -c "echo 'V' > /dev/watchdog" But the question is which (zombie) process is triggering the watchdog then if it is not watchdog.service or wd_keepalive.service ? It is not my program. There are no services which are obvious candidates.
As a final verification - this python code will trigger a reboot after 15 second:
Code:
import time# Open the watchdog devicewith open("/dev/watchdog", 'w') as watchdog: print("Watchdog activated. The system will reboot in 15 seconds if no ping is sent.") time.sleep(30) # Wait for 30 seconds, longer than the timeout of 15 secondsSo it all adds up except - which process is running in the background pinging the watchdog?
I am unsure of the difference between /dev/watchdog and /dev/watchdog0 - so far I have not seen any difference.
Comments are welcomed.
Code:
pi@pi119:~ $ uname -aLinux pi119 6.6.51+rpt-rpi-v8 #1 SMP PREEMPT Debian 1:6.6.51-1+rpt3 (2024-10-08) aarch64 GNU/LinuxStatistics: Posted by h.larsen — Sat Oct 12, 2024 8:37 pm — Replies 5 — Views 79