Porteus

We occasionally have trouble PXE booting the kiosk. It then gets stuck at a "kernel panic" page. We'd like to make the device reboot when this happens, instead of just sit there. This seems possible to do as described here:

https://www.techrepublic.com/blog/linux ... nel-panic/

But I can't find this directory on the image. I assume it's inside one of the laarge files there, that would then need to be unpacked, patched and re-packaged. Any pointers here would be most welcome.

-JM

Hello JM,

As per:
https://www.systutorials.com/how-to-mak ... nel-panic/

1) Please use this parameter when generating PXE Boot image in the kiosk wizard:
kernel_parameters=panic=10


2) Or edit /boot/pxelinux.cfg/default directly and add 'panic=10' parameter to relevant config line.

Confirm by running 'cat /proc/cmdline' over SSH on PXE booted kiosk that 'panic=10' is there.

Thanks

Thank you! We will investigate to see if this will solve our PXE boot issues.

-JM

I have set the kernel_parameters=panic=10 in the kiosk wizard and confirmed it is there with cat /proc/cmdline. Is there also a way to force a kernel panic from ssh?

On Ubuntu I have used this method when fiddeling with WatchDogs:

# echo "1" > /proc/sys/kernel/sysrq

# sudo echo c > /proc/sysrq-trigger

Is there a simular method that can be used in Porteus Kiosk?

Hello jerrymattias,

I'm not aware of any method which would allow to test kernel panics.

If you find it then please share as it may come useful for debugging.

Thank you.

I added this option in the kiosk wizard, as suggested:
However, it seems we can still run into situations during PXE boot where a player ends up with a Kernel Panic without an automatic re-boot after the specified time. When in the boot process are such options from the kiosk wizard applied? Would options from any sysctl.conf file be applied "earlier" in the boot process, or is there no difference? Any other thoughts on how we could avoid Kernel Panic during PXE boot, or how to ensure that a player is indeed always rebooted after such as mishap?

-JM

"When in the boot process are such options from the kiosk wizard applied?"
Immediately when the kernel image is extracted.

"Any other thoughts on how we could avoid Kernel Panic during PXE boot, or how to ensure that a player is indeed always rebooted after such as mishap?"
Perhaps the PXE server could SSH into each client mins after boot and toggle the power in case of a failure?
That would require some scripting on the server side and ability to cycle the power.

Or maybe you could find why the panics occur in the first place?
Please use: and link a photo of the kiosk screen with the error visible on it.

Thanks

"When in the boot process are such options from the kiosk wizard applied?"
Immediately when the kernel image is extracted.

Isn't it then strange that we still can get into a situation with a Kernel Panic message on screen that is NOT causing a reboot after the specified time?

Or maybe you could find why the panics occur in the first place?

Yes, always best of course. But we have not been able to reproduce this in house. It only happen at clients' site, and is rare there too. So hardest kind of problem to pin down and fix

-JM

"Isn't it then strange that we still can get into a situation with a Kernel Panic message on screen that is NOT causing a reboot after the specified time?"
Maybe there is another kernel option which must be enabled.
I actually doubt this as it would be enabled automatically, this is how kernel configurator works.

We have the panic timeout set to '0' (never reboot) by default: We need to prevent rebooting loops (it could affect the hardware in long run).

You still should be able to change default timeout value using our 'kernel_parameters='.


I'm afraid we wont be able to debug this problem properly until we find a way to manually trigger the panic.

Thanks