21.08 Upgrade - Divide By Zero Error

[paul56]

Hi all,

Just upgraded from SCALE 21.06 to 21.08 via the upgrade button in the GUI. Upon reload I got the attached red screen of death.
I can't do much to get any further logs as it won't boot now but hopefully this error will mean something to someone. I will reinstall from fresh. P.S Noy sure why the forim keeps turning the picture upside down lol

ra

 

[paul56]

OK I've now re-installed 21.08 from USB stick. It detected the old version and asked if I wanted to upgrade so I tried that first, but it failed. So I then chose fresh install and all was fine. I then uploaded the configuration backup file I made prior to the upgrade and all restored including my pools/shares and SCALE apps. So no big deal just a bit of a pain, but hope this error message can help someone for future upgrades.

 

[paul56]

Maybe spoken too soon. Now I'm on 21.08 if I shutdown/reboot the system, upon boot I am getting the same Divide By Zero error, and have to reinstall from USB and then restore settings to get it back online.
Everything is fine while running but now I am unable to reboot without a reinstall so not ideal.
I'm happy to log a bug report, but I only have the screenshot attached as to get it back up and running so don't know how useful that is. I would give logs but not sure what would be needed, and can only get them prior to reboot/red screen anyway. Can someone advise what to provide in a bug report?

 

[Kris Moore]

Please do open at ticket and include the screenshot, along with any other relevant system information you can think to provide. We'll take a look and see if its something we can fix.

 

[paul56]

OK thanks Kris,
I have now filed the bug report: https://jira.ixsystems.com/browse/NAS-112151

 

[paul56]

Hi Kris,

I did log the Jira ticket but it was closed before I could provide all the extra diagnostic information as Ryan felt it was an issue with the server's firmware rather than with Truenas/Linux. But in gathering the requested information I'm pretty sure I have disproven it being a server firmware issue as it happens after the Linux Grub menu not during the server initialisation.
I added the following to the Jira ticket but I'm not sure it will get read as it is closed. How do I get this reopened so the new information can be taken into account? I'm able to help troubleshoot for a short while longer, so I hope someone can take a look while I'm around.

"Hi Ryan,
As requested I've now found the time to gather more data you asked for.
I can now confirm that after the server boots, it shows the boot menu briefly, then for a fraction of a second the "Welcome to Grub" black screen shows. You have to be fast to catch it though as it's so quick. Normally after that it switches to the TrueNas blue bootloader screen for a second and then starts booting Linux. However, once the crash occurs after the very fast Grub screen shows it goes straight ot he red "Divide By Zero" screen instead of the usual Truenas blue bootloader screen.

This additional information would seem to prove it is not the server's firmware having this issue as suspected above as far as I can tell, but happy to be corrected if you think that is possible even after the Welcome to Grub screen.

A few more relevant pieces of information I have found:
1) This error only occurs if I select Reboot from the Truenas shutdown menu. If I select Shutdown and then turn the server back on again all is fine (I have tested this over half a dozen times now so confidence is high that is safe now).
2) This error also affects the new 21.08 BETA2 in the same way as previous release.


So to summarise, this issue is avoidable usually if I remember to never use the Reboot option. Unfortunately upgrading to higher releases from the GUI is out of the question as this automatically reboots afterwards instead of shutting down so causes the issue.
A fresh install from USB stick is fine as that gives the option to shutdown once completed instead of reboot.

N.B. Once the Divide By Zero error has occurred once, it never boots Truenas again even if power down/up completely, as Divide By Zero comes back every single time. Also once you have rebooted and got that error, the usb boot upgrade previous installation always fails so you can only install it fresh while choosing the format the boot device option.
I have not yet tried a full shutdown and then boot to new version on USB and then an upgrade via that method but I suspect it would work. I will try it when 21.10 is out.


Thanks!