Ok, so here goes. I installed FB on an Atomic Pi SBC, and it has been working well for several months now, but recently, it doesn’t want to boot because of a problem with the BIOS. It seems that the AMIBios on this board has a known issue with trouble booting to preexisting filesystems on microsd card after the RTC has been reset, not sure why. To add to this problem, the Borg backup failed on my install, also not sure why, but I did file a bug report on the matter. So here is what I am thinking to do: Take the files and folders from the (working, non-booting) installation, and place them into a new install on new hardware (I have already ordered a raspberry pi 3 b+ to replace it). Can this be done? Would I be better served to “start over”, just ditching everything and starting again from scratch? I have my domain hosted on this FB install, security certs, pagekite, radicale, minidlna, syncthing, openvpn, mediawiki, and tt-rss. I have already spent quite a bit of time repairing and replacing when this happened several months ago, and would prefer not to have to go through all that PITA again.
Problem Description
I have an FB install that won’t boot because of a bad BIOS.
Steps to Reproduce
Power down & Reboot, not booting.
Expected Results
For the system to boot to my FB install and run normally.
Actual results
System repeatedly(always, every reboot, shutdown and power up cycle) goes into BIOS settings when it cannot boot.
Screenshot
No need, just a BIOS screen.
Information
FreedomBox version: 20.9, I think
Hardware: Atomic Pi
How did you install FreedomBox?: Installed Debian buster first, then followed instructions for unattended installation.
If you want to do a manual backup and restore, you have to simulate manually what FreedomBox does automatically.
You can look at the manifest.py file for each of the applications you have installed and copy over the files to the new installation. You also have to ensure that the file permissions are exactly the same.
You can probably ignore the config files if you didn’t make any changes to the config.
For Syncthing, data is not backed up. It is expected that the files will be synced from its peers.
For TT-RSS, a pgdump and restore is required if you want to get the full database back. Or if you don’t care about your read history, you can just import the feeds from an OPML file.
BTW, if you weren’t jumping architectures from amd64 to ARM, you could have just inserted the old disk into new machine (or image copy it). So, another way is to do this is to get the Atomic Pi disk working with a VM on your laptop/desktop. Then install the backup app (after fixing it, let us know) take a backup and restore it on the new setup.
Sunil, thanks for the reply, which leads me to another question: Can the .img file of the old install be run in a VM on the ARM machine? I did in fact, make an .img file of the old install, and if I could run it in a VM on the ARM platform, that would be very cool indeed. I suspect that it would take more resources than the Raspberry Pi model 3 B+ would be capable of, but I could certainly be wrong there, my thoughts are that it would be something that I could do with the RPi 4, with >1GB of onboard memory, once that platform is “Debian Ready”. Please let me know your thoughts on this. Also, just for clarity, am I to use the manifest.py files located in /usr/lib/python3/dist-packages/plinth/modules/* as the reference files that Joseph was talking about in the first reply? One more thing, I will give that procedure you were talking about a try, I just need to install the VM software on my workstation to make that happen, and thanks for the help!
I don’t have experience running amd64 VMs on ARM. I would not recommend it for production setup since the resources are constrained.
You should be able to create a VirtualBox machine on your Laptop/Desktop machine with .img file you have. See manual page for VirtualBox on how to forward ports etc. Once it is running, take a backup, and install properly on Raspberry Pi.
