That will definitely make you not get hardware accelerated video Make sure to follow the 3 commands to install the custom repo, its certificate and higher priority: sudo wget http://apt.undo.it:7242/apt.undo.it.asc -O /etc/apt/trusted.gpg.d/apt.undo.it.asc . /etc/os-release && echo "deb http://apt.undo.it:7242 $VERSION_CODENAME main" | sudo tee /etc/apt/sources.list.d/apt.undo.it.list $ echo -e "Package: *\nPin: release o=apt.undo.it\nPin-Priority: 600" | sudo tee /etc/apt/preferences.d/apt-undo-it sudo apt update Then check that the new custom dpkgs are available for install/upgrade apt list ffmpeg-whatever Here's how I force install a specific version: sudo apt install packagename=nn.nn.n

Thanks, all's clear I might be such random person lol. However it would need either a lot of time or information delivered when it has been broken; open code is nice but there are too many kernel versions and distributions. That matter would need some deeper conversation. I think you was the person managing this area so maybe you know where to search the issue to correct it or you know the person who might know. I'm affraid I found the reason of long image generation time: my laptop fans are to silent today and it might be the reason...

Thinking about a better automated option to revert (e.g. just power off like @The Tall Man suggested) while still not touching uboot or even lower level layers the initrd might be an option based on the current setup. I assume we have very little probability for kernel+initrd load problems. Most issues will probably occur after the rootfs was mounted and handover. Integrating additional scripts inside the initrd could be used to auto revert the symlink in case of failure. Before we are going to reboot we place a status file inside the boot partitions boot directory that we just updated. For this first reboot we call this testboot1. When the new system boots up initrd script checks for this file and renames it to testboot2. If the system boots up correctly or can be accessed remotely /boot/testboot2 is deleted. But if the initrd script sees a testboot2 file during its boot process it knows something went wrong. It then changes the /boot symlink to the other boot folder deletes testboot2 and reboots instead of handing over to the rootfs. If the new system got stuck after it handed over to the rootfs the initrd script should have renamed the file to testboot2 already and a hard restart should revert. This hard reset could probably also be fully automated if the SBC has a hardware watchdog. If the watchdog module is integrated in the initrd it could set it to a reasonable value. The watchdog is then only disabled or changed to normal operation once the system is reliable working or if disabled via remote access. This concept would build on top of what exist. Any community member could choose to test the idea and implement management scripts. I guess everything can be done in bash/sh. It could also lead to a new armbian package: simpleAB which is not perfect but might be enough for many. It would not need many uboot or architectural changes. I guess it also works for all armbian supported devices? But as @Igor already said. One needs to invest.

"apt policy ffmpeg" FTW ;-) apt also has a command-line option to force the installation of a particular package version. To automate this, you can put a file under /etc/apt/preferences.d/

This is not recommended as its not controlled environment. It might work, but no warranties. HDMI output is broken in general (on this particular SoC) regardless of variant, desktop, cli, minimal. Its known bug, but we currently don't have anyone maintaining Allwinner A20 series so its up to random person to fix this. Code is open, so anyone can. Then it should be o.k. since HDMI output is not needed.