After multiple attempts, finally made my own stable image of Bookworm 12.4 for BeaglePlay. With xfce desktop even.
For now, I'm saving every build log, pass or fail, mostly on the sheer wtf of some of the failures.
Little known fact: my learning curve thrives on finding all the ways I can make something fail. That's basically the fastest and easiest way for me to learn something; break it. This is fun for me, so fuck yes those logs are amazing.
Even if I didn't cause it.
In logs for part one of initial build using image-creator, errors that I did not cause and vanished on their own:
1.) Pauses because the password of the default user of that build had expired. The build had not yet even gotten past the first script; there was no rootfs yet.
I still have no idea wtf happened there and it happened twice right after the user creation function.
2.) resolv.conf vanished right before the final package installation. No reason. No idea what was up with that.
3.) It didn't like the test function I added the first time. Second time it did. Just--what?
No problems after this, until we get to writing it to an sd card with baletcha and/or Raspberry Pi Imager. All bets were off.
Using the same final image:
Baletcha: I went through thirteen perfectly good cards; it failed every time at the end of validation at 98%. I do not know how many times I re-imaged each of those cards but it was a lot.
Raspberry Pi imager with the custom image option: failed all but the last time. Fine, whatever.
It worked perfectly when I booted it with #BeaglePlay
Due to sheer joy, I forked a copy of image-builder so as to play more thoroughly with the functions and also to work out how it works end to end. Also, to work out how to integrate my customizations.
So far, full success in adding:
1.) update configs of nanorc, ssh, samba
2.) create directories and mount points for remote shares
3.) download common files including my mount files for remote shares.and authorized_key file for ssh.
Sometimes-successful, sometimes not:
1.) add mount files to systemd, enable and start.
2.) Copy over all personal scripts from one of those shares.
3.) Add welcome script that displays at ssh login.
In progress:
1.) add my username and home directory
These I can't do until my user is already there.
1.) configure skel file with some more options including uncomment force-color-prompt=yes in .bashrc, add .bash_aliases, and an .ssh folder so key authenticaion is enabled at first boot.
2.) Compile some needed programs from git
What I really want to do now though:
1.) Create a custom file in /etc with basic build info.
2.) add copy of build.log in /etc so if while using it things do weird things I didn't expect, i'ts right htere for me to check.
3.) Add a metric ton of echo statements (log and debug) both as a practical way to map connections but also because I am super into logging and makes the build.log super super fun to read instead of just super fun.
4..) Add a metric ton of comments to the scripts because I love reading comments.
5.) Recompile the kernel because fuck yes that's how I have fun on Saturday nights.
Question that's bothering me: is there any way to get detailed information on why balena/pi imager failed? Or is there an imager II can use that will give me that information?
It's not just happening with my images; it's also happening with about a third of the canonical builds posted to the beagleboard forum for all the Beagle Boards. One thing I did notice: it seems to happen more often with an image over 10 GB and it's never happened with any image, mine or canonical, below 4 GB.
Seperis
LANdalf the Grey