Bug#974838: grub: GRUB as of 2020-November.15: Failure to boot multiple systems - seemingly random

[waxhead]

Source: grub
Severity: normal
X-Debbugs-Cc: waxhead at dirtcellar.net

Dear Maintainer,

First I hope I am reporitng this on the correct package. Apologize if not.

I have various systems all running BTRFS as the rootfs. BTRFS is installed on partitions that are offset of the first sector=2048
The systems I use range from 8 to 20 drives. BTRFS is set up with data=raid1 , and metadata =raid1c4
After updates which update the grub package GRUB fails very often on my systems.
I can't remember all the error codes I have had,  which makes this a rather poor bug report, but the latest one (today) was:

error: symbol 'grub_calloc' not found.
grub rescue>

I also know that I have had 'embedding not possible', 'error <num>' and various other errors that did not even give a rescue prompt.
Sure , I realize that GRUB is free for me to use and I am thankful for that , but it is quite frustrating to have GRUB seemingly randomly fail on not
only one, but on sereval independent systems after an update. And it is not even the same update that breaks all systems.
Even after a reboot and a successfull boot, a new reboot (without any update or changes) renders GRUB unbootable (sorry , can't remember error codes).
It is not all bad , since I can get it working again relatively reasy - but it is an extra hassle that might take an hour or more depending on how
many reboots are neccesary to get things working.

When I have a working GRUB that boots my system , I expect that an update would not change that to an unbootable system without
giving me an error on update-grub.

I realize that this bugreport is not quite good, and I admith to being in an "AAARGH!" mode when writing this , but the last year or so I have had
nothing but random issues with grub. Yes, I run Debian testing, but GRUB used to worok perfectly all the time. It does not anymore.

I whish I could uninstall grub from ALL drives except one to try to debug , but apparently that is not possible without dd'ing over the bootblock?

   * What led up to the situation?

I was just using GRUB normally, e.g. with that I mean it was installed and I just did a regular update in Debian.
that is aptitude, update packages, mark upgradeable, look that nothing breaks, upgrade, and at some point later reboot.

   * What exactly did you do (or not do) that was effective (or
     ineffective)?

I had to boot a live CD. chroot to my rootfs,
update-grub                : rebooted, ineffective
install-grub               : rebooted, ineffective
update-initramfs -k all -u
and
update-grub                : rebooted, effective

   * What was the outcome of this action?

I was able to restore grub in such a way that it booted my system

   * What outcome did you expect instead?

That GRUB should not render my system unbootable


-- System Information:
Debian Release: bullseye/sid
  APT prefers testing
  APT policy: (500, 'testing')
Architecture: amd64 (x86_64)
Foreign Architectures: i386

Kernel: Linux 5.9.0-1-amd64 (SMP w/2 CPU threads)
Kernel taint flags: TAINT_PROPRIETARY_MODULE, TAINT_OOT_MODULE, TAINT_UNSIGNED_MODULE
Locale: LANG=en_US.UTF-8, LC_CTYPE=en_US.UTF-8 (charmap=UTF-8), LANGUAGE=en_US:en
Shell: /bin/sh linked to /usr/bin/dash
Init: systemd (via /run/systemd/system)
LSM: AppArmor: enabled