[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Dangerous installation of bullseye: What shall i do next?



On 3/26/22 15:16, Datakanja de Bruyn wrote:
Just a bit of context:
I am old + handicapped + pretty much isolated, thus certainly not an expert.
But i am happily using debian stable (oldstable by now) since several
years. But since more and more software got outdated, i was interested
to move to bullseye.
In order to have 2 bootable instances (oldstable + stable), i installed
and tested refind, in ordeer to have some safety during the migration
period. I did test the whole setup, which went well. Then installed
bullseye (11.2), but after that, apparently nothing did work any longer.
It took me several days to find out, what destroyed my configuration:

1. bullseye installer had installed grub2 over refind, thereby killing
my setup made for safety purpose.
2. Also, when i tried to boot into the oldstable by hand, it failed to
come up due to some error in fstab, bcoz the installer, while
reformatting the free partition, assigned a new partuuid to it, which no
longer corresponded to the entry in fstab.
3. Furthermore, it installed a grub2 version, that is buggy and which
cannot boot the bootentries, i was used to resort to in case of trouble
(a.k.a. booting straight from an ISO image as an emergency system. The
version installed was known to fail to boot on my kind of hardwae since
several years, and i assumed (my mistake), that a stable debian would
have been fixing the issue by now. (I refer to the links at the bottom).

Ok. After days in panic, i was able to straigthen out my old system and
get it to boot again. But since then, i am totally undecided (and a bit
overwhelmed) with the options, i have to decide about now. What shall i
focus on next?

1. Try the whole process once again and manually downgrade grub2 in
order to have the ISO-boot at hand? (What risk would that involve?)
2. Report a bug (but honestly, i am not skilled enough to even determine
the package(s) causing the mess i encountered. I suspect at least one of
grub2 packages to be involved, but also the installer itself does a
pretty careless job IMHO. (I learned to create proper assertion checks
before shooting a working configuration to death.)
3. Continue to work with oldstable, which increasingly causes problems
due to the outdated software involved. That is, what i am using right now.

Or is there a better option? - Like maybe someone willing to assist in
the process or at least guiding me some steps further?
But i am scared to show the details of what i am doing, as i am a ZFS
user since many years, which is pretty much non-standard!

The bug, i mentioned seems to be related to grub2 2.04 and UEFI booting,
which is necessary on my machine:
https://superuser.com/questions/755641/grub2-boot-error-out-of-memory
https://bugzilla.redhat.com/show_bug.cgi?id=1838633
https://bugs.launchpad.net/ubuntu/+source/grub2/+bug/1851311
https://ubuntuforums.org/showthread.php?t=2430437

Any hint will be greatly appreciated.
DdB


I suggest that you get an entry-level server. Install a small SSD. Do a fresh install of your server OS of choice. Use the recommended filesystem -- e.g. ext4 on Debian. Do not create an unprivileged user account during installation; or create a generic account if required (e.g. "debian"). Enable SSH service. Add site users and groups with planned UID's and GID's.


Get an external HDD and take a raw binary image of the OS drive. Create a checksum file for the image file. Repeat periodically and as needed.


Install two large HDD's (avoid SMR). Create a ZFS mirror. Install zfs-auto-snapshot and create cron job. Consider enabling default compression (lz4). Avoid deduplication. Consider adding a fast SSD as a cache device.


Choose a networked version control system (I use CVS). Create a ZFS filesystem for the repository. Enable service and connect to repository. Install client. Check in your modified OS configuration files and any working files you maintain.


Create a ZFS filesystem for Samba data. Enable Samba service and share filesystem. Avoid NFS. Collect and organize all of your data.


Get two or more large external HDD's and implement backups in rotation.


Archive important images and/or backups periodically and as needed.


Finally, remove the OS drive in your existing computer. Install a blank SSD. Do a fresh install of your desktop OS of choice. If you want more than one OS, repeat with another blank SSD; e.g. avoid multi-boot. Install SSH client. Install version control system client. Create users, groups, UID's, and GID's to match server. Mount the Samba data share in a local directory. Add computer to image, backup, and archive procedures.


David


p.s. I prefer FreeBSD for my SOHO servers, in part because of available documentation:

https://www.pearson.com/us/higher-education/program/Mc-Kusick-Design-and-Implementation-of-the-Free-BSD-Operating-System-The-2nd-Edition/PGM224032.html

https://mwl.io/nonfiction/os


Reply to: