People,
A while ago I was in a hurry to start using F40 (Sway) and installed Rawhide from a live usb to get started.
Everything went pretty well and I had minimal problems up until after f40 was release. However now, when I tried to do a:
dnf system upgrade download --releasever=40
I get a lot of errors - trying to do "dnf --erase" the problem rpms ends up making the problem worse.
So, if I try to reinstall from the current f40 live usb - can I do that without touching the backup subvolume? ie:
/dev/nvme0n1p3 btrfs 1,951,850,496 464,659,184 1,486,438,224 24% / /dev/nvme0n1p3 btrfs 1,951,850,496 464,659,184 1,486,438,224 24% /backup
For decades I have just been using a separate partition for /home with ext[2|3|4] and this was conveniently not touched if I did a custom install to upgrade - I want to do something similar for this backup subvolume.
Thanks,
Phil.
On Mon, Jun 3, 2024 at 4:43 PM Philip Rhoades via users < users@lists.fedoraproject.org> wrote:
People,
A while ago I was in a hurry to start using F40 (Sway) and installed Rawhide from a live usb to get started.
Everything went pretty well and I had minimal problems up until after f40 was release. However now, when I tried to do a:
dnf system upgrade download --releasever=40
I get a lot of errors - trying to do "dnf --erase" the problem rpms ends up making the problem worse.
You need to tell us what errors you encounter. Are they filesystem problems, package conflicts, etc.?
So, if I try to reinstall from the current f40 live usb - can I do that without touching the backup subvolume? ie:
/dev/nvme0n1p3 btrfs 1,951,850,496 464,659,184 1,486,438,224 24% / /dev/nvme0n1p3 btrfs 1,951,850,496 464,659,184 1,486,438,224 24% /backup
Unlike earlier filesystems, btrfs doesn't silently ignore bitrot. Corrupt
btrfs filesystems are often a result of hardware faults.
For decades I have just been using a separate partition for /home with
ext[2|3|4] and this was conveniently not touched if I did a custom install to upgrade - I want to do something similar for this backup subvolume.
I would never consider a btrfs subvolume a suitable location for a backup. The btrfs snapshots provide a way to go back in time after data are lost due to a software glitch or command-line typo. Using a subvolume, when the drive fails, both the original and backups may be lost. If a drive is close to failure, a linux install often pushes it over the edge (due to amount of activity), so I like to make sure backups are good before an upgrade or install.
George,
On 2024-06-04 09:37, George N. White III wrote:
On Mon, Jun 3, 2024 at 4:43 PM Philip Rhoades via users users@lists.fedoraproject.org wrote:
People,
A while ago I was in a hurry to start using F40 (Sway) and installed
Rawhide from a live usb to get started.
Everything went pretty well and I had minimal problems up until after f40 was release. However now, when I tried to do a:
dnf system upgrade download --releasever=40
I get a lot of errors - trying to do "dnf --erase" the problem rpms ends up making the problem worse.
You need to tell us what errors you encounter. Are they filesystem problems, package conflicts, etc.?
I think it will be too much to sort out the errors . .
So, if I try to reinstall from the current f40 live usb - can I do that without touching the backup subvolume? ie:
/dev/nvme0n1p3 btrfs 1,951,850,496 464,659,184 1,486,438,224 24% / /dev/nvme0n1p3 btrfs 1,951,850,496 464,659,184 1,486,438,224 24% /backup
Unlike earlier filesystems, btrfs doesn't silently ignore bitrot. Corrupt btrfs filesystems are often a result of hardware faults.
Hmm . . not good . . this is a new card at least . .
For decades I have just been using a separate partition for /home with ext[2|3|4] and this was conveniently not touched if I did a custom install to upgrade - I want to do something similar for this backup subvolume.
I would never consider a btrfs subvolume a suitable location for a backup. The btrfs snapshots provide a way to go back in time after data are lost due to a software glitch or command-line typo. Using a subvolume, when the drive fails, both the original and backups may be lost. If a drive is close to failure, a linux install often pushes it over the edge (due to amount of activity), so I like to make sure backups are good before an upgrade or install.
OK - things to think about (thanks for that!) - but my immediate question has not been answered - is it actually possible to re-install from a Live-USB while preserving what is in the /backup subvolume - like I used to do with the ext4 /home partition or will the zubvolume get zapped?
Thanks,
Phil.
On Mon, Jun 3, 2024, at 3:42 PM, Philip Rhoades via users wrote:
So, if I try to reinstall from the current f40 live usb - can I do that without touching the backup subvolume? ie:
/dev/nvme0n1p3 btrfs 1,951,850,496 464,659,184 1,486,438,224 24% / /dev/nvme0n1p3 btrfs 1,951,850,496 464,659,184 1,486,438,224 24% /backup
Yes. You need to do a custom installation. This is not official documentation, it's intended for Fedora QA purposes, it has macros that show Rawhide versions because - well it's for testing :D so you just stick with the version you have, and adapt it for your usecase.
https://fedoraproject.org/wiki/QA:Testcase_partitioning_custom_btrfs_preserv...
It sounds like you would just create a new root mountpoint, thus a new root subvolume, and not create a /home mountpoint. You can optionally click on the existing backup subvolume and then assign it a mountpoint /backup to have the installer do this for you and add it to fstab. Or you can do it post-install yourself.
Note that unless you explicitly delete the current (broken?) root in the intstaller, it will live on. And you also won't be able to navigate to it from / - you can either mount the top level of the file system and delete it, or you can delete directly by subvolume ID without mounting the top level, see man btrfs subvolume
If you have questions find me (cmurf) on matrix https://matrix.to/#/#fedora:fedoraproject.org