It is not really clear what happens there… at least what I can say about this is that if a message like “Dependency failed for Local Filesystem” appear, then the device names in /etc/fstab are not correct. You should be able to correct them.
# manjaro-chroot -a
==> Detected systems:
--> 0) Windows
--> 1) Windows1
--> 2) Manjaro:linux:btrfs:UUID=abd07649-4af0-49fa-906b-c8cc116b98e6
--> 3) Windows2
==> Select system to mount [0-3] :
3
==> ERROR: You can't mount 3!
[manjaro /]# cat /etc/fstab
# /etc/fstab: static file system information.
#
# Use 'blkid' to print the universally unique identifier for a device; this may
# be used with UUID= as a more robust way to name devices that works even if
# disks are added and removed. See fstab(5).
#
# <file system> <mount point> <type> <options> <dump> <pass>
UUID=901e9437-31d8-4a4f-ac41-d7e585b7cc79 /boot ext4 defaults,noatime 0 2
UUID=abd07649-4af0-49fa-906b-c8cc116b98e6 / btrfs subvol=@,defaults,noatime,space_cache 0 1
UUID=abd07649-4af0-49fa-906b-c8cc116b98e6 /home btrfs subvol=@home,defaults,noatime,space_cache 0 2
This is a partition layout that may lead to problems with manjaro. You can’t rollback securely this way !!!
Why ?
Your /boot is on another partition then your / . When you update your system, the kernel will be updated at the same time.
on / you will have all revisions of all the different modules for all kernel-versions you need to rollback
on your /boot every kernel-version will replace each other. you really only have the newest kernel-version there !
When you roll back then there is a kernel in /boot with his new initramdisk.
BUT
later in the boot-process this new kernel will not find its own modules. because you rolled back / into an older snapshot.
I don´t know if this is the problem you face at this time. But i got burned. So i learned: With manjaro you have frequent changing kernels. To do a save rollback you can’t have /boot as extra partition. This used to be a good idea years ago. but nowadays it is best to not have /boot as partition
Thanks for your advice. I wanted to try to solve the problem according to your methods according to the link, but I thought to solve it with megavolt and not run from advice to advice of different people. But he doesn’t seem to want to deal with my problem)
I do not understand everything that you are saying because of the translation. I would like to know if it is possible to restore my system?
No thats not the case… If I have no idea at this moment, then I don’t answer. And it is bad if only one person answers. More heads produce better results… (something like that). So @andreas85 explained it in an excellent way.
So on /boot there are initramfs. These initramfs are not rolled back and therefore it crash. I guess an easy fix could be:
You do all right, in only following one advice at a time! megavolt is very competent like a lot of members in this forum. I must be honest. you wrote the most of this in your first post. But i did not recognize some of it
So what do we know until now:
your /boot is an extra partition
you did a rollback with timeshift
the /boot is not corrupted
Lets assume that “Failed to start Load Kernel Modules” is because the Kernel-Modules on /lib/modules don’t match the kernel in /boot
There seem to be several ways to rescue your install:
try to make the corresponding kernel modules in btrfs available again
try to install new kernel, modules and initramfs (without updating the system)
try to undo the rollback (this will only work, if Timeshift did not delete the newest snapshot)
try to backup the data you need and to do a fresh install (this may be very time-comsuming)
maybe i can come up with others (this may take some time)
As i said, i got burned this way already.
But i was able to rescue my installation.
After this i did remove the /boot and included it into /
This is indeed the first(best) way to go. It may be additionally necessary to replace the kernel with mkwd-kernel so that kernel and /lib/modules/* have the same version.
I don’t think so… Let’s assume he have installed an update. It updates the kernel, creates a new initramfs image and updates grub.
Now he rolled it back. But since /boot is not on the same partition (and therefore not included in the snapshot), it is not rolled back and stays the same. Now the new images after update stay and everything else is rolled back… So if he install a new kernel with mhwd, then the initramfs image will be recreated anyway, but the other stay.
Oh man… that is getting complicated…
On BTRFS /boot should be included and not a separated partition. Less headache, when using snapshots. Fully agree here with @andreas85
I think in /lib/modules on the disk there may be an old version of the modules from the rollback. When mkinitcpio runs it will try to use these. but the kernel is an newer version. At the moment the kernel and the initramfs may be the same version.
But You may be right. And it does not harm to try this. There will be no damage
I don’t have /boot (but i did configure /boot so that btrfs does not use compression there )
i do use zstd with initramfs (to save space even in /boot because kernels actually do decompress even zstd)
The only thing btrfs can’t do at this time is to store the last boot-entry on btrfs (GRUB_DEFAULT=saved does not work with btrfs)