Default GRUB boot entry not working.

Debian_SuperUser

Active Member
Joined
Mar 18, 2024
Messages
164
Reaction score
41
Credits
1,999
I installed Fedora 44 Sway Atomic and on the first boot itself it failed to boot. I found out that I have to boot from $root/boot/ostree and not $root/ostree. I was configuring my partitions and mount points manually. Did I mess up there? It seems that I can't permanently change the boot config as it is overwritten on every ostree deployment, so how should I fix it?

btw how are u supposed to install a C compiler without needing to reboot your system, use the potentially awkward --apply-live option, or use a toolbox which is more awkward and isn't good for scripts and AI agents?
 


You need to give more & detailed info of what you're doing.
I just installed Fedora. That's it. I configured my partitions and mount points manually during setup. Now it doesn't boot with the default settings using GRUB.

This for instance it's hard to tell what's your question, you install compiler with sudo apt install build-essential
I clearly listed Fedora Atomic not Debian.
 
I clearly listed Fedora Atomic not Debian.

While I didn't, likely due to how I read web pages, I could see where someone would read your post quickly and get confused because of your username.
 
During your manual partitioning are you sure that you allocated enough for the /boot, /root, swap, /var/home, /opt, /tmp....etc partitions?
Does your efi boot partition have the boot flag?

If Grub successfully installed to your hard drive then your fresh install should just boot to the log in.
Ostree should have updated grub.cfg to point to the current image.


Users have reported that they couldn't install Fedora Atomic Desktop on an EFI system.

Can you boot into the Live session of Fedora that you used to install it and run this and post back?
Code:
sudo fdisk -l
 

I just installed Fedora. That's it. I configured my partitions and mount points manually during setup. Now it doesn't boot with the default settings using GRUB.
It's impossible to determine cause of this without analyzing logs and conf files which you didn't provide. All we know is that you installed immutable (atomic) variant of Fedora.

I suggest to reinstall and use automatic partitioning, unless you are prepared to get technical, report bugs to Red Hat, and determine cause of this.
 
I was configuring my partitions and mount points manually. Did I mess up there? It seems that I can't permanently change the boot config as it is overwritten on every ostree deployment, so how should I fix it?
Fedora Atomic doesn't support custom partitioning.
But there's documentation if you want to try it, where it mentions again that manual partitioning is not supported.

btw how are u supposed to install a C compiler without needing to reboot your system, use the potentially awkward --apply-live option, or use a toolbox which is more awkward and isn't good for scripts and AI agents?
Fedora Atomic is a container based distribution and so is the workflow, so either toolbox or distrobox and if you have to layer something you will need to reboot unless you use --app-live.
 
It's impossible to determine cause of this without analyzing logs and conf files which you didn't provide. All we know is that you installed immutable (atomic) variant of Fedora.

I suggest to reinstall and use automatic partitioning, unless you are prepared to get technical, report bugs to Red Hat, and determine cause of this.
Agreed.
Allowing the partition manager to handle the partitioning normally goes well.

Manual partitioning takes time to master.
 
tl;dr (and read 5th paragraph)-
Code:
NAME   MAJ:MIN RM   SIZE RO TYPE MOUNTPOINTS
sda      8:0    0 238.5G  0 disk
├─sda1   8:1    0   128M  0 part
├─sda2   8:2    0     1G  0 part
├─sda3   8:3    0  80.7G  0 part
├─sda4   8:4    0  39.9G  0 part
├─sda5   8:5    0    16M  0 part
├─sda6   8:6    0  57.3G  0 part
├─sda7   8:7    0  58.8G  0 part /var
│                                /boot
│                                /sysroot/ostree/deploy/fedora/var
│                                /sysroot
│                                /etc
└─sda8   8:8    0 613.2M  0 part /sysroot/boot/efi
                                 /boot/efi
zram0  251:0    0   7.6G  0 disk [SWAP]

I have 2 EFI partitions. sda1 and sda8.
During your manual partitioning are you sure that you allocated enough for the /boot, /root, swap, /var/home, /opt, /tmp....etc partitions?
I didn't setup these many. Only /, /boot, and /boot/efi.
Does your efi boot partition have the boot flag?
I don't know what that means. But from what I can infer, the problem probably seems to be the mount configuration which lead to a wrong GRUB config. I could be wrong.
Users have reported that they couldn't install Fedora Atomic Desktop on an EFI system.
That isn't helpful. What am I supposed to do, boot from legacy BIOS? I don't have that. EFI installation should obviously work.
Can you boot into the Live session of Fedora that you used to install it and run this and post back?
To clarify again, I can boot into the Fedora installation. I am writing this from it. In the GRUB boot menu, I just have to edit $root/ostree/... to $root/boot/ostree/. The ostree directory is under /boot which it doesn't want. But it seems that I also have a /ostree anyway? Seems like only the kernel and initramfs are stored in /boot/ostree while /ostree contains the whole deployment and states. Oh, is /boot supposed to be under /ostree/boot? Because also under /ostree/ I have boot.1/ and boot.1.0/, clearly trying to not collide with "boot".

automatic partitioning
I had partitions on my drive. I am triple booting. I had to choose the manual partitioning method.
Fedora Atomic doesn't support custom partitioning.
??
Fedora Atomic is a container based distribution and so is the workflow, so either toolbox or distrobox and if you have to layer something you will need to reboot unless you use --app-live.
Yeah I know, and it is tiring and very inefficient to reboot. I use --apply-live but I think it is not a very intended way. Maybe containerization is the intended way. Even on Android I can install gcc in Termux and that is containerization. And why do I think layering is also not really that intended? That's why at least for GUI programs I am using Flatapk and oh my how much space it takes. At least though, I get containerization and bloat management which I really like. One delete operation and nothing remains behind. But besides bloat management, I don't think there is much of a reason to use Flatpak. I am using Fedora in the first place because it has a very stable ABI and package compatibility. All packages are supposed to work together. And Flatpak subsystem itself has some overhead. So I might not end up using Flatpak. Or at least for the programs I will have installed most of the time. And I guess I could manually containerize which I haven't tried.

If anyone remembers I said I wanted to try out Fedora Atomic for the delta updates. It still downloads more data than I expected but still significantly lower than Arch Linux. I just downloaded like a 500 MB update (or it was the extracted size, I don't remember). And I think before I had downloaded like a gigabyte of updates. Fedora updates much faster than I expected. Compared that on Arch, I could have to unpack almost 10 GB of data frequently. Ironically, I find that Windows's updates are more size efficient than Fedora Atomic. I honestly don't remember what used to happen on Ubuntu, because that seems to be the most efficient, or it looks to be at least. Maybe it used to update in background without me noticing. Or it updated much slower than Fedora Atomic. OSTree also has its own overhead.
 
Fedora Atomic doesn't support custom partitioning.
But there's

documentation if you want to try it, where it mentions again that manual partitioning is not supported.
Sorry I worded it wrong, must be the heat and the lack of sleep. I meant to say something else, but in short as stated on the two pages linked it's not fully functional and has it's limitations which you will might be running into because you are also dual-booting.

Yeah I know, and it is tiring and very inefficient to reboot. I use --apply-live but I think it is not a very intended way. Maybe containerization is the intended way. Even on Android I can install gcc in Termux and that is containerization. And why do I think layering is also not really that intended?
It's not a mortal sin to layer something if you really need it that way, but from my time using Silverblue I've always seen people say use containerized as much as possible whether that be toolbox, distrobox or flatpak and if you have no other choice just layer it.

Btw I just remembered another possible solution for you. It's also possible to use "brew" on Linux.
Then you can install gcc without having to reboot.
Code:
brew install gcc

The other option is the create your own image based on Silverblue and then just add everything in the image you need so that you don't need to layer anything. The Ublue project has a nice template for this.
 
You need to give more & detailed info of what you're doing.


This for instance it's hard to tell what's your question, you install compiler with sudo apt install build-essential
I clearly listed Fedora Atomic not Debian.
While I didn't, likely due to how I read web pages, I could see where someone would read your post quickly and get confused because of your username.
Maybe change your username to just "SuperUser", as I've made that mistake in other topics of yours before ;)
 


Follow Linux.org

Staff online


Latest posts

Top