Installation failed due to a GRUB installation issue

Wouldn’t it be better to use --removable in this case? Using --no-nvram will leave you with an unbootable system I think. --removable will write the default UEFI entry so the system can boot.

--no-nvram is more for subsequent installs where there is already a valid entry in nvram.

could be i am not sure here at all… as it looks like NVRAM is not writable at all … and it only boot on

No BootOrder is set; firmware will attempt recovery

all together.

but yes --removable is the better option to use for newly install a system …

I do still think it is the nvram that needs to get rewritten/reset by try to reflash the firmware/bios.

The other question would be if it shows any boot entries when booting or inside the firmware interface ?
I saw people where it show the same output for efibootmgr but there were entries in the nvram that you can choose on boot…

On the other NUC where I managed to install an encrypted EOS Qtile along with a not encrypted Lubuntu (for guests) I have :

[crocefisso@crocefisso-pc ~]$ sudo efibootmgr -v -v
[sudo] Mot de passe de crocefisso : 
Could not read variable 'BootNext': No such file or directory
error trace:
 efivarfs.c:268 efivarfs_get_variable(): open(/sys/firmware/efi/efivars/BootNext-8be4df61-93ca-11d2-aa0d-00e098032b8c): No such file or directory
 lib.c:140 efi_get_variable(): ops->get_variable failed: No such file or directory
Could not read variable 'BootCurrent': No such file or directory
error trace:
 efivarfs.c:268 efivarfs_get_variable(): open(/sys/firmware/efi/efivars/BootCurrent-8be4df61-93ca-11d2-aa0d-00e098032b8c): No such file or directory
 lib.c:140 efi_get_variable(): ops->get_variable failed: No such file or directory
Could not read variable 'Timeout': No such file or directory
error trace:
 efivarfs.c:268 efivarfs_get_variable(): open(/sys/firmware/efi/efivars/Timeout-8be4df61-93ca-11d2-aa0d-00e098032b8c): No such file or directory
 lib.c:140 efi_get_variable(): ops->get_variable failed: No such file or directory
Could not read variable 'BootOrder': No such file or directory
error trace:
 efivarfs.c:268 efivarfs_get_variable(): open(/sys/firmware/efi/efivars/BootOrder-8be4df61-93ca-11d2-aa0d-00e098032b8c): No such file or directory
 lib.c:140 efi_get_variable(): ops->get_variable failed: No such file or directory
 efibootmgr.c:372 read_order(): efi_get_variable failed: No such file or directory
No BootOrder is set; firmware will attempt recovery
Could not read variable 'MirrorCurrent': No such file or directory
error trace:
 efivarfs.c:268 efivarfs_get_variable(): open(/sys/firmware/efi/efivars/MirrorCurrent-7b9be2e0-e28a-4197-ad3e-32f062f9462c): No such file or directory
 lib.c:140 efi_get_variable(): ops->get_variable failed: No such file or directory
Could not read variable 'MirrorRequest': No such file or directory
error trace:
 efivarfs.c:268 efivarfs_get_variable(): open(/sys/firmware/efi/efivars/MirrorRequest-7b9be2e0-e28a-4197-ad3e-32f062f9462c): No such file or directory
 lib.c:140 efi_get_variable(): ops->get_variable failed: No such file or directory
[crocefisso@crocefisso-pc ~]$

However, I don’t think that the issue is coming from firmware, because I just tried to install EOS on my laptop (Acer Swift 3 SF314-54) and it failed with the same issue.

I managed to install an unencrypted EOS XFCE for guests. But when I try to install an encrypted EOS Gnome for myself, it fails.

Here is the log.

I think the issue comes from wanting an encrypted EOS on a partition and an unencrypted EOS on the same machine (so the guests can boot the machine without any passphrase).

It almost definitely is a firmware issue.

This is a totally different issue. This is a file conflict caused by not formatting the partition holding /boot

If I tick format on /boot partition, won’t it erase the kernels of EOS XFCE?

Yes. But if you are sharing a /boot between two distros the kernels and ucode would install on top of each other anyway. In short, you can’t do that.

Why are you trying to share a /boot?

1 Like

Because I want to have 2 EOS installed.

  • One for me, which is encrypted.
  • One for guests, that is not encrypted, so they can boot without any passphrase or password.

None of that requires a shared /boot

The unencrypted install doesn’t need a separate /boot at all.

/boot is where the kernels and initrams are installed. In your partition setup from above, there is nothing special about it.

1 Like

That doesn’t require any special configuration. It should “just work”.

It is failing because you are trying to share a /boot partition between the two. If you want a separate, unencrypted /boot in the encrypted installation to avoid the grub decryption prompt, that is an option. However, the unencrypted install doesn’t need that.

Keep in mind, in a UEFI installation, the EFI partition is where the bootloader is being launched from, not /boot. By default, we don’t use a separate /boot at all.

If for some reason, you want a separate /boot for both installations, that is fine as well. Just create two partitions, one for each installation.

1 Like

Sorry, I think my GNU/Linux knowledge is too limited to fully understand. I’m trying to install EOS Gnome on /dev/sda3 as you can see in this picutre.

If I don’t need to create another partition, what shall I do not to have the EOS Gnome installation fail?

If I need to create another partition to install EOS Gnome, could you provide me with details on this partition (size, flag, mount point, system file)?

Only use / and /boot/efi. Ignore /dev/sda2. Format /, don’t format /boot/efi if this is the second installation.

1 Like

It is indeed the second installation, the first one worked (EOS XFCE on /dev/sd4).

I followed your instructions and it failed. Here is the log.

It can’t write to the disk. Are you sure you selected to format /?

I didn’t. Shall I format it to ext4, luks or luks2?

If you want it encrypted, select ext4 and check the encryption check box.

Ok, shall I also format /dev/sda /boot/efi?

No.

The installation on my laptop ended successfully. However, booting didn’t work as expected:

  • luks passphrase is asked in qwerty when my keyboard is in azerty.
  • EOS XFCE kernels are showing up after luks passphrase instead of before it.
  • When I select an EOS XFCE kernel, the boot fails. So I can only boot on EOS Gnome kernels

I managed to install EOS properly on my laptop, having /boot

  • on a specific partition (/dev/sda2) for the encrypted EOS (/dev/sda3)
  • on the same partition as the clear EOS (/dev/sda4)

Regarding the initial issue I had with my NUC, I changed Calamares script according to you prescription (see picture), but it didn’t worked. Installation failed, here is the log.

This topic was automatically closed 2 days after the last reply. New replies are no longer allowed.