I’m still not entirely clear on why I should reinstall the bootloader. The PC is actually booting the bootloader installed on the external hard drive. rEFInd shows the kernels I have installed, and I can even select and start one. However, when trying to boot a kernel, I encounter issues. Could you help clarify why the bootloader needs to be reinstalled if the system is able to show the kernels and allow selection?
Meaning what? As it is running from a removable device - it is sheer luck you didn’t run into any issues before now.
I understand this as: the installation on the removable device is working as expected.
While it is possible and with the increasingly better USB connections it may be getting attractive to use a second disk instead of dual booting - but the configuration mechanisms provided by various installers are - kind’a - expecting a disk configuration that does not change.
When it is suggested that you reinstall the boot loader it is because the installer is expected to fix the inconsistencies causing the current configuration to malfunction.
In this case you are assuming that because have not had any incidents previously and thus configuration is supported and updates are expected to not change that. That has never been the case and now the loader configuration is causing the loader to malfunction.
I have not been fiddling much with rEFInd, if I recall correct it can only boot from EFI. It can also boot kernel images if they are on readable file system - therefore rEFInd does not support booting from LUKS encrypted volumes.
You should verify that the $esp is indeed a an EFI partition with the correct partition type - in gdisk it is 0xEF00
- if it is not I don’t expect rEFInd to recognise the partition as the efi system partition.
The information given loosely explains how to install rEFInd to the $ESP
on a fixed disk rather than a USB device. It was purely informational, after seeing comments that rEFInd is installed to a USB device, and also;
With the last comment I am in agreement. The information provided seems inconsistent. Nonetheless, I’ll add more information that might be useful.

I have replaced GRUB with the rEFInd bootloader, and it was installed on the external hard drive. Nevertheless, the problem with mounting the EFI partition persists.
If you did indeed replace GRUB with rEFInd, that is another issue in itself. Plus, as far as I’m aware, booting from a USB device isn’t a supported scenario.
Usually rEFInd is fully capable of chainloading most other boot loaders in your system, including GRUB, so, the typical scenario is to allow rEFInd to do just that.
If you have actually removed/replaced GRUB with rEFInd, it cannot.

I suspect there might be an issue with the partitioning or the mount options for /boot/efi, as this partition seems to fail to mount.
What specifically makes you think that? I ask because it could just as easily be an issue with GRUB.

As you can see, /boot/efi is located on /dev/sdb1
For the sake of clarity, /boot/efi
is a mountpoint to the EFI System Partition (ESP) ($ESP/EFI
) and not the partition itself. Call it the ESP… because, that’s what it is.

You mentioned restoring the boot loader, which is a possibility, but since I’m using rEFInd instead of GRUB, I wasn’t sure whether this applies in the same way.
This only enhances the reasoning for giving the information you queried; whether you use it, or not, is immaterial.

BootCurrent: 0008 Timeout: 0 seconds BootOrder: 0007,0002,0001,0000,0008,0009,0003,0004 Boot0000* Windows Boot Manager HD(2,GPT,32c5a3c8-fbf0-4ee6-a298-957aaa8a5053,0xc8800,0x32000)/\EFI\Microsoft\Boot\bootmgfw.efi57494e444f5753000100000088000000780000004200430044004f0042004a004500430054003d007b00390064006500610038003600320063002d0035006300640064002d0034006500370030002d0061006300630031002d006600330032006200330034003400640034003700390035007d0000006f000100000010000000040000007fff0400 Boot0001* manjaro HD(1,GPT,92ee68f5-5982-4269-8f18-930cc57625d8,0x28,0x64000)/\EFI\manjaro\grubx64.efi Boot0002* rEFInd Boot Manager HD(1,GPT,92ee68f5-5982-4269-8f18-930cc57625d8,0x28,0x64000)/\EFI\refind\refind_x64.efi Boot0003* Onboard NIC(IPV4) PciRoot(0x0)/Pci(0x1d,0x0)/Pci(0x0,0x0)/MAC(a4bb6d53e03e,0)/IPv4(0.0.0.0,0,DHCP,0.0.0.0,0.0.0.0,0.0.0.0)0000424f Boot0004* Onboard NIC(IPV6) PciRoot(0x0)/Pci(0x1d,0x0)/Pci(0x0,0x0)/MAC(a4bb6d53e03e,0)/IPv6([::],0,Static,[::],[::],64)0000424f Boot0007* rEFInd HD(1,GPT,92ee68f5-5982-4269-8f18-930cc57625d8,0x28,0x64000)/\EFI\refind\refind.efi Boot0008* UEFI: SanDisk PciRoot(0x0)/Pci(0x14,0x0)/USB(1,0)/CDROM(1,0x743260,0x8000)0000424f Boot0009* UEFI: SanDisk PciRoot(0x0)/Pci(0x14,0x0)/USB(1,0)/HD(2,MBR,0x0,0x743260,0x2000)0000424f
You appear to have two entries for rEFInd, and neither appear to be on the $ESP
of a USB device, as you claim. Which of these instances is working?
Judging by the file names, one is 64bit, and the other isn’t; of the two, the 64bit entry (specifying refind_x64.efi
) should be used. If working, this should be set as the default bootloader in your BIOS.

Boot0009* UEFI: SanDisk PciRoot(0x0)/Pci(0x14,0x0)/USB(1,0)/HD(2,MBR,0x0,0x743260,0x2000)0000424f
This Sandisk USB has an MBR
partition scheme, so it would be unable to boot a UEFI boot loader such as rEFInd in any case.
Additional;
MANJARO-REFIND-INSTALLER
A special Manjaro GUI application exists called manjaro-refind-installer which allows to specify the default kernel to use when booting from rEFInd:
sudo pacman -S manjaro-refind-installer
Despite it’s name, this package doesn’t actually install rEFInd, but it does offer a convenient GUI method to choose the kernel that rEFInd should reference to boot Manjaro via the kernel stub.
This may or may not be useful to you.
Overall, my opinion is that a fresh install might be in order. Please see links (below) in case they are useful.
Regards.
Ventoy A New Bootable USB Solution Boot with a Ventoy USB and ISO files are automatically listed in the Ventoy menu. Any ISO present can then be launched and booted directly. The obvious advantage here is that instead of constantly writing a Manjaro Installer ISO to a new USB any time you might need it, every ISO you desire can be stored on the Ventoy USB and launched the same way. A 32GB capacity USB would allow ample space to store several ISOs of your choice; an 8GB capacity USB might …
Multiboot Windows and Linux …on Separate Disks Firstly, this guide is intended for multibooting with each OS on a separate disk. If you have a laptop with only one disk, or If you wish to install an OS on it’s own partition on the same disk, this guide is not for you (but read on regardless, if interested). Secondly, this guide assumes each OS will be installed using UEFI and not UEFI’s CSM or Legacy mode. This guide does not support the traditional BIOS boot method. Additionally, all disks …
I recently ran into boot issues with my Manjaro XFCE system. After some troubleshooting, I was able to resolve everything, so I’m sharing the steps here in case it helps others.
Situation
- My system was no longer booting correctly.
- I had previously used rEFInd as the bootloader.
- Goal: switch back to GRUB and restore normal boot behavior.
Steps to Solve
1. Removed rEFInd
- Deleted the EFI boot entry:
efibootmgr -b XXXX -B
- Removed the folders:
/boot/efi/EFI/refind/
/boot/efi/EFI/tools/ (if present)
2. Reinstalled GRUB
Booted into a Live ISO and used manjaro-chroot
:
sudo manjaro-chroot -a
Then reinstalled GRUB:
sudo pacman -S grub
grub-install --target=x86_64-efi --efi-directory=/boot/efi --bootloader-id=manjaro
grub-mkconfig -o /boot/grub/grub.cfg
Set boot order:
efibootmgr -o 0001
3. Booted via GRUB Fallback Mode
GRUB worked, but the standard kernel didn’t boot.
From the GRUB menu, I selected:
Advanced options for Manjaro > Fallback initramfs
4. Installed a New Kernel
Once booted into the fallback image:
sudo mhwd-kernel -i linux612
Error During Kernel Installation
During the mkinitcpio
phase, I received this warning:
findmnt: /etc/fstab: Einlesefehler in Zeile 1 -- ignoriert
Cause:
The first line in /etc/fstab
had an invalid syntax:
UUID=67E3-17ED /boot/efi vfat umask=0077 defaults 0 2
Here, the mount options (umask=0077
and defaults
) were split into two columns, which is not allowed.
Fix:
I corrected the line to:
UUID=67E3-17ED /boot/efi vfat defaults,umask=0077 0 2
Tested the file with:
sudo mount -a
→ No more errors!
Final Reboot & Result
After rebooting, the system booted normally and stably with the new kernel and GRUB.
Optional tweak (to always show GRUB menu):
ini
GRUB_TIMEOUT_STYLE=menu
GRUB_TIMEOUT=5
Summary
Removed rEFInd
Reinstalled and reconfigured GRUB
Booted via fallback initramfs
Installed new kernel (linux612)
Fixed
/etc/fstab
syntax error (line 1)System is now stable and boots normally
Big thanks to the community for all the guidance!
Mod edit:- Removed bolded “[Solved]”; that’s not how it’s done on the Manjaro forum. Use the Solution button instead.