Month: April 2026

Nested Partitions grub-update Mess…

Posted on by dk

Having imaged an old MBR-based partitioned disk to a VHD and back to bare metal on a now UEFI, GPT-based NVME SSD, update-grup was complaining every time:

# update-grub
Sourcing file `/etc/default/grub'
Generating grub configuration file ...
grub-probe: warning: Discarding improperly nested partition (hostdisk//dev/nvme0n1,gpt2,msdos1).
grub-probe: warning: Discarding improperly nested partition (hostdisk//dev/nvme0n1,gpt2,msdos2).
grub-probe: warning: Discarding improperly nested partition (hostdisk//dev/nvme0n1,gpt2,msdos3).
grub-probe: warning: Discarding improperly nested partition (hostdisk//dev/nvme0n1,gpt2,msdos4).

<above 4 lines repeats 11 times>

/usr/sbin/grub-probe: warning: Discarding improperly nested partition (hostdisk//dev/nvme0n1,gpt2,msdos1).
/usr/sbin/grub-probe: warning: Discarding improperly nested partition (hostdisk//dev/nvme0n1,gpt2,msdos2).
/usr/sbin/grub-probe: warning: Discarding improperly nested partition (hostdisk//dev/nvme0n1,gpt2,msdos3).
/usr/sbin/grub-probe: warning: Discarding improperly nested partition (hostdisk//dev/nvme0n1,gpt2,msdos4).

<above 4 lines repeats 29 times>

Found linux image: /boot/vmlinuz-6.17.0-22-generic
Found initrd image: /boot/initrd.img-6.17.0-22-generic
/usr/sbin/grub-probe: warning: Discarding improperly nested partition (hostdisk//dev/nvme0n1,gpt2,msdos1).
/usr/sbin/grub-probe: warning: Discarding improperly nested partition (hostdisk//dev/nvme0n1,gpt2,msdos2).
/usr/sbin/grub-probe: warning: Discarding improperly nested partition (hostdisk//dev/nvme0n1,gpt2,msdos3).
/usr/sbin/grub-probe: warning: Discarding improperly nested partition (hostdisk//dev/nvme0n1,gpt2,msdos4).

<above 4 lines repeats 23 times>

And so on… and so forth…

The great, wise ChatGPT was convinced that this was the work of the devil and that the “Microsoft Reserved Partition” partition was without a “soul” (file system) and would need to be “cast out” (excluded) by grub-prober… And then went on to instruct me to “add the file system UUID” to my /etc/default/grub via the GRUB_OS_PROBER_SKIP_LIST="<UUID>@/dev/nvme0n1p2" line… Never mind it just told me there is no file system for the MSR partition (and therefore no file system UUID)…

Continue reading →

Rebuilding Ubuntu LTS on WSL2 and Interfering Docker Integrations

Posted on by dk

In an attempt to get some AMD iGPU accelerated apps to run via Ubuntu LTS on WSL2, I had rebuild my Ubuntu LTS “VM” under WSL2.

  • I had to first remove the existing Ubuntu LTS (named “UbuntuLTS”): wsl --unregister UbuntuLTS

WARNING: Although multiple sources state that the source .vhdx file does not get deleted, it actually does get deleted for me! 🙄

  • I re-downloaded the .wsl file directly from Ubuntu, then installed the VM manually:

wsl --install --from-file "ubuntu-24.04.4-wsl-amd64.wsl" --name UbuntuLTS --location "D:\WSL\UbuntuLTS" --no-launch

Note that if you have Docker already installed, the default “distro” WSL2 may have been set to Docker’s docker-desktop:

  • set the correct default: wsl --set-default UbuntuLTS
  • run wsl.exe and it will launch into the newly created distribution, set a new (default) user and password… and ignore the stupid <3>WSL (299 - Relay) ERROR: operator():579: getpwuid(1000) failed error 🙄
    • NOTE: the error will “go away” after a wsl --shutdown and restart…


  • turn on the integration in Docker again…  and watch errors pop up…
    An error occurred while running the command. DockerDesktop/Wsl/ExecError: c:\windows\system32\wsl.exe -d -e sh -c cat - > ~/.docker/config.json: exit status 1 (stderr: , wslErrorCode: DockerDesktop/Wsl/ExecError)

Continue reading →

AMD Radeon Software Hanging Windows Explorer…

Posted on by dk

So, another day, another IT f*ck-up…

AMD’s in the crosshairs today. Due to having to muck around with AMD iGPUs in/on a WSL2-hosted Ubuntu LTS VM, I had to install an “older” AMD Radeon 26.2.2 driver…

Now, every time I right-clicked on a drive, a folder or a file in Windows Explorer, the stupid Radeon dashboard launches… and proceeds to “hang” Windows Explorer (with your mouse cursor turning into a spinning cursor) until something times out and hands Windows Explorer back its head…

  • No, killing the dashboard immediately does not give Windows Explorer back its head
  • No, attempting the registry hacks found across the ‘net did not help
  • No, disabling all non-Microsoft “Context Menu” items via ShellExView did not help

What did help was one of the suggestions by Google’s “AI Mode”!

i.e. F*ck Rename C:\Program Files\AMD\CNext\CNext\RadeonSoftware.exe (as an Administrator, of course).

“Unrooting” WSL…

Posted on by dk

WSL starts as root user be default.

Here’s how to ensure it runs as a “non-root user” (assumedly already created), which you then have to su and/or sudo <something> to act as root:

  • either/and:
    • modify a shortcut or change Windows Terminal’s Profile to set the distribution (if required) and the user name, and starting in the user’s home directory:
      • wsl -d <distro> -u <user> -cd ~
    • edit /etc/wsl.conf to always start with a specific user:
      • [user]
        default=<user>

Simple! (Not!)

Unifi Controller

Posted on by dk

I had a docker container of the jacobalberty/unifi “all-in-one” Unifi Network Application, and I somehow garbled the password…

The fix?

  • enter the container:
    • docker exec -it <container_name> /bin/bash
  • run MongoDB CLI:
    • mongo --port 27117
  • find the administrator user name/s:
    • use ace;
    • db.admin.find().forEach(printjson);
  • you should see one or more name entry/entries:
    • "name" : "<user_name>"
  • set/reset the password (to “password“):
    • db.admin.update( { "name" : "<user_name>" }, { $set : { "x_shadow" : "$6$ybLXKYjTNj9vv$dgGRjoXYFkw33OFZtBsp1flbCpoFQR7ac8O0FrZixHG.sw2AQmA5PuUbQC/e5.Zu.f7pGuF7qBKAfT/JRZFk8/" } } )

You should be able to log right back in with the reset password.

LAGGing pfSense… Manually…

Posted on by dk

Throughput through my KVM-hosted pfSense was abysmal for a 10Gbps link – despite the Intel Core i7 155h host w/128GB 5600Mhz RAM doing nothing else but hosting the pfSense VM (replete with VFIO-passthrough)…

I could get ~8-9Gbps iperf3 performance on from my PC to the WAN interface, which was “OK”, but iperf3 from pfSense out to various 10Gbps-capable public iperf3 servers was bad, as was Ookla speedtest.net results (from my PC).

In an attempt to figure out if the virtualisation was part of the issue, I attempted to run pfSense bare metal (via dual-booting – which had a whole storied journey itself)…

First boot brought up all the interfaces (thankfully) for selection – but no way to set up LAG then choose that for my WAN/LAN interface…

Continue reading →