!oNSIfazDqEcwhcOjSL:matrix.org

disko

353 Members
disko - declarative disk partitioning - https://github.com/nix-community/disko89 Servers

You have reached the beginning of time (for this room).


SenderMessageTime
21 Mar 2025
@disco_stick:matrix.orgfood style edible product changed their display name from It's scary when I'm focused to OpenAI brand ChatGPT®.15:23:11
@marcusramberg:matrix.orgMarcus joined the room.20:19:03
22 Mar 2025
@disco_stick:matrix.orgfood style edible product changed their display name from OpenAI brand ChatGPT® to Lizard in a Skin Suit.19:52:47
24 Mar 2025
@nicoty:kde.orgnicoty

Hello. I'm currently working on improving disko's support for bcachefs. So far, I have made a PR to add encryption support. Currently, I am working on adding subvolume support. However, I am having trouble getting this to work. This is the output when I run nix build --no-link .#checks.x86_64-linux.bcachefs-subvolumes --show-trace -L -v.

I suspect that a reason that this could happening might be due the order in which mountpoints are mounted. My theory is that descendant mountpoints (e.g., /ancestor/descendant) could be being mounted before ancestor mountpoints (e.g., /ancestor) have been created/mounted.

Does disko mount filesystems such that ancestor mountpoints are mounted before descendants, or is this behaviour dependent on how the _mount option (made with diskoLib.mkMountOption) for the relevant filesystem type is implemented?

I have tried to debug my implementation using nix build .#checks.x86_64-linux.bcachefs-subvolumes.driverInteractive; result/bin/nixos-test-driver --keep-vm-state, but this doesn't seem to be working properly, as it doesn't show the bcachefs filesystem that should have been created:

$ lsblk -f
NAME FSTYPE FSVER LABEL UUID                                 FSAVAIL FSUSE% MOUNTPOINTS
fd0                                                                         
sr0                                                                         
vda  ext4   1.0   nixos a7ac900e-9d30-45da-b745-56d7d190aedd  888.3M     2% /
vdb  

In fact, it shows that same lsblk output for various other examples I tried it with (I also tried it with btrfs-subvolumes and simple-efi).

Is there another way I can go about debugging this?

10:47:05
@projectinitiative:matrix.orgprojectinitiativeHey! I was working on adding advanced options for Bcachefs a few months ago. May want to poke around here: https://github.com/nix-community/disko/pull/961 There are some current gotchas that cause some of the issues you are facing. Using the bcachesfs fs formatter is limited which is why I went the route of creating a custom type similar to mdadm and zfs. There are some systemd bugs that prevent the FS ground getting mounted properly/consistently during reboots12:45:51

Show newer messages


Back to Room ListRoom Version: 10