mkosi VS btrfs-progs

I work with a free and open-source software community called Fedora Project. I had the opportunity to moderate the talk of one of the maintainers of the Systemd suite during the annual contributor conference, Flock To Fedora 2023 where he talked about a tool named Mkosi.

For testing i prefer systemd-nspawn containers with mkosi. A neat tool for running your other fav. distro in a terminal. Works like a charm and integrates nicely in your system. Eg. logs and systemd services or CI testing.

- https://github.com/systemd/mkosi

- man:systemd-nspawn(1)

- man:machinectl(1)

you're gonna have to build this on an x86 pc. sudo dnf install arch-install-scripts bubblewrap gdisk qemu-user-static rsync systemd-container python3 -m pip install --user git+https://github.com/systemd/mkosi.git git clone https://github.com/leifliddy/asahi-fedora-usb.git cd asahi-fedora-usb

There's also mkosi: https://github.com/systemd/mkosi. This one outputs an iso or similar image file and supports many base distributions.

[1] https://github.com/systemd/mkosi/issues/376

System's mkosi is worth checking out too: https://github.com/systemd/mkosi I don't think it generates docker/OCI images directly, but it definitely can generate a tarball of the final image contents and then crane of a similar tool could package it up into an appropriate image. For just docker usage it's probably overkill, the main advantage would be it can build other image types like adding a kernel and init to be a fully bootable iso of VM image.

Constellation ensures that all K8s nodes run on AMD-based Confidential VMs (CVMs). CVMs are strongly isolated from the host and remain encrypted in memory at runtime. Constellation also ensures that all nodes run the same minimal mkosi-based node image.

At first glance I thought your project is a frontend for mkosi but then I saw that you support non-systemd targets too. Mentioning it here because it may be relevant to other users/developers.

Not an error I've seen before or that makes much sense to me at first glance, might want to take this to https://github.com/kdave/btrfs-progs/issues or the mailing list?

You can test it by allocating the space to see if it can use it all: e.g. fallocate -l 3.8t /mnt/elfstone2/butts should succeed. If it does, then it's probably fine even if the balance didn't do what it should have. If not and you run out of space with some unallocated space left on only one device, not so much. If its the same bug I saw that means if your fs was created pre kernel 5.19, it will not allocate correctly on >5.19 until the array is rebuilt under the newer kernel.

Mount options should be documented either at https://btrfs.readthedocs.io/en/latest/Administration.html or at the manual page section 5 (man 5 btrfs). Autodefrag is in section https://btrfs.readthedocs.io/en/latest/Administration.html#btrfs-specific-mount-options . In case you find more information missing, unclear, lacking cross-references please open an issue at https://github.com/kdave/btrfs-progs/issues .

thanks. I filled a feature request... https://github.com/kdave/btrfs-progs/issues/610

Resume sort of doesn't work, when you resumed it will have started a fresh balance with a high usage value (check your syslog to confirm).

That's the hint. It's an unfixed bug with the accounting when scrub is paused and resumed, the relevant issue is I think https://github.com/kdave/btrfs-progs/issues/227 .

[1] - https://github.com/kdave/btrfs-progs/issues/277

Hoping for a filesystem clone option at some point.