Instead of running meson install and hoping for the best, let's build
distribution packages from the downstream packaging specs. This gets
us the following:
- Vastly simplified mkosi scripts since we don't need a separate initrd
image anymore but can just reuse the default mkosi initrd.
- Almost everything can move to the base image as its not the basis
anymore for the initrd and as such we don't need to care about the
size anymore.
- The systemd packages that get pulled in as dependencies of other
packages get properly uninstalled and replaced with our packages that
we built instead of just installing on top of an existing systemd
installation with no guarantee that everything from that previous
installation was removed.
- Much better testing coverage as what we're testing is much closer
to what will actually be deployed in distributions.
- Immediate feedback if something we change breaks distribution packaging
- We get integration with the distribution for free as we'll automatically
use the proper directories and such instead of having to hack this
into a mkosi build script.
- ...
The runner has a lot of useless things installed, taking ~10GB, and
jobs have started to fail when booting images due to lack of disk
space, so delete some directories to make room.
2024-02-27T20:20:58.0998709Z ##[warning]You are running out of disk space. The runner will stop working when the machine runs out of disk space. Free space left: 0 MB
Co-authored-by: Daan De Meyer <daan.j.demeyer@gmail.com>
Both building and booting a directory image is much faster than
building or booting a disk image so let's default to a directory
image.
In CI, we stick to a disk image to make sure that keeps working as
well.
The only extra dependency this introduces is virtiofsd which is
packaged in all distributions except Debian stable. For users
hacking on systemd on Debian stable, a disk image can be built by
writing the following to mkosi.local.conf:
```
[Output]
Format=disk
```
The mkosi github action doesn't set up the host machine for building
full images anymore. Instead, only sufficient packages are installed
to be able to build tools trees so we configure a fedora tools tree
to build the actual images.
- Use mkosi.images/ instead of mkosi.presets/
- Use the .chroot suffix to run scripts in the image
- Use BuildSources= match for the kernel build
- Move 10-systemd.conf to mkosi.conf and rely on mkosi.local.conf
for local configuration
This will bring in the fix for rawhide/tumbleweed builds (new libsolv
capable of handling zstd). If all goes well it will migrate to jammy
proper in a week and it can be reverted
mkosi detects whether /dev/kvm is available and uses it if it is. But
some GHA hosts have it, but it's broken and not supported, so we need
to explicitly disable it.
If the systemd version on the host is too old and there's no local
build directory, use the default tools tree which will build an
image containing all the tooling required to build systemd and use
that to build the other presets.
Configuration now takes priority over CLI options so we have to
configure the defaults for settings that we want to allow overriding
from the CLI. We also explicitly set some other settings so that they
can't be overridden from the CLI anymore. For example the base and
initrd image should never be made bootable so we set Bootable=no
explicitly for both.
The mkosi Arch CI doesn't work as the keyring package is out
of date and cannot be built due to various build toolchain
issues. Disable the job as it always fails and confuses
submitters.
This update introduces the explicit Dependencies= setting, instead
of relying on implicit dependencies via alphanumerical ordering.
We also take the opportunity to rename the "final" preset to the
"system" preset, which seems like a better name.
When the mkosi CI fails to boot, it just sits there waiting at the emergency
console until the job times out. Add a drop-in for emergency.service in the
CI configuration so that instead it exists immediately.
mkosi now supports CentOS SIGs natively so we drop our own definition
of that and use the mkosi builtin one. We also enable hyperscale for
both CentOS 8 and CentOS 9 for consistency and add epel-next as well
which is a requirement for Hyperscale.
We now run repart before starting systemd-nspawn to make sure that
the root partition is also generated when we boot the image in a
container instead of a VM.
To make sure we start from scratch for both the container boot and
the VM boot, we also enable Ephemeral to make sure all changes to
the image are ephemeral.
vsock should work properly after the latest release of mkosi. But
to make sure it works, let's exit with 123 in case of success and
check for that in Github Actions.
We only really care about lowering the device timeout so we get to
a shell faster when the root device doesn't appear so let's only
lower that timeout instead of lowering all default timeouts.
Let's start moving towards a more involved partitioning setup to
test our stuff more when using mkosi.
The root partition is generated on boot with systemd-repart.
CentOS supports neither erofs nor btrfs so we use squashfs and xfs
instead.
We also enable SecureBoot= locally for additional coverage. This
and the use of verity means users need to run `mkosi genkey` once
to generate the keys necessary to do secure boot and verity.
Let's make sure we log more of what mkosi's doing so we can debug
issues better. Note this also makes mkosi set SYSTEMD_LOG_LEVEL=debug
when running programs so we'll get all the systemd debug logging as
well.
Instead of building the initrds for the mkosi images with dracut,
let's switch to using mkosi presets to build the initrd with mkosi
as well.
This commit splits up our single image build into three separate
mkosi presets:
1. The "base" preset. This image contains systemd and all its runtime
dependencies. The sole purpose of this image is to serve as a base image
for the initrd and the final image. It's also responsible for building
systemd from source with the build script. The results are installed into
the base image. Note that we install the systemd and udev packages into this
image as well to prevent package managers from overriding the systemd we built
from source with the distro packaged systemd if it's pulled in as a dependency
by another package from the initrd or final profiles.
2. The "initrd" preset. This image provides the initrd. It's trivial and does
nothing more than packaging the base image up as a zstd compressed initramfs and
adds /init and /etc/initrd-release symlinks to the image.
3. The "final" preset. This image builds on top of the base image and adds
a kernel and extra packages that are useful for testing and debugging.
We also split out the optional kernel build into a separate set of config files
that are only included if a kernel to build is actually provided.
Note that this commit doesn't really change anything about how mkosi is used.
The commands remain the same, except that mkosi will now build all the presets
in order. "mkosi summary" will show the summary of all the presets. "mkosi qemu,
boot, shell" will always boot the final preset. With "-f", all presets will be
built and the final one is booted. "-i" makes a cache of each preset.
The only thing to keep in mind is that specifying config via the mkosi CLI will
apply to each of the presets. e.g. any extra packages added with "-p" will be
installed in both the initrd and the final image. To apply local configuration
to a single preset, create a file 00-local.conf in
mkosi.presets/<profile>/mkosi.conf.d and put all the preset specific configuration
in there.
Daan De Meyer
Luca BoccassiandDaan De Meyer
dependabot[bot]
Jan Janssen