A recent commit (8a7b6d3) fixed the ostree config stage and thus
we actually properly set the readonly flag for the deployment.
This broke the image since we did not specify the `rw` kernel flag
and as a consequence the ostree mounting code in the initrd broke.
If the object's id does not match with the one supplied for the
commit, we create a clone. Otherwise we store the tree.
The code path is arranged in a way that we always go through
`Object.store_tree` so we always call `Object.finalize` as a
prepration for the future, where we might actually do something
meaningful in the finalizer, like reset the *times or count the
tree size.
Remove copy-on-write support from `objectstore.Object`. The main
reason for introducing copy-on-write was to save an additional
copy in the non DAG-pipeline model[1]. With the introduction of
the latter and the explicit `--export` option, we can achieve the
same result without the complexity of copy-on-write semantics.
[1] See commit 39213b7, part of 3b7c87d5..42a365d1 changeset.
There is little use in sharing the store between test, quite to
opposite: all tests expect a clean store and some currently set
that up themselves. Create a fresh store for each test.
When committing an object to the store, clone it if the current
stage is not the latests stage, i.e. `todo` has still entries.
This is the second step of the removal of copy-on-write support
in `Object`.
Add a new `clone` parameter to the `commit` method on `ObjectStore`
that when used will clone the object to the store instead of using
the `store_tree` method which moves the object and resets it. This
is the first step of removing copy-on-write support from `Object`.
In newer version of the container storage package the config file
moved from `/etc/containers` to `/usr/containers/`. The later is
not marked as config, so we don't want to change it.
The current containers code[1] will read _either_ a file in `usr`
or in `etc` depending on the existence of the latter. This we can
not just write the keys we want into a file in `/etc/containers`
without losing all other defaults set in the config file.
A new option `filebase` is therefore added, that when given will
be read and form the bases of the configuration data. Then data
from the target file (given via `filename`) will be merged into
and finally the actual configuration will be applied on top.
[1] 232bf398bd/types/options.go (L85)
If the existing config does not have a section that is being set
by the stage it currently would crash since we assume that it
exists. Default to an empty dict in that case.
Create a squashfs image from the root file system and append it
to the existing initram fs. Passing `root=live:/rootfs.img`
is enough to use the existing dmsquash machiner in dracut.
Add an new module with utility functions to inspect PE32+ files,
mainly listing the sections and their addresses and sizes.
Include a simple test to check that we can successfully parse the
EFI stub contained in systemd (systemd-udev package).
add the ability to configure the grub timeout. Today it defealts to 1
minute and we can't configure it from osbuild-composer either (the
simplified-installer is mainly using this afaict and 1 min is just too
much).
Signed-off-by: Antonio Murdaca <antoniomurdaca@gmail.com>
According to the OCI Image Index specification the mediaType field is not mandatory
Assume that it is a list if mediaType is not set while manifests field is
Signed-off-by: Ygal Blum <ygal.blum@gmail.com>
This code was clearly influenced/copied from rpm-ostree,
since it's now duplicated let's cross-link to help ensure that
if someone wants to change this they hopefully consider changing
rpm-ostree too.
This stage allows to configure a systemd-journald
config drop-in file located in
`/usr/lib/systemd/journald.conf.d`.
Currently it supports the following options for
the 'Journal' section: Storage, Compress, SplitMode,
MaxFileSec, MaxRetentionSec, SyncIntervalSec, Audit.
Signed-off-by: Irene Diez <idiez@redhat.com>
From machine-id(5): machine ID is a single *newline-terminated*, hexadecimal,
32-character, lowercase ID.
This commit fixes that by adding a newline.
This broke us on Fedora 38:
This pull request added `set -e` to kernel-install:
https://github.com/systemd/systemd/pull/23881
This caused F38 pipeline failing because the script also runs
`read -r MACHINE_ID </etc/machine-id`. `read` returns 1 if the input doesn't
contain a newline, thus after the addition of `set -e`, the script started
crashing with our bad machine-id. This caused crashing of the whole image build
in the grub2 stage which requires kernel-install to work (because kernel-install
is responsible for putting vmlinuz into /boot).
