The new "latest" aka v1.51.1 started throwing unused errors during runs,
blocking all the other tests, to quickly unblock that until a proper
fix/rewrite is found, pin the version to the previous one.
Signed-off-by: Antonio Murdaca <antoniomurdaca@gmail.com>
I use this tool quite a lot and I often want to use the CDN content, so
I would very much appreciate RHSM support. :)
Signed-off-by: Ondřej Budai <ondrej@budai.cz>
The _dnfrepo is full of branches, turn off warning.
The keyfile doesn't use 'with' because the file needs to remain
available when the function exits. Cleanup of persistdir will clean up
the temporary file used for the key.
This allows verification of repository metadata signatures.
The gpgkeys field is a list of key urls, or the gpg key itself, starting
with '-----BEGIN PGP PUBLIC KEY BLOCK-----'. These will be written to a
temporary file, and that file:// url will be passed to dnf.
DNF supports more than one GPG key. It is possible that one may be used for
signing packages, and another to sign the repository metadata. This
renamed GPGKey to GPGKeys internally. It does not change the on-disk
repository json format.
When the image definitions were updated to the new framework, I failed
to update the dracut modules for Anaconda installers to match the
existing ones.
The changes in the manifest are at commit
c4af0a1886.
The nvdimm module and the additional drivers were removed.
The nvdimm module in particular is required for http boot but should
only be specified for RHEL 9. In RHEL 8 it is part of the default set
of modules.
See 02bb7a0b4f and
dc95382ba3 for the original commits that
introduced these changes.
Similarly to the change made for rhel9, adding the sos package
gives users a built-in way to gather system logs and debug info.
Signed-off-by: Irene Diez <idiez@redhat.com>
Adds the sos package to the edge commit package set so that
users have built-in way to gather system logs and debug info.
Signed-off-by: Irene Diez <idiez@redhat.com>
Add the rhui-azure rpmrepo snapshots to aarch64 repositories.
The repos are labelled as x86_64 but they're actually 'noarch' and we
name them x86_64 when creating the snapshots.
The previous error didn't make it clear where the issue was coming from.
Now it explains that the problem is that a partition table for a given
architecture isn't specified on the image type.
Rework the stage to not reimplement `FilesInput` as
`RPMStageInput`, but instead use the one common
`FilesInput` implementation and its supported
references.
Signed-off-by: Tomáš Hozza <thozza@redhat.com>
Rework the stage to not reimplement `FilesInput` as
`IgnitionStageInput`, but instead use the one common
`FilesInput` implementation and its supported
references.
Signed-off-by: Tomáš Hozza <thozza@redhat.com>
Rework the stage to not reimplement `FilesInput` as `QEMUStageInput`,
but instead use the one common `FilesInput` implementation and its
supported references.
Signed-off-by: Tomáš Hozza <thozza@redhat.com>
Rework the stage to not reimplement `FilesInput` as
`FDOStageInput`, but instead use the one common
`FilesInput` implementation and its supported
references.
Signed-off-by: Tomáš Hozza <thozza@redhat.com>
The `FilesInputs` was since the beginning an XZ-specific implementation
of the input, but it was implemented in the `files_input.go` in a false
hope that it could be used as a generic stage inputs by any stages. It
turned out that various stages require different implementation of
its input. Specifically there is usually a stage-specific key, which has
assigned a common input type. For XZ stage, the key is `file`.
Remove `FilesInputs` and instead implement `XzStageInputs` which is now
accepted by the XZ stage.
Fix all affected pipeline implementations that use XZ stage.
Signed-off-by: Tomáš Hozza <thozza@redhat.com>
Rework files input implementation to support all reference types
supported by the input schema. Also implement helper functions to
generate supported reference types. In some cases, the reference
supports e.g. referencing multiple pipelines in the stage inputs, but
this is currently not implemented, since no pipeline in composer uses
it.
Rework the files input unit tests to cover the new functionality.
Adjust all code affected by the changes made to files input.
Signed-off-by: Tomáš Hozza <thozza@redhat.com>
Updated manifests for RHEL 9.1.
The changes added here would have been part of commit
73882b7289 but the lack of repositories
defined for the distro prevented test case generation.
In commit 7c8b9f4da8 I intended to make a
copy of the RHEL 9.1 repos under the key "rhel-9" so that the
RHEL 9 distribution, which is currently an alias to RHEL 9.1 (the
current GA) would have valid test repositories available for the test
case generator. Instead, I accidentally replaced the "rhel-91"
repositories with the ones for "rhel-9".
This commit fixes that mistake.
Changes:
- Removed unnecessary packages from build roots. Build packages are
added on-demand.
- Some stages changed order in the OS pipeline for azure-rhui. This
will have no functional effect.
- RHSM Facts stage added to all RHEL 7 manifests. This wont be used
since RHEL 7 is not built in the cloud service. The RHSM fact is
added by the test case generator to all image types to test that the
option is working properly.
- Azure archive pipeline renamed to xz like in all the other rewritten
image types that use the xz pipeline.
- Azure image filename before archiving has changed from 'disk.vhd' to
'image.vhd'. Again, this will have no functional effect because the
name of the final artifact is unchanged (disk.vhd.xz).
The RHEL 7 images need to enable the force_autorelabel option for the
SELinux stage in osbuild. This option should almost never be used but
it was added specifically for RHEL 7. With the rewrite to the new
definitions and the sharing of pipeline code between all distros, we
need to add support to all stages of the pipeline generation to be able
to enable it.
The RHEL 7 vpc subformat in qemu does not support force_size so we need
to be able to disable it. The parameter in all parts is defined as a
pointer because the default value is 'true'. Not specifying it will
keep the option in the osbuild stage as 'nil', falling back to 'true' in
osbuild.
Older OS versions (RHEL 7) with older versions of grub2 don't support
BLS entries. Setting NoBLS to true configures the bootloader with
traditional menu entries through the grub2.legacy osbuild stage. This
requires specifying extra information for the OS to the pipeline:
version, product, and nick.
Add the partition tool as an option on the Raw pipeline. Set it to the
old value (sfdisk) by default.
Expose the option up through the liveImage image kind so that the
distribution can set it if needed.
For RHEL 7, set it to sgdisk.
- Replace Manifest() and PackageSets() imageType methods with (adapted)
copies from RHEL 8.
- Replace pipeline functions with liveImage image function.
- Specify xz compression for Azure RHUI.
- Add similar package name overrides as we did in RHEL 8. For RHEL 7,
we need to modify the capitalisation of python3-pyyaml.
Moved the qcow2 image type definition to the top of the file for
consistency with the other image type files.
Separated the default image config struct from the base image type
definition to make it easier to read.
Moved the Azure image type definition to the top of the file for
consistency with the other image type files.
Separated the default image config struct from the base image type
definition to make it easier to read.
Update the implementation of the distro.Distro interface to match the
one in RHEL 8, 9, and Fedora. The main change is that the runner is a
runner.Runner and not a string.
The runner name is now rhel79 (changed from rhel7). This is
functionally equivalent based on osbuild's runner version fallback
logic.
The qcow2 image type for RHEL 7 doesn't have packageset chains defined.
This means that the blueprint packages are never merged into the os
pipeline.
This is unnecessary right now because of the upcoming rewrite, but it
will minimise the differences that will show up in the manifest.
The qcow2-customize manifest has an added block of options for the
grub2.legacy stage because now the dracut-config-rescue package is being
installed in the image.
