Add a helper function that collects all the manifest list digests from a
list of container specs and returns a FilesInput to be used with the
stage.
Use the function in the OS pipeline when adding containers. The
manifests input to the stage constructor will be empty if there are no
manifest lists in the container specs.
While resolving a manifest list digest, store the list digest to return
with the resolvedIds.
This is done for both types of manifest list:
application/vnd.docker.distribution.manifest.list.v2+json
and
application/vnd.oci.image.index.v1+json
Add the ListDigest to the container Spec struct and all its copies so we
can store list digests when they are available and pass them on to the
appropriate osbuild stages, sources, and inputs.
Copy the value whenever a spec is moved to a different representation.
The skopeo stage in osbuild supports an second optional set of inputs
called `manifest-lists`. This is an array of files, i.e.,
`org.osbuild.files` type input.
To support this we need a new type for the skopeo stage inputs that can
encompass both input types, images and manifest-lists.
Let's move all SQL queries together. In the following commit, we will actually
put all of them into a transaction in order to ensure atomicity.
This isn't a functional change, just code shuffling.
Signed-off-by: Ondřej Budai <ondrej@budai.cz>
If inserting a heartbeat or querying dependencies fail, we don't want to
actually dequeue the job from the database.
The failures may be:
- context timeout/cancellation
- network issues
Signed-off-by: Ondřej Budai <ondrej@budai.cz>
when it discovers "missing distros" b/c this check isn't suitable for
downstream RHEL anyway. See comment in script for more details!
Related: COMPOSER-1943
Update the version of the worker in the old worker/new composer
regression test. The test, up until now, was using a pinned version
of osbuild-worker (v51) which made use of dnf-json as a socket.
Testing a new composer version with such an outdated worker instance
is very unrealistic since the workers and composer are normally
updated in lock step, or in some cases the worker may be a version
behind.
This commit takes the current worker version and then downgrades it
to two versions ago (out of caution) to perform the tests. This
provides a more realistic scenario.
The size of the page blob is defined on creation and the blob is
zero-initialized. Therefore, we can just skip all the pages that contain
only zeros. This should save a lot of bandwidth if used on sparse files as
e.g. operating system images. (:
https://github.com/Azure/azure-storage-blob-go/ is deprecated, the main SDK
should be now used instead. Let's migrate the code. There should be no
functional changes.
Signed-off-by: Ondřej Budai <ondrej@budai.cz>
This adds a new field `Warnings` to the `ComposeReply`
struct, allowing to send back any warnings (e.g. deprecation
notices) generated during the `checkOptions` step of the
manifest initialization.
See also https://github.com/osbuild/weldr-client/pull/99 which
handles the weldr-client side of things.
Signed-off-by: Irene Diez <idiez@redhat.com>
This changes the `Manifest` function of the `ImageType`
interface so that any warnings detected during the
`checkOptions` step of the manifest initialization can
be propagated back to the Weldr-API (see next commit).
Signed-off-by: Irene Diez <idiez@redhat.com>
We sometimes skip testing customizations for image types that don't
support them, other times, we check for a specific error. If we don't
care about the specific error message that's received, running the test
and checking for error is better than skipping, because it ensures that
we're skipping the test for the right reasons and not accidentally
skipping a test for an image type that should support the
customizations.
Skip customizations tests for the EAP image.
Create a map of unsupported image types for each test to make adding new
ones easier.
Use a `nil` struct pointer for the pipeline names test instead of an
empty struct. Non-nil customizations can't be compared.
