It turned out that the target was never adopted by the service, thus it
is being deleted as part of upload code consolidation.
Signed-off-by: Tomáš Hozza <thozza@redhat.com>
This method is not particularly useful anymore. Its purpose was to
initialise the ImageOptions from an ImageRequest with the appropriate
size and partitioning mode. However, the partitioning mode was also
being set later using request.GetPartitioningMode(). More importantly,
setting the size on the ImageOptions caused issues with the interaction
between filesystem and partitioning customizations as well as the image
request size (see #4705). The correct thing to do here is to map the
ImageRequest.Size directly onto ImageOptions.Size, without taking into
account ImageType or the Blueprint Customizations. The rest are
considered when generating the manifest in images, either when preparing
the Manifest() call or when generating the partition table. This makes
it easier to trace and reason about the effect of each option. This
kind of decision making in the API layer makes it difficult to maintain
the logic, since it requires duplicating the decision making or, as we
had now, making certain specific combinations impossible.
Fixed distribution list type, function signatures (later versions parse
uuids in the generated code), defaults actually being set and added
x-go-type where needed.
Previously, the worker was determining the GCE image guest OS Features
on its own, based on the OS name. This caused problems, in case the
osbuild-composer was of a newer version than the worker.
Example:
osbuild-composer contained support for c10s GCE image type and its
implementation also contained the proper guest OS Features list for it.
However, when the worker got the osbuild job, it built it and tried to
fetch the guest OS Features for the distro. Since its implementation was
too old, it didn't contain the code that added the actual support for
c10s GCE images and got no guest OS features list (which is the default
for unsupported distros). The image was successfully uploaded and
shared, but it does not boot in GCP, because it does not know that it
should use UEFI to boot it.
This behavior could be considered a bug. The worker should be dumb. It
should not be making decisions about the image features, but instead it
should take them from the upload target options. And composer should be
the authoritative source of truth for this. Because otherwise, we
basically have two components that need to be updated in sync to add
support for GCE images on a new distro.
Move the GCE image guest OS features to the GCP upload target options.
The worker will just take what is specified there and use it when
importing the image to GCP. As a compatibility layer for the case when
the composer would be older than the worker (unlikely, but still),
worker will try to determine the image guest OS features in case the
list in the upload target options is empty.
Extend the GCP functional tests to check that the imported image has at
least some guest OS features set.
Signed-off-by: Tomáš Hozza <thozza@redhat.com>
Read the upload target types and options in the UploadTargets array of
the ImageRequest and initialise the Target array. If the top-level
(old) UploadOptions are also specified, prepend them to the array using
the image type's default target type.
Each upload target type is checked against a support map for
compatibility.
Add an array of targets in the imageRequest and return an array from
ImageRequest.GetTargets() (renamed from GetTarget()). Currently, the
function still only returns one target, the default for the image type
with the top level upload options.
Separate the target selection in GetTarget() into two steps. First
determine the default target name for the image type and then use the
name to initialise the target object. This is a bit more work (and
double switching) but will be needed to support selecting targets
externally.
Separate the handling of each individual target type into its own
function called by GetTarget()'s case switch. This makes the function
more readable and the target object creation reusable.
Added an empty line after each creation of irTarget to make it easier to
visually distinguish the cases that fall through.
This moves some of the code from the PostCompose function in handler.go
into methods on the OpenAPI ComposeRequest and ImageRequest structs.
In compose.go I have added several methods.
GetBlueprintWithCustomizations takes the ComposeRequest customizations
and builds a Blueprint struct.
GetPayloadRepositories returns the custom payload repos.
GetSubscription returns the ImageOptions setup with optional
subscription information from the request.
In imagerequest.go I have added GetTarget which takes the upload
options and returns a Target. This moves the giant switch statement,
which may also benefit from further simplification at some point.
GetOSTreeOptions returns the OSTree ImageOptions if there are ostree
settings in the ImageRequest.
GetImageOptions returns the distro.ImageOptions with the size set.
This commit only moves the code, making PostCompose easier to read. All
tests still pass.
cloudapi: Move the size handling to a method on ImageRequest
