Open martinmo opened 1 month ago
Btw, KaaS v1 has the same behavior, but I am not sure if I should open an issue for it as well, because it's already deprecated.
@jschoone I assigned you, so it appears in the Kanban board that we review tomorrow, I hope that's ok.
TBH, I didn't expect the discussion in the Container call to be that passionate and thought this would be a quick and easy change. To support the point from the standard's point of view a bit:
The original intent for using the space character is coming from the Image Metadata Standard in https://docs.scs.community/standards/scs-0102-v1-image-metadata/#naming
We suggest plain OS images to be named "Distribution Version", e.g. "openSUSE Leap 15.3" or "Ubuntu 20.04" or "CentOS 8", "Windows Server 2012R2". We do not normally recommend to add more detailed patch levels into the name.
The Standard Images standard works with an image spec file (with the format of that YAML specified in this standard). The IaaS certificate scope version is tied to a specific version of the image spec file, as you can see in the last row of the table at https://docs.scs.community/standards/scs-compatible-iaas (for IaaS v4, it is scs-0104-v1-images.yaml).
I created https://github.com/osism/openstack-image-manager/issues/820 https://github.com/osism/openstack-image-manager/issues/821 to help overcoming this issue...
/kind bug
What steps did you take and what happened: Workload clusters created with cluster stacks will use node image names such as
instead of
The latter is what the scs-0102-v1 and scs-0104-v1 require (note the space instead of the dash).
This means that a compliant public image such as
ubuntu-capi-image v1.28.11
uploaded by an operator will not be used for the K8s nodes. Instead, an image will be uploaded on-the-fly (if it wasn't uploaded during a previous run) and marked asprivate
. This image is missing a lot of the properties that are usually set, such asimage_source
(already reported in #23). Also, sinceqcow2
seems to be used, in the case of Ceph this will circumvent the ability to use copy-on-write RBD clones and use a lot of space.What did you expect to happen: For Kubernetes vX.Y.Z, cluster stacks should use
ubuntu-capi-image vX.Y.Z
as the image name.Anything else you would like to add: [Miscellaneous information that will assist in solving the issue.]
Environment:
kubectl version
)/etc/os-release
):