PoC: use OCM K8s toolkit for component consumption

[reshnm]

Both the OpenMCP bootstrapper as the OpenMCP controller need a way to consume a OpenMCP component.
The OpenMCP component describes a complete OpenMCP delivery that is uesed to set-up a custom tailored OpenMCP landscape.
The OpenMCP component contains all resources like OCI images that are needed to deploy and run a OpenMCP landscape. E.g.:

graph TD
    OpenMCP --> GardenerClusterProvider["GardenerClusterProvider<br/>OCI Image: gardener:latest"]
    OpenMCP --> ProjectWorkspaceOperator["ProjectWorkspaceOperator<br/>OCI Image: pwo:latest"]
    OpenMCP --> LandscaperProvider["LandscaperProvider<br/>OCI Image: landscaper:latest"]

Loading

In order to consume the resources of all OpenMCP sub components, especially the OCI images, it is required that they are stored in a way so that they are concumable by the K8s container runtime. To ensure that, the OpenMCP component resources shall be replicated in a local OCI registry.
This will also make the OpenMCP product more robust against unavaiability of the upstream OCI respository.

To fulfill these requirements we want to investigate the OCM K8s Toolkit:

https://github.com/open-component-model/ocm-k8s-toolkit

[reshnm]

Currently blocked because of open-component-model/ocm-k8s-toolkit#131

[In-Ko]

blocker seems fixed according to the link above. But you mentioned a follow-up problem this morning, right ? Can you quickly elaborate ?

[reshnm]

Blocker is fixed. We are now waiting for the ocm-k8s-toolkit to use the new oci storage backend for being usable in our context.

[In-Ko]

And is there a timeline for the new OCI storage backend to get usable by us ?

I believe we should avoid any hard-dependencies towards other teams implementations, if we are talking several sprints until this becomes available.
Is it possible to come up with something we can use as long as the OCM OCI Storage stuff is not available, even if it is just a workaround ?

[In-Ko]

Talked to the OCM folks. Timeline for the OCI storage backend will most likely not fit to our timelines (will likely come later than H1/25). So seems we have to go for a workaround for now.

[In-Ko]

Follow-up tasks have been defined:

• EPIC: Implement OpenMCP bootstrapper without the usage of OCM K8S Toolkit #45
• Implement OpenMCP bootstrapper with the usage of OCM K8S Toolkit #46

This PoC task can be closed now.