Goal:

Today Satellite ships with multiple compute providers. The documentation splits it into cloud providers and virtualization providers, but internally we just have compute providers:

• OpenStack (default, foreman-openstack package)
• EC2 (default, foreman-ec2 package)
• Google Compute Engine (default, foreman-plugin-google package)
• Microsoft Azure (default, foreman-plugin-azure_rm package)
• libvirt (default, foreman-libvirt package)
• vmware (default, foreman-vmware package)
• OpenShift Virtualization (optional tech preview, foreman-plugin-foreman_kubevirt package)

In upstream they are all optional and none are installed by default.

Acceptance criteria:

• All of the above compute providers are present in the container image
• All the compute providers built into core (OpenStack, EC2, libvirt, vmware) are enabled unconditionally
• Users can enable each of the plugin compute resources individually. This implies they are individual features.

Open questions:

• Do we enable the current set of compute resources by default? Can we solve that via profiles?
• How do we document this? https://docs.redhat.com/en/documentation/red_hat_satellite/6.18/html/provisioning_hosts/provisioning_virtual_machines_kubevirt_kubevirt-provisioning#adding-kubevirt-connection_kubevirt-provisioning has an instruction to enable the plugin and that needs to be updated. If we also don't enable all by default, the instructions also need to be updated.

Additional Info:

• This was found out during the investigation of Spike SAT-38963 and discussed during refinement meeting.
• The Containerfile already contains a mechanism to enable/disable individual plugins
• For the exact feautre naming https://issues.redhat.com/browse/SAT-38964?focusedId=28410992&page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#comment-28410992 has some suggestions. During refinement the consensus was that this was the best path forward and we don't prefix it with compute- or similar.