Creating a workload CAPX cluster spanning Prism Element clusters¶
The scenario and features described on this page are experimental. It's important to note that they have not been fully validated.
This page will explain how to deploy CAPX-based Kubernetes clusters where worker nodes are spanning multiple Prism Element (PE) clusters.
All the PE clusters must be managed by the same Prism Central (PC) instance.
The topology will look like this:
- One PC managing multiple PE's
- One CAPI management cluster
- One CAPI workload cluster with multiple
Refer to the CAPI quickstart to get started with CAPX.
To create workload clusters spanning multiple Prism Element clusters, it is required to create a
NutanixMachineTemplate resource for each Prism Element cluster. The Prism Element specific parameters (name/UUID, subnet,...) are referenced in the
- Create a management cluster that has the CAPX infrastructure provider deployed.
- Create a
cluster.ymlfile containing the workload cluster definition. Refer to the steps defined in the CAPI quickstart guide to create an example
By default there is only one machine template and machine deployment defined. To add nodes residing on another Prism Element cluster, a new
NutanixMachineTemplateresource needs to be added to the yaml file. The autogenerated
NutanixMachineTemplateresource definitions can be used as a baseline.
Make sure to modify the
Apply the modified
cluster.ymlfile to the management cluster.