Goal:

As an administrator, I would like to deploy an OpenShift 4 cluster to an on-premise AWS Outposts environment.

Problem:

Customers were able to deploy to AWS Outposts with OCP 3.11, but our global configuration in OCP 4 doesn't support this.

At a bare minimum, we should be able to provide a user-provisioned infrastructure (UPI) workflow for deploying OpenShift to AWS Outposts. However, it's likely many customers will want a fully-automated deployment method, so we should try to also deliver an installer-provisioned infrastructure workflow.

Why is this important:

• Many of our customers would like to move to using on-premise cloud infrastructure, but lack of support for AWS Outposts is prevent them from being able to deploy OpenShift.

 
Lifecycle Information:

• Core

Previous Work:

Dependencies:

• Custom API endpoint support w/ CA
• Cloud / Machine API
• Image Registry
• Ingress
• Kube Controller Manager
• Cloud Credential Operator
• others?

• Requires access to local AWS Outposts environment
• Working CI environment for GA product support

Prioritized epics + deliverables (in scope / not in scope):

• Allow on-premise AWS Outposts regions to be selected/specified for OpenShift cluster deployment
• Provide mechanism for defining custom API endpoints necessary for cluster deployment
• Enable customers to use their own managed internal/cluster DNS solutions due to provider and operational restrictions

Estimate (XS, S, M, L, XL, XXL):

Open questions:

Demo: https://drive.google.com/file/d/1QDUpJwCnPUskFoWOh9XgjVYXrW0p3NOj/view