KCP-Stable Compute Internal Service

Slack: #forum-kcp, #forum-acm-singapore-gateway-service

Problem statement

Engineering teams need access to KCP Workspace environments backed by real compute to develop their respective applications.  Engineers need to get accustomed to using hosted control planes to develop on and also reduce setup and “time to development” for teams.

Goals

• Use RHACM, HyperShift, and an OCM-KCP integration prototype to provide compute services in the kcp-stable environment
• Empower engineering teams to work directly on KCP without having to run KCP locally on their workstations

Non-Goals

• Open this service up externally for community use

Background

• KCP (Control Plane Service and Workload Service) - doc
• Prototype open-cluster-management integration with KCP - https://issues.redhat.com/browse/ACM-1073 
• Open Cluster Management + KCP demo - recording slides doc

The ACM team built a prototype that demonstrated how ACM could register and connect physical clusters to running KCP workspaces and deploy the KCP syncers.  This internal service would leverage and deploy that prototype.

Proposal

The ACM engineering team will:

• Deploy a RHACM hub cluster alongside the kcp-stable environment and configure the ocm-kcp integration prototype controllers
• Create a to-be determined amount of HyperShift hosted clusters on the RHACM hub clusters
• Connect the hosted clusters to one or more KCP locations
• Provide SRE support for the hub and hosted clusters
• Lifecycle the kcp syncer agents as the stable environment gets updated
• TBD:  add *ks clusters to the location for further testing

Security Considerations

• Do we need to restrict the users access to the HyperShift hosted cluster?

Resources

Original Google Doc