Introduction

Experimental and unsupported: this is a proof-of-concept operator path, not a production platform. Review the security and release limitations before creating infrastructure.

UMKIM operates independent K3s workload clusters from one OCM hub. The hub provides management services; it is not itself a workload ManagedCluster.

For features added after the frozen 0.2.0 doc snapshot, see Recent changes.

Public onboarding path

1. Review prerequisites and prepare a fresh hub inventory.
2. Install the hub with an encrypted vault.
3. Create a registry site (with ingress entries as needed), then a FleetClusterJoin cluster in the authenticated console or with umkim.
4. Download the cluster bootstrap bundle through the authenticated console (or umkim cluster bootstrap) and treat it as sensitive enrollment material.
5. When the cluster is Ready, install a Helm chart with FleetApp onto that cluster.

Use the lifecycle pages for resource behavior and recovery limits; this section intentionally keeps the first-run path short.

For the full operational introduction (hub → YAIDAP lab → HA → GPU → LiteLLM) with verify checkpoints and screenshots, see Operational introduction.

For a screenshot tour of the same flow in the console (sites, bootstrap, Apps, Authelia, HA failover), see the console walkthrough. Screenshots may show the older Catalog UI; the v0.2 flow uses Apps and Library instead.