Multicluster Engine (MCE) is required for all HCP tests.
Important — pod capacity per node (not just CPU/RAM): By default, a cluster is installed with a maximum of 250 pods per node. A hosted cluster adds roughly 30–40 extra pods plus 2+ worker VM pods (virt-launcher). On an SNO that is both control plane and worker, the node-wide pod limit (Too many pods) can be hit before CPU or RAM runs out. Before HCP tests, check: oc describe $(oc get nodes -o name) | grep 'Non-terminated Pods' and raise the pod limit if necessary.
1OpenShift Console → Ecosystem → Software Catalog → Operators → search for
Multicluster Engine for Kubernetes → install (skip if PRE-14 already installed MCE)2Create MultiClusterEngine CR (
availabilityConfig: Basic for SNO)3Enable HyperShift local hosting component (patch on MCE CR)
4Verify:
oc get pods -n hypershift → operator Running5Install hcp CLI — do not use hypershift from GitHub Releases (that is the dev tool, not the product CLI): the product hcp binary is served by the cluster via a ConsoleCLIDownload Route
oc get ConsoleCLIDownload hcp-cli-download -o jsonpath='{range .spec.links[*]}{.text}{"\t"}{.href}{"\n"}{end}'→ pick the URL for your platform, then:
curl -kL <url> -o hcp.tar.gz && tar xvzf hcp.tar.gz && chmod +x hcp && sudo mv hcp /usr/local/bin/hcpIf DNS/TLS issues during download: see Commands/YAML tab, section Troubleshooting Download
6Verify:
hcp version — client/server versions must match (Server Supports OCP Versions should include your OCP version)