| name | cluster-provision |
| description | Provision an ephemeral OpenShift cluster via the OpenShift CI Gangway API. Claims a cluster from a Hive ClusterPool, downloads and decrypts kubeconfig, and validates connectivity. Requires GANGWAY_TOKEN and KUBECONFIG_ENCRYPTION_KEY env vars. Clusters auto-expire after ~4 hours. Use when you need a real OpenShift cluster for blackbox testing or integration work.
|
| argument-hint | [KUBECONFIG_PATH] [OCP_VERSION] |
| allowed-tools | ["Bash(bash .claude/scripts/cluster-provision.sh *)","Bash(oc --kubeconfig=*)","Read"] |
Cluster Provision
Provision an ephemeral OpenShift cluster from a Hive ClusterPool via the OpenShift CI Gangway REST API.
Arguments
Parse $ARGUMENTS into at most two values before invoking Bash:
- Arg 1: KUBECONFIG path (default:
/tmp/k)
- Arg 2: OCP version (default:
4.18)
Set variables from $ARGUMENTS:
KUBECONFIG_PATH = first arg or /tmp/k
OCP_VERSION = second arg or 4.18
Step 1: Provision the cluster
bash .claude/scripts/cluster-provision.sh up "$KUBECONFIG_PATH" "$OCP_VERSION"
The script handles everything: triggering the Gangway API, polling for readiness, downloading the kubeconfig, and validating connectivity. Wait for the === Cluster Ready === output.
Step 2: Verify
oc --kubeconfig="$KUBECONFIG_PATH" whoami
oc --kubeconfig="$KUBECONFIG_PATH" get nodes
Step 3: Report
Summarize connection status and remind user:
- The cluster is ready
- Kubeconfig path
- Cluster server URL
- Cluster auto-expires in ~4 hours
- Prow job URL for reference
Status
bash .claude/scripts/cluster-provision.sh status
Teardown
bash .claude/scripts/cluster-provision.sh down
Troubleshooting
| Error | Fix |
|---|
GANGWAY_TOKEN is not set | Run: oc login https://api.ci.l2s4.p1.openshiftapps.com:6443 --web && export GANGWAY_TOKEN=$(oc whoami -t) |
HTTP 401 | Token expired — re-authenticate with oc login |
KUBECONFIG_ENCRYPTION_KEY is not set | Contact the CI team for the decryption passphrase |
| Timeout after 40 min | Pool exhausted or job stuck — check the Prow job URL |
