Installation¶
Prerequisites¶
- Kubernetes cluster v1.26+
- kubectl v1.26+
- metrics-server installed (Hybernate reads pod CPU/memory via the Kubernetes Metrics API)
- Helm v3 (if using Helm install)
Install¶
Install directly from the OCI registry:
helm install hybernate oci://ghcr.io/okedeji/charts/hybernate \
--version v0.1.6 \
--namespace hybernate-system \
--create-namespace
Helm Values
| Value | Default | Description |
|---|---|---|
replicaCount |
1 |
Number of operator replicas |
image.repository |
ghcr.io/okedeji/hybernate |
Container image |
image.tag |
latest |
Image tag |
leaderElection.enabled |
true |
Enable HA leader election |
metrics.secure |
true |
Serve metrics over HTTPS |
resources.limits.cpu |
500m |
CPU limit |
resources.limits.memory |
128Mi |
Memory limit |
Apply the all-in-one installer manifest directly:
This installs the CRDs, RBAC, and the operator Deployment into the hybernate-system namespace.
Verify Installation¶
# Check the operator is running
kubectl get pods -n hybernate-system
# Verify CRDs are installed
kubectl get crd managedworkloads.hybernate.io
kubectl get crd workloadpolicies.hybernate.io
kubectl get crd hybernatereports.hybernate.io
Uninstall¶
# Remove all CRs first
kubectl delete managedworkloads --all --all-namespaces
kubectl delete workloadpolicies --all --all-namespaces
# Remove the operator
make undeploy
# Remove CRDs
make uninstall
Warning
Deleting CRDs removes all ManagedWorkload, WorkloadPolicy, and HybernateReport resources from the cluster. Workloads that were paused (scaled to zero) will remain at zero replicas. You must manually restore them.
Next Steps¶
Follow the Quickstart to manage your first workload.