Basic Usage

Verify installation¶

kubectl get pods -n cruisekube-system

Expect cruisekube-controller-manager-*, cruisekube-webhook-server-*, and (if enabled) cruisekube-frontend-* in Running state.

Access the dashboard¶

kubectl port-forward -n cruisekube-system svc/cruisekube-frontend 3000:3000

Open http://localhost:3000. See Dashboard for UI concepts.

Apply recommendations (Recommend vs Cruise)¶

Optimization is controlled per workload in the UI—there is no separate global “dry-run mode” toggle.

1. Open Workloads → CruiseKube mode & priority (see Dashboard).
2. Leave workloads on Recommend while you validate suggestions against metrics and SLOs.
3. Switch trusted workloads to Cruise when you want the controller and webhook to apply right-sizing.

CruiseKube doesn't apply recommendations unless cruise mode is enabled. In Cruise mode, Memory recommendation application remains enabled by default. To disable memory recommendation application, set the following disable flags to true:

• cruisekubeController.env.CRUISEKUBE_RECOMMENDATIONSETTINGS_DISABLEMEMORYAPPLICATION=true

helm upgrade --install cruisekube oci://tfy.jfrog.io/tfy-helm/cruisekube --namespace cruisekube-system --create-namespace \
--set cruisekubeController.env.CRUISEKUBE_DEPENDENCIES_INCLUSTER_PROMETHEUSURL="http://prometheus-kube-prometheus-prometheus.monitoring.svc:9090" \
--set cruisekubeController.env.CRUISEKUBE_RECOMMENDATIONSETTINGS_DISABLEMEMORYAPPLICATION=true

You can check all the available environment variables in the values.yaml file.

Optional Helm knobs (for example disabling memory application while tuning CPU) still exist on the controller and webhook—see charts/cruisekube/values.yaml and Configuration. Align changes with Policies & modes before wide rollout.