Keda Manager reconciles KEDA deployment based on the watched Keda custom resources (CRs):
-
Apply Keda CR (sample) to have KEDA installed.
kubectl apply -f https://github.com/kyma-project/keda-manager/releases/latest/download/keda-default-cr.yaml
After a while, you have KEDA installed, and you see its workloads:
NAME READY UP-TO-DATE AVAILABLE AGE keda-manager 1/1 1 1 3m keda-operator 1/1 1 1 3m keda-operator-metrics-apiserver 1/1 1 1 3m
Now you can use KEDA to scale workloads in the Kubernetes cluster. Check the demo application.
-
Remove Keda CR to have KEDA uninstalled.
kubectl delete -f config/samples/keda-default-cr.yaml
This uninstalls all KEDA workloads but leaves Keda Manager.
NOTE: Keda Manager uses finalizers to uninstall the Keda module from the cluster. It means that Keda Manager blocks the uninstallation process of KEDA until there are user-created CRs (for example, ScaledObjects).
-
Update the specification of Keda CR to change the Keda installation
The configuration example shows how to modify the Keda properties using the
keda.operator.kyma-project.io
CR.cat <<EOF | kubectl apply -f - apiVersion: operator.kyma-project.io/v1alpha1 kind: Keda metadata: name: default spec: logging: operator: level: "debug" resources: operator: limits: cpu: "1" memory: "200Mi" requests: cpu: "0.5" memory: "150Mi" metricServer: limits: cpu: "1" memory: "1000Mi" requests: cpu: "300m" memory: "500Mi" EOF