The vertical-pod-autoscaler-operator manages deployments and configurations of the OpenShift Vertical Pod Autoscaler's three controllers. The three controllers are:
- Recommender, which monitors the current and past resource consumption and provides recommended values for containers' CPU and memory requests.
- Admission Plugin, which sets the correct resource requests on new pods using data from the Recommender. The recommended request values will be applied to new pods which are being restarted (after an eviction by the Updater) or by any other pod restart.
- Updater, which checks which of the managed pods have incorrect resources set, and evicts any it finds so that the pods can be recreated by their controllers with the updated resource requests.
The operator manages the following custom resource:
-
VerticalPodAutoscalerController: This is a singleton resource which controls the configuration of the cluster's VPA 3 controller instances. The operator will only respond to the VerticalPodAutoscalerController resource named "default" in the managed namespace, i.e. the value of the
WATCH_NAMESPACE
environment variable. (Example)Many of fields in the spec for VerticalPodAutoscalerController resources correspond to command-line arguments of the three VPA controllers and also control which controllers should be run. The example linked above results in the following invocation:
Command: recommender Args: --safetyMarginFraction=0.15 --podMinCPUMillicores=25 --podMinMemoryMb=250 Command: admission-plugin Command: updater
## Build, Test, & Run
$ make build
$ make test
$ export WATCH_NAMESPACE=openshift-vertical-pod-autoscaler
$ ./bin/vertical-pod-autoscaler-operator -alsologtostderr
The Vertical Pod Autoscaler Operator is designed to be deployed on
OpenShift by the Cluster Version Operator, but it's possible to
run it directly on any vanilla Kubernetes cluster.
To do so, apply the manifests in the install/deploy
directory:
kubectl apply -f ./install/deploy
This will create the openshift-vertical-pod-autoscaler
namespace, register the
custom resource definitions, configure RBAC policies, and create a
deployment for the operator.
You can run the e2e test suite with make test-e2e
. These tests
assume the presence of a cluster not already running the operator, and
that the KUBECONFIG
environment variable points to a configuration
granting admin rights on said cluster.