machine-health-check.md

Machine Health Check

This controller responsible to monitor nodes and create the MachineRemediation object when the node has unhealthy condition, when unhealthy condition can be definied by a user.

How to define MachineHealthCheck

apiVersion: machineremediation.kubevirt.io/v1alpha1
kind: MachineHealthCheck
metadata:
  name: workers
  namespace: openshift-machine-api
spec:
  selector:
    matchLabels:
      machine.openshift.io/cluster-api-machine-role: worker
      machine.openshift.io/cluster-api-machine-type: worker

This health check will start to monitor all nodes with labels under the matchLabels.

How to monitor custom node conditions

By default, the machine health check controller recognize only NotReady condition and will remove unhealthy machine after 5 minutes. If you want to customize unhealthy conditions you can create node-unhealthy-conditions config map, for example:

apiVersion: v1
kind: ConfigMap
metadata:
  name: node-unhealthy-conditions
  namespace: openshift-machine-api
data:
  conditions: |
    items:
    - name: NetworkUnavailable
      timeout: 5s
      status: True