• configure-alertmanager-operator
  • Summary
  • Cluster Readiness
  • Metrics
  • Alerts
  • Testing
    • Building
    • Deploying
      • Prevent Overwrites
      • Replace the Image

The Configure Alertmanager Operator was created for the OpenShift Dedicated platform to dynamically manage Alertmanager configurations based on the presence or absence of secrets containing a GoAlert URLs, Pager Duty RoutingKey, and Dead Man's Snitch URL. When the secret is created/updated/deleted, the associated Receiver and Route will be created/updated/deleted within the Alertmanager config.

The operator contains the following components:

• Secret controller: watches the openshift-monitoring namespace for any changes to relevant Secrets or ConfigMaps that are used in the configuration of Alertmanager. For more information on this see Secret Controller below.

• Types library: these types are imported from the Alertmanager Config library and pared down to suit our config needs. (Since their library is intended for internal use only).

The Secret Controller watches over the resources in the table below. Changes to these resources will prompt the controller to reconcile.

To avoid alert noise while a cluster is in the early stages of being installed and configured, this operator waits to configure Pager Duty -- effectively silencing alerts -- until a predetermined set of health checks, performed by osd-cluster-ready, has completed.

This determination is made through the presence of a completed Job named osd-cluster-ready in the openshift-monitoring namespace.

The Configure Alertmanager Operator exposes the following Prometheus metrics:

The operator creates a Service and ServiceMonitor named configure-alertmanager-operator to expose these metrics to Prometheus.

The following alerts are added to Prometheus as part of configure-alertmanager-operator:

• Mismatch between DMS secret and DMS Alertmanager config.
• Mismatch between GoAlert secret and GoAlert Alertmanager config.
• Mismatch between PD secret and PD Alertmanager config.
• Alertmanager config secret does not exist.

Tips for testing on a personal cluster:

You may build (make docker-build) and push (make docker-push) the operator image to a personal repository by overriding components of the image URI:

• IMAGE_REGISTRY overrides the registry (default: quay.io)
• IMAGE_REPOSITORY overrides the organization (default: app-sre)
• IMAGE_NAME overrides the repository name (default: managed-cluster-validating-webhooks)
• OPERATOR_IMAGE_TAG overrides the image tag. (By default this is generated based on the current commit of your local clone of the git repository; but make docker-build will also always tag latest)

For example, to build, tag, and push quay.io/my-user/configure-alertmanager-operator:latest, you can run:

make IMAGE_REPOSITORY=my-user docker-build docker-push

Note: This step requires elevated permissions

This operator is managed by OLM, so you must switch that off, or your changes to the operator's Deployment will be overwritten:

oc scale deploy/cluster-version-operator --replicas=0 -n openshift-cluster-version
oc scale deploy/olm-operator --replicas=0 -n openshift-operator-lifecycle-manager

NOTE: Don't forget to revert these changes when you have finished testing:

oc scale deploy/olm-operator --replicas=1 -n openshift-operator-lifecycle-manager
oc scale deploy/cluster-version-operator --replicas=1 -n openshift-cluster-version

Edit the operator's deployment (oc set image deployment configure-alertmanager-operator -n openshift-monitoring *=<IMAGE>), replacing the image: with the URI of the image you built above. The deployment will automatically delete and replace the running pod.

NOTE: If you are testing coordination with the osd-cluster-ready job, you may need to set the MAX_CLUSTER_AGE_MINUTES environment variable in the deployment's configure-alertmanager-operator container definition. For example, to ensure the osd-cluster-ready Job is checked in a cluster less than 1048576 minutes (~two years) old:

        containers:
        - command:
          - configure-alertmanager-operator
          env:
          - name: WATCH_NAMESPACE
            valueFrom:
              fieldRef:
                apiVersion: v1
                fieldPath: metadata.namespace
          - name: POD_NAME
            valueFrom:
              fieldRef:
                apiVersion: v1
                fieldPath: metadata.name
          - name: OPERATOR_NAME
            value: configure-alertmanager-operator
          ### Add this entry ###
          - name: MAX_CLUSTER_AGE_MINUTES
            value: "1048576"
          image: quay.io/2uasimojo/configure-alertmanager-operator:latest
          imagePullPolicy: Always
          name: configure-alertmanager-operator
          resources: {}
          terminationMessagePath: /dev/termination-log
          terminationMessagePolicy: File