endpoint-metrics-operator

README

endpoint-monitoring-operator

Overview

The endpoint-monitoring-operator is a component of ACM observability feature. It is designed to install into Spoke Cluster.

Developer Guide

The guide is used for developer to build and install the endpoint-monitoring-operator . It can be running in [kind][install_kind] if you don't have a OCP environment.

Prerequisites

• git
• go version v1.15+
• docker version 17.03+
• kubectl version v1.16.3+
• kustomize version v3.8.5+
• operator-sdk version v1.4.2+
• access to a Kubernetes v1.16.0+ cluster

Build the Operator

1. Check out the endpoint-metrics-operator repository.

$ git clone git@github.com:stolostron/endpoint-metrics-operator.git

1. Build the endpoint-metrics-operator image and push it to a public registry, such as quay.io:

$ make -f Makefile.prow docker-build docker-push IMG=quay.io/<YOUR_USERNAME_IN_QUAY>/endpoint-metrics-operator:latest

Deploy this Operator

1. Create the open-cluster-management-addon-observability namespace if it doesn't exist:

$ kubectl create ns open-cluster-management-addon-observability

1. Create the secret named hub-kube-config in namespace open-cluster-management-addon-observability about the hub cluster information:

$ cat << EOF | kubectl apply -n open-cluster-management-addon-observability -f -
kind: Secret
apiVersion: v1
metadata:
  name: hub-kube-config
type: Opaque
data:
    kubeconfig: ***
EOF

Note: the content of kubeconfig is base64-encoded content of kubeconfig for the hub cluster.

1. Create the secret named hub-info-secret in namespace open-cluster-management-addon-observability about the hub cluster information:

$ cat << EOF | kubectl apply -n open-cluster-management-addon-observability -f -
kind: Secret
apiVersion: v1
metadata:
  name: hub-info-secret
type: Opaque
data:
    clusterName: ***
    hub-info.yaml: ***
EOF

Note: the content of clusterName is base64-encoded yaml of the hub cluster name, while the content of hub-info.yaml is base64-encoded yaml that contains the observatorium api gateway URL, hub alertmanager URl and hub router CA which is exposed on the hub cluster. The original yaml content resembles below:

endpoint: "http://observatorium-api-open-cluster-management-observability.apps.stage3.demo.red-chesterfield.com/api/v1/receive"
hub-alertmanager-endpoint: "https://alertmanager-open-cluster-management-observability.apps.stage3.demo.red-chesterfield.com"
hub-alertmanager-router-ca: |
-----BEGIN CERTIFICATE-----
xxxxxxxxxxxxxxxxxxxxxxxxxxx
-----END CERTIFICATE-----

1. Create the configmap named observability-metrics-allowlist in namespace open-cluster-management-addon-observability:

$ kubectl apply -n open-cluster-management-addon-observability -f https://raw.githubusercontent.com/stolostron/multicluster-observability-operator/main/manifests/base/config/metrics_allowlist.yaml

1. Update the value of environment variable COLLECTOR_IMAGE in the endpoint-metrics-operator deployment, for example: quay.io/stolostron/metrics-collector:2.3.0-SNAPSHOT-2021-04-08-09-07-10

$ sed -i 's~REPLACE_WITH_METRICS_COLLECTOR_IMAGE~quay.io/stolostron/metrics-collector:2.3.0-SNAPSHOT-2021-04-08-09-07-10~g' config/manager/manager.yaml

1. Update the value of environment variable HUB_NAMESPACE with the actual hub namespace, for example: cluster1

$ sed -i 's~REPLACE_WITH_HUB_NAMESPACE~cluster1~g' config/manager/manager.yaml

1. Replace the operator image and deploy the endpoint-metrics-operator:

$ make -f Makefile.prow deploy IMG=quay.io/<YOUR_USERNAME_IN_QUAY>/endpoint-metrics-operator:latest

1. Deploy the endpoint-metrics-operator CR:

$ kubectl -n open-cluster-management-addon-observability apply -f config/samples/observability.open-cluster-management.io_v1beta1_observabilityaddon.yaml

Verify the Installation

After installed successfully, you will see the following pod are running:

# kubectl -n open-cluster-management-addon-observability get pod
NAME                                               READY   STATUS    RESTARTS   AGE
endpoint-observability-operator-7cf545f45c-cfjlk   1/1     Running   0          136m
metrics-collector-deployment-6dc9998cb-f2wd7       1/1     Running   0          136m

You should also see the CR created in the cluster:

# kubectl -n open-cluster-management-addon-observability get observabilityaddon
NAME                  AGE
observability-addon   137m

Notice: To deploy the observabilityaddon CR in local managed cluster just for dev/test purpose. In real topology, the observabilityaddon CR will be created in hub cluster, the endpoint-monitoring-operator should talk to api server of hub cluster to watch those CRs, and then perform changes on managed cluster.

View metrics in dashboard

Access Grafana console in hub cluster at https://{YOUR_DOMAIN}/grafana, view the metrics in the dashboard named "ACM:Managed Cluster Monitoring"

Rebuild Image: Tue May 17 14:38:14 EDT 2022