Multi Cluster Observability AddOn

Description

The multicluster-observability-addon is a pluggable addon working on OCM based on the extensibility provided by addon-framework which automates the collection and forwarding of observability signals to central stores.

This is achieved through the installation of the spoke clusters of dedicated operators for each observability signal:

• For Metrics it's required that the multicluster-observability-operator is installed. Once enalbed the addon will create the necessary resources to configure metrics collection.

• For Logs the operator installed will be cluster-logging-operator. The addon will also configure an instance of ClusterLogForwarder to forward logs to a configured store.

• For Traces the operator installed will be Red Hat build of OpenTelemetry. The addon will also configure an instance of OpenTelemetryCollector to forward traces to a configued store.

The logging-ocm-addon consists of one component:

• Addon-Manager: Not only manages the installation of the AddOn on spoke clusters. But also builds the manifests that will be deployed to the spoke clusters.

Getting started

Prerequisite

• OCM registration (>= 0.5.0)

Steps

Installing via Kustomize

1. Install the AddOn using Kustomize

   make install-crds
   kubectl apply -k deploy/

2. The addon should now be installed in you hub cluster

   kubectl get ClusterManagementAddOn multicluster-observability-addon

3. The addon will install automatically in spoke clusters once the resources referenced in ClusterManagementAddOn are created.

Installing via MCO

In 2.12, multicluster-observability-operator has the ability to install MCOA using the capabilities field.

1. Create a MultiClusterObservability resource and configure capabilities

   apiVersion: observability.open-cluster-management.io/v1beta2
   kind: MultiClusterObservability
   metadata:
     name: observability
   spec:
     capabilities:
       platform:
         logs:
           collection:
             enabled: true
       userWorkloads:
         logs:
           collection:
             clusterLogForwarder:
               enabled: true
         traces:
           collection:
             instrumentation:
               enabled: true
             openTelemetryCollector:
               enabled: false
     observabilityAddonSpec: {}
     storageConfig:
       metricObjectStorage:
         name: thanos-object-storage
         key: thanos.yaml

   Note: Deploy a custom image by adding the annotation: mco-multicluster_observability_addon-image: quay.io/YOUR_ORG_HERE/multicluster-observability-addon:YOUR_TAG_HERE

2. The addon should now be installed in you hub cluster

   kubectl get ClusterManagementAddOn multicluster-observability-addon

3. The addon will install automatically in spoke clusters once the resources referenced in ClusterManagementAddOn are created.

Default configurations references

apiVersion: addon.open-cluster-management.io/v1alpha1
kind: ClusterManagementAddOn
spec:
  installStrategy:
    type: Placements
    placements:
      - name: <placement_name> # Use global for selecting all clusters
        namespace: open-cluster-management-global-set
        configs:
          - group: observability.openshift.io
            resource: clusterlogforwarders
            name: instance
            namespace: open-cluster-management-observability
          - group: opentelemetry.io
            resource: opentelemetrycollectors
            name: instance
            namespace: open-cluster-management-observability
          - group: opentelemetry.io
            resource: instrumentations
            name: instance
            namespace: open-cluster-management-observability
          # Default metrics forwarding configuration for the ACM platform metrics collector
          - group: monitoring.coreos.com
            resource: prometheusagents
            name: acm-platform-metrics-collector-default
            namespace: open-cluster-management-observability
          - group: monitoring.coreos.com
            resource: scrapeconfigs
            name: platform-metrics-default
            namespace: open-cluster-management-observability
          - group: monitoring.coreos.com
            resource: prometheusrules
            name: platform-rules-default
            namespace: open-cluster-management-observability
          # Default metrics forwarding configuration for the ACM user workload metrics collector
          # There are no default configurations for the scrapeConfigs and prometheusRules
          - group: monitoring.coreos.com
            resource: prometheusagents
            name: acm-user-workload-metrics-collector-default
            namespace: open-cluster-management-observability

References

• Open-Cluster-Management: https://github.com/open-cluster-management-io/ocm
• Addon-Framework: https://github.com/open-cluster-management-io/addon-framework
• Multicluster-Observability-Operator: https://github.com/stolostron/multicluster-observability-operator
• Cluster-Logging-Operator: https://github.com/openshift/cluster-logging-operator
• OpenTelemetry-Operator: https://github.com/open-telemetry/opentelemetry-operator