Verrazzano Enterprise Container Platform – Install Multicluster Verrazzano

Docs: Install Multicluster

Mon, 01 Jan 0001 00:00:00 +0000

Prerequisites

Before you begin, read this document, Verrazzano in a multicluster environment.
To set up a multicluster Verrazzano environment, you will need two or more Kubernetes clusters. One of these clusters will be admin cluster; the others will be managed clusters. For instructions on preparing Kubernetes platforms for installing Verrazzano, see Platform Setup.

NOTE

If Rancher is not enabled, then refer to Verrazzano multicluster installation without Rancher because additional steps are required to register a managed cluster.

The following instructions assume an admin cluster and a single managed cluster. For each additional managed cluster, simply repeat the managed cluster instructions.

Install Verrazzano

To install Verrazzano on each Kubernetes cluster, complete the following steps:

On one cluster, install Verrazzano using the dev or prod profile; this will be the admin cluster.
On the other cluster, install Verrazzano using the managed-cluster profile; this will be a managed cluster. The managed-cluster profile contains only the components that are required for a managed cluster.
NOTE: You also can use the dev or prod profile.

For detailed instructions on how to install and customize Verrazzano on a Kubernetes cluster using a specific profile, see the Installation Guide and Installation Profiles.

Register managed clusters using the console

To register a cluster, complete the following steps:

Enable syncClusters.
For information about syncClusters, see Enable syncClusters.
```
kubectl apply -f - <<EOF
apiVersion: install.verrazzano.io/v1beta1
kind: Verrazzano
metadata:
  name: admin
spec:
  profile: prod
  components:
    clusterOperator:
      overrides:
      - values:
          syncClusters:
            enabled: true
EOF
```
On the admin cluster, open the Rancher console.
You can find the Rancher console URL for your cluster by following the instructions here.
Select Cluster Management, and then click Import Existing Cluster.
Provide a name for your managed cluster, for example, managed1.
Optional. In your Verrazzano configuration, if you specified a cluster label selector, then under Labels & Annotations provide the label and value for the cluster. The label and value information must match the cluster selection matchExpression in your Verrazzano configuration for selective cluster registration to occur.
Click Create.
On the next screen, follow the on-screen instructions to complete the registration by running the provided command against the managed cluster.

After the cluster reaches the Active state in the console, synchronization with Verrazzano will happen automatically and a VerrazzanoManagedCluster resource will be created in the verrazzano-mc namespace.

Run the following command to view the details and status of Verrazzano’s multicluster initialization operations:

$ kubectl get vmc -n verrazzano-mc <Rancher_cluster_name> -o yaml

For more information, see Registering Existing Clusters in the Rancher documentation.

NOTE: You can also register managed clusters using kubectl, see Register Managed Clusters using kubectl.

Next steps

Verify your multicluster Verrazzano environment set up by following the instructions at Verify Multicluster Installation.
Deploy multicluster example applications. See Examples of using Verrazzano in a multicluster environment.

NOTE

To deregister a managed cluster, see Deregister a Managed Cluster.

Docs: Verify Multicluster Installation

Mon, 01 Jan 0001 00:00:00 +0000

Verify that managed cluster registration has completed

You can perform all the verification steps on the admin cluster.

Verify that the managed cluster can connect to the admin cluster. View the status of the VerrazzanoManagedCluster resource on the admin cluster, and check whether the lastAgentConnectTime, prometheusHost, and apiUrl fields are populated. This may take up to two minutes after completing the registration steps.
```
# On the admin cluster
$ kubectl --kubeconfig $KUBECONFIG_ADMIN --context $KUBECONTEXT_ADMIN \
    get vmc managed1 -n verrazzano-mc -o yaml

# Sample output showing the status field
spec:
  ....
  ....
status:
  apiUrl: https://verrazzano.default.172.18.0.211.nip.io
  conditions:
  - lastTransitionTime: "2021-07-07T15:49:43Z"
    message: Ready
    status: "True"
    type: Ready
  lastAgentConnectTime: "2021-07-16T14:47:25Z"
  prometheusHost: prometheus.vmi.system.default.172.18.0.211.nip.io
```
Verify that the managed cluster is successfully registered with Rancher. When you perform the registration steps, Verrazzano also registers the managed cluster with Rancher. View the Rancher console on the admin cluster. If the registration with Rancher was successful, then your cluster will be listed in Rancher’s list of clusters, and will be in the Active state. You can find the Rancher console URL for your cluster by following the instructions for Accessing Verrazzano.

Verify that managed cluster metrics are being collected

Verify that the admin cluster is collecting metrics from the managed cluster. The Prometheus output will include records that contain the name of the Verrazzano cluster (labeled as verrazzano_cluster).

You can find the Prometheus console URL for your cluster by following the instructions for Accessing Verrazzano. Run a query for a metric (for example, node_disk_io_time_seconds_total).

Sample output of a Prometheus query

An alternative approach to using the Prometheus console is to query metrics from the command line. Here is an example of how to obtain Prometheus metrics from the command line. Search the output of the query for responses that have the verrazzano_cluster field set to the name of the managed cluster.

# On the admin cluster
$ prometheusUrl=$(kubectl --kubeconfig $KUBECONFIG_ADMIN --context $KUBECONTEXT_ADMIN \
                 get verrazzano -o jsonpath='{.items[0].status.instance.prometheusUrl}')
$ VZPASS=$(kubectl --kubeconfig $KUBECONFIG_ADMIN --context $KUBECONTEXT_ADMIN \
           get secret verrazzano --namespace verrazzano-system \
           -o jsonpath={.data.password} | base64 --decode; echo)
$ curl -k --user verrazzano:${VZPASS} "${prometheusUrl}/api/v1/query?query=node_disk_io_time_seconds_total"

Verify that managed cluster logs are being collected

Verify that the admin cluster is collecting logs from the managed cluster. The output will include records which have the name of the managed cluster in the cluster_name field.

You can find the OpenSearch Dashboards URL for your cluster by following the instructions for Accessing Verrazzano. Searching the verrazzano-system data stream for log records with the cluster_name set to the managed cluster name yields logs for the managed cluster.

Sample output of a OpenSearch Dashboards screen

An alternative approach to using the OpenSearch Dashboards is to query OpenSearch from the command line. Here is an example of how to obtain log records from the command line. Search the output of the query for responses that have the cluster_name field set to the name of the managed cluster.

# On the admin cluster
$ OS_URL=$(kubectl --kubeconfig $KUBECONFIG_ADMIN --context $KUBECONTEXT_ADMIN \
                 get verrazzano -o jsonpath='{.items[0].status.instance.openSearchUrl}')
$ VZPASS=$(kubectl --kubeconfig $KUBECONFIG_ADMIN --context $KUBECONTEXT_ADMIN \
           get secret verrazzano --namespace verrazzano-system \
           -o jsonpath={.data.password} | base64 --decode; echo)
$ curl -k --user verrazzano:${VZPASS} -X POST -H 'kbn-xsrf: true' "${OS_URL}/verrazzano-system/_search?size=25"

Docs: Deregister a Managed Cluster

Mon, 01 Jan 0001 00:00:00 +0000

NOTE: The following procedure is for a cluster in which Rancher is enabled on the admin cluster. If Rancher is not enabled, then additional steps are required to deregister a managed cluster, see Deregister a managed cluster without Rancher.

If you want to deregister a managed cluster because you no longer want it to be part of a Verrazzano multicluster environment, then log in to the Rancher console and delete the managed cluster. To delete a cluster in Rancher, see What if I don’t want my registered cluster managed by Rancher? This results in the cluster being deregistered from Verrazzano. The associated VerrazzanoManagedCluster resource will be automatically deleted, and, if present, then the Argo CD registration of the managed cluster also will be removed.

Alternatively, you can deregister a managed cluster by deleting the VerrazzanoManagedCluster resource. This will result in the automatic cleanup of the Rancher cluster, as well as the Argo CD registration, if it is present.

# On the admin cluster
$ kubectl --kubeconfig $KUBECONFIG_ADMIN --context $KUBECONTEXT_ADMIN \
    delete vmc -n verrazzano-mc managed1

NOTE: Even after deregistration, any applications that you deployed previously to the managed cluster will continue running on that cluster.

Docs: Advanced Features

Mon, 01 Jan 0001 00:00:00 +0000