Chapter 4. Creating Data Grid clusters

Create Data Grid clusters running on OpenShift with the Infinispan CR or with the native Data Grid CLI plugin for oc clients.

4.1. Infinispan custom resource (CR)

Data Grid Operator adds a new Custom Resource (CR) of type Infinispan that lets you handle Data Grid clusters as complex units on OpenShift.

Data Grid Operator listens for Infinispan Custom Resources (CR) that you use to instantiate and configure Data Grid clusters and manage OpenShift resources, such as StatefulSets and Services.

Infinispan CR

apiVersion: infinispan.org/v1
kind: Infinispan
metadata:
  name: infinispan
spec:
  replicas: 2
  service:
    type: DataGrid

FieldDescription

apiVersion

Declares the version of the Infinispan API.

kind

Declares the Infinispan CR.

metadata.name

Specifies a name for your Data Grid cluster.

spec.replicas

Specifies the number of pods in your Data Grid cluster.

spec.service.type

Specifies the type of Data Grid service to create.

4.2. Creating Data Grid clusters

Create Data Grid clusters with the native CLI plugin, kubectl-infinispan.

Prerequisites

  • Install Data Grid Operator.
  • Have kubectl-infinispan on your PATH.

Procedure

  1. Run the infinispan create cluster command.

    For example, create a Data Grid cluster with two pods as follows:

    oc infinispan create cluster --replicas=3 -Pservice.type=DataGrid infinispan
  2. Watch Data Grid Operator create the Data Grid pods.

    oc get pods -w

Next steps

After you create a Data Grid cluster, use the oc to apply changes to Infinispan CR and configure your Data Grid service.

You can also delete Data Grid clusters with kubectl-infinispan and re-create them as required.

oc infinispan delete cluster infinispan

4.3. Verifying Data Grid cluster views

Confirm that Data Grid pods have successfully formed clusters.

Prerequisites

  • Create at least one Data Grid cluster.

Procedure

  • Retrieve the Infinispan CR for Data Grid Operator.

    oc get infinispan -o yaml

    The response indicates that Data Grid pods have received clustered views, as in the following example:

    conditions:
      - message: 'View: [infinispan-0, infinispan-1]'
        status: "True"
        type: wellFormed
Tip

Do the following for automated scripts:

oc wait --for condition=wellFormed --timeout=240s infinispan/infinispan

Retrieving cluster view from logs

You can also get the cluster view from Data Grid logs as follows:

oc logs infinispan-0 | grep ISPN000094
INFO  [org.infinispan.CLUSTER] (MSC service thread 1-2) \
ISPN000094: Received new cluster view for channel infinispan: \
[infinispan-0|0] (1) [infinispan-0]

INFO  [org.infinispan.CLUSTER] (jgroups-3,infinispan-0) \
ISPN000094: Received new cluster view for channel infinispan: \
[infinispan-0|1] (2) [infinispan-0, infinispan-1]

4.4. Modifying Data Grid clusters

Configure Data Grid clusters by providing Data Grid Operator with a custom Infinispan CR.

Prerequisites

  • Install Data Grid Operator.
  • Create at least one Data Grid cluster.
  • Have an oc client.

Procedure

  1. Create a YAML file that defines your Infinispan CR.

    For example, create a my_infinispan.yaml file that changes the number of Data Grid pods to two:

    cat > cr_minimal.yaml<<EOF
    apiVersion: infinispan.org/v1
    kind: Infinispan
    metadata:
      name: infinispan
    spec:
      replicas: 2
      service:
        type: DataGrid
    EOF
  2. Apply your Infinispan CR.

    oc apply -f my_infinispan.yaml
  3. Watch Data Grid Operator scale the Data Grid pods.

    oc get pods -w

4.5. Stopping and starting Data Grid clusters

Stop and start Data Grid pods in a graceful, ordered fashion to correctly preserve cluster state.

Clusters of Data Grid service pods must restart with the same number of pods that existed before shutdown. This allows Data Grid to restore the distribution of data across the cluster. After Data Grid Operator fully restarts the cluster you can safely add and remove pods.

Procedure

  1. Change the spec.replicas field to 0 to stop the Data Grid cluster.

    spec:
      replicas: 0
  2. Ensure you have the correct number of pods before you restart the cluster.

    oc get infinispan infinispan -o=jsonpath='{.status.replicasWantedAtRestart}'
  3. Change the spec.replicas field to the same number of pods to restart the Data Grid cluster.

    spec:
      replicas: 6