Install ScyllaDB Operator in Red Hat OpenShift¶

This procedure describes how to install and subscribe to ScyllaDB Operator from software catalog by using the OpenShift Container Platform web console.

Procedure

1. Navigate to Ecosystem -> Software Catalog.

2. Search for ScyllaDB Operator and select the Certified version by setting the Source filter to Certified or ensuring that the ScyllaDB Operator tile has the Certified tag next to it.

3. In the description dialog, read the information about the operator and click Install.

4. In the Install Operator dialog, configure your ScyllaDB Operator installation:

   • For clusters on cloud providers with token-based authentication:

     • AWS Security Token Service (STS Mode in the web console): enter the Amazon Resource Name (ARN) of the AWS IAM role of your service account in the role ARN field.

   • Select the stable Update Channel and version 1.20.0 from the list.

   • Select the All namespaces on the cluster installation mode.

   • Select the Operator recommended installed namespace: scylla-operator.

     Warning

     ScyllaDB Operator must run in a reserved scylla-operator namespace. It is not currently possible to run it in a different namespace.

   • Select the Manual update approval strategy to manually approve ScyllaDB Operator upgrades when new versions are available.

     Caution

     Always review the Upgrading ScyllaDB Operator guide and check for additional steps for specific versions before starting the upgrade process.

5. Click Install to install and subscribe to ScyllaDB Operator.

6. In the Install Plan dialog, review the manual install plan and click Approve to approve it.

7. Log in to the OpenShift cluster in the terminal. Ensure that kubectl is configured to communicate with your OpenShift cluster.

This procedure describes how to install and subscribe to ScyllaDB Operator from software catalog by using the CLI.

Procedure

1. Log in to the OpenShift cluster in the terminal. Ensure that kubectl is configured to communicate with your OpenShift cluster.

2. Ensure that the ScyllaDB Operator package is available:

   kubectl get -n=openshift-marketplace packagemanifest scylladb-operator
   

   Expected output:

   NAME                CATALOG               AGE
   scylladb-operator   Certified Operators   2d22h
   

3. Create the scylla-operator namespace:

   Warning

   ScyllaDB Operator must run in a reserved scylla-operator namespace. It is not currently possible to run it in a different namespace.

   kubectl create namespace scylla-operator
   

4. Create an OperatorGroup object in the scylla-operator namespace:

   kubectl apply --server-side -n=scylla-operator -f=- <<EOF
   apiVersion: operators.coreos.com/v1
   kind: OperatorGroup
   metadata:
     name: scylladb-operator
     namespace: scylla-operator
   EOF
   

5. Create a Subscription object to install and subscribe to ScyllaDB Operator:

   Clusters on cloud providers with token-based authentication require you to include relevant cloud-provider specific fields in the Subscription object’s config section. Apply the appropriate manifest based on your environment. In the below table, manifests are provided for AWS STS mode and generic installations.

   AWS STSGeneric

   If the cluster is in AWS STS mode, include the role ARN details in the Subscription manifest.

   kubectl apply --server-side -n=scylla-operator -f=- <<EOF
   apiVersion: operators.coreos.com/v1alpha1
   kind: Subscription
   metadata:
     name: scylladb-operator
     namespace: scylla-operator
   spec:
     channel: stable
     installPlanApproval: Manual
     name: scylladb-operator
     source: certified-operators
     sourceNamespace: openshift-marketplace
     startingCSV: scylladb-operator.v1.20.0
     config:
       env:
       - name: ROLEARN
         value: "<role_arn>"
   EOF
   

6. Review and approve the generated InstallPlan to start the installation:

   • Locate the generated InstallPlan object and note its name:

     kubectl -n=scylla-operator get installplan -l=operators.coreos.com/scylladb-operator.scylla-operator=""
     

     Example expected output:

     NAME            CSV                         APPROVAL   APPROVED
     install-tw6bv   scylladb-operator.v1.20.0   Manual     true
     

   • Describe the InstallPlan object and review the changes that will be applied:

     kubectl -n=scylla-operator describe installplan <install-plan-name>
     

   • Approve the reviewed InstallPlan to start the installation:

     kubectl -n=scylla-operator patch installplan --type=merge -p='{"spec":{"approved":true}}' <install-plan-name>
     

     Example expected output:

     installplan.operators.coreos.com/install-tw6bv patched
     

7. Wait for the ScyllaDB Operator to be installed:

   Wait for ClusterServiceVersion to be created:

   kubectl -n=scylla-operator wait --for=create --timeout=10m csv/scylladb-operator.v1.20.0 
   

   Expected output:

   clusterserviceversion.operators.coreos.com/scylladb-operator.v1.20.0 condition met
   

   Wait for ClusterServiceVersion to reach Succeeded phase:

   kubectl wait -n=scylla-operator --timeout=5m --for=jsonpath='{.status.phase}'=Succeeded clusterserviceversions.operators.coreos.com/scylladb-operator.v1.20.0
   

   Expected output:

   clusterserviceversion.operators.coreos.com/scylladb-operator.v1.20.0 condition met
   

The following manifest creates a RAID0 array from the available NVMe devices, formats it with the XFS filesystem, and enables performance tuning recommended for production-grade ScyllaDB deployments on the selected nodes.

apiVersion: scylla.scylladb.com/v1alpha1
kind: NodeConfig
metadata:
  name: scylladb-nodepool-1
spec:
  localDiskSetup:
    filesystems:
    - device: /dev/md/nvmes
      type: xfs
    mounts:
    - device: /dev/md/nvmes
      mountPoint: /var/lib/persistent-volumes
      unsupportedOptions:
      - prjquota
    raids:
    - name: nvmes
      type: RAID0
      RAID0:
        devices:
          modelRegex: Amazon EC2 NVMe Instance Storage
          nameRegex: ^/dev/nvme\d+n\d+$
  sysctls:
  - name: fs.aio-max-nr
    value: "30000000"
  - name: fs.file-max
    value: "9223372036854775807"
  - name: fs.nr_open
    value: "1073741816"
  - name: fs.inotify.max_user_instances
    value: "1200"
  - name: vm.swappiness
    value: "1"
  - name: vm.vfs_cache_pressure
    value: "2000"
  placement:
    nodeSelector:
      scylla.scylladb.com/node-type: scylla
    tolerations:
    - effect: NoSchedule
      key: scylla-operator.scylladb.com/dedicated
      operator: Equal
      value: scyllaclusters

You can apply it using the following command:

kubectl -n=scylla-operator apply --server-side -f=https://raw.githubusercontent.com/scylladb/scylla-operator/master/examples/openshift/rosa/nodeconfig.yaml

The following manifest creates a loop device, formats it with the XFS filesystem, and enables performance tuning on the selected nodes.

apiVersion: scylla.scylladb.com/v1alpha1
kind: NodeConfig
metadata:
  name: scylladb-nodepool-1
spec:
  localDiskSetup:
    loopDevices:
    - name: persistent-volumes
      imagePath: /var/lib/persistent-volumes.img
      size: 80Gi
    filesystems:
    - device: /dev/loops/persistent-volumes
      type: xfs
    mounts:
    - device: /dev/loops/persistent-volumes
      mountPoint: /var/lib/persistent-volumes
      unsupportedOptions:
      - prjquota
  sysctls:
  - name: fs.aio-max-nr
    value: "30000000"
  - name: fs.file-max
    value: "9223372036854775807"
  - name: fs.nr_open
    value: "1073741816"
  - name: fs.inotify.max_user_instances
    value: "1200"
  - name: vm.swappiness
    value: "1"
  - name: vm.vfs_cache_pressure
    value: "2000"
  placement:
    nodeSelector:
      scylla.scylladb.com/node-type: scylla
    tolerations:
    - effect: NoSchedule
      key: scylla-operator.scylladb.com/dedicated
      operator: Equal
      value: scyllaclusters

kubectl -n=scylla-operator apply --server-side -f=https://raw.githubusercontent.com/scylladb/scylla-operator/master/examples/generic/nodeconfig-alpha.yaml