3.1 Setting the NDB Cluster's Configuration

NDB Operator relies on a custom resource definition (CRD) named NdbCluster to obtain the MySQL Cluster configuration data that it needs to start. Whenever a user creates, modifies, or deletes a Kubernetes object of type NdbCluster, NDB Operator receives a change event, and updates the NDB Cluster running in Kubernetes Cluster accordingly. (See Chapter 1, Introduction to NDB Operator, for a description of this mechanism.)

The NdbCluster CRD defines a Kubernetes resource type which can be used to specify the configuration of an NDB Cluster. See Section 5.1, “NdbCluster Resource”, for more information.

The docs/examples directory in the NDB Operator source tree contains several examples, including example-ndb.yaml. This file contains an NdbCluster specification having the characteristics shown here, specified using YAML format:

apiVersion: mysql.oracle.com/v1
kind: NdbCluster
metadata:
  name: example-ndb
spec:
  redundancyLevel: 2
  dataNode:
    nodeCount: 2
  mysqlNode:
    nodeCount: 2

spec.dataNode.nodeCount sets the number of data nodes.

spec.redundancyLevel specifies the number of replicas as well as the number of management nodes (ndb_mgmd processes). Since this is greater than 1, the NDB Cluster is created with two management nodes.

Note

The number of management nodes is not directly configurable; it is determined solely by the value of redundancyLevel.

spec.dataNode.nodeCount determines the number of data nodes in the NDB Cluster.

spec.mysqld.nodeCount determines the number of MySQL Servers attached to the NDB Cluster as SQL nodes, providing an SQL front end to the NDB Cluster data nodes.