Adding Nodes

This chart supports two methods for adding new nodes to an existing installation:

1. Using Spock to add a node via logical replication
2. Using CloudNativePG to add a node via bootstrap

Adding a node via spock

In order to add a node after installing the chart, add a new entry to the nodes list with configuration for the new node.

For example, if you have the chart installed with 2 nodes, like in this configuration:

pgEdge:
  appName: pgedge
  nodes:
    - name: n1
      hostname: pgedge-n1-rw
    - name: n2
      hostname: pgedge-n2-rw
  clusterSpec:
    storage:
      size: 1Gi

You can add a new node simply by introducing it to the nodes list, and including the mode: spock property to bootstrap it from another node:

pgEdge:
  appName: pgedge
  nodes:
    - name: n1
      hostname: pgedge-n1-rw
    - name: n2
      hostname: pgedge-n2-rw
    - name: n3
      hostname: pgedge-n3-rw
      bootstrap:
        mode: spock
        sourceNode: n1
  clusterSpec:
    storage:
      size: 1Gi

From there, perform a helm upgrade to deploy the new node.

helm upgrade \
--values examples/configs/single/values.yaml \
    --wait \
    pgedge ./

The init-spock job will run during the upgrade, ensuring that replication configuration is established across the new and existing nodes, and use Spock subscription's sync_data and sync_structure to align the new node via logical replication.

Note

In order to ensure the health of replication across your nodes, you should:

• Stop writes to all nodes before performing the upgrade.
• Ensure all previous writes have replicated to other nodes by monitoring replication lag via spock.

This will ensure that all nodes remain aligned during the update, and replication can continue successfully once you resume writes.

Adding a node via CloudNativePG bootstrap

As an alternative approach to adding a node, you can also bootstrap the new node using CloudNativePG's Bootstrap from another cluster capability.

Here is an example of adding a node n3 using the Barman Cloud CNPG-I plugin to bootstrap the node from the existing node n1 which has backups and wal archiving configured in S3.

pgEdge:
  appName: pgedge
  nodes:
    - name: n1
      hostname: pgedge-n1-rw
      clusterSpec: 
        plugins:
        - name: barman-cloud.cloudnative-pg.io
          isWALArchiver: true
          parameters:
            barmanObjectName: s3-store
    - name: n2
      hostname: pgedge-n2-rw
    - name: n3
      hostname: pgedge-n3-rw
      bootstrap:
        mode: cnpg
      clusterSpec: 
        bootstrap:
          initdb: null
          recovery:
            source: pgedge-n1
        externalClusters:
        - name: pgedge-n1
          plugin:
            name: barman-cloud.cloudnative-pg.io
            parameters:
              barmanObjectName: s3-store
              serverName: pgedge-n1

  clusterSpec:
    storage:
      size: 1Gi

This builds upon the example configuration started in Performing Backups.

The subsequent init-spock job reconfigures the restored node, ensuring replication remains healthy after performing the update.

Regardless of the CloudNativePG bootstrap approach you take, you should ensure that the data being restored on the new node aligns with the state of the other nodes.