Configure replication between the sites¶

The sites must have the same copy of data. To do so, configure the replication between them so that sites are always in sync. The replication is defined via a replication channel where you specify which site is the source of data and which site receives it.

Prepare the primary site¶

Your replica site needs to connect to your primary site to replicate data from it. For this, each database Pod on the primary site must have an external IP addresses to be reached directly. This is done by exposing the database cluster Pods using the LoadBalancer service type. Read more about exposing a cluster.

Since the primary site is already running, we will patch its configuration with the following command. Replace the <namespace> placeholder with your namespace:

$ kubectl patch pxc cluster1 -n <namespace> --type=merge --patch '{
  "spec": {
  "pxc": {
    "expose": {
      "enabled": true,
      "type": "LoadBalancer"
    }
  }}}'

Configure the replication channel on the primary site. Specify the following Custom Resource options in the spec.pxc.replicationChannels subsection in the deploy/cr.yaml file:
- pxc.replicationChannels.[].name is the name of the channel,
- pxc.replicationChannels.[].isSource defines what cluster the data is replicated from. Set the value to true.
Run the following command to add this configuration:
```
$ kubectl patch pxc cluster1 --type=merge --patch '{
  "spec": {
    "pxc": {
      "replicationChannels": [
        {
          "name": "pxc1_to_pxc2",
          "isSource": true
        }
      ]
    }
  }}'
```

Check that the Pods are exposed by listing the services:

  $ kubectl get services -n <namespace>

Expected output

NAME                              TYPE           CLUSTER-IP       EXTERNAL-IP      PORT(S)
cluster1-pxc-0                    LoadBalancer   34.118.227.242   104.197.82.173   3306:32522/TCP                          7m5s
cluster1-pxc-1                    LoadBalancer   34.118.236.108   34.44.97.95      3306:32361/TCP                          7m5s
cluster1-pxc-2                    LoadBalancer   34.118.236.170   35.222.208.249   3306:31607/TCP

Store the public IP addresses of your Pods. You will need them during the replica site setup.

Prepare the replica site¶

Configure the replication channel on the replica site. Specify the following Custom Resource options in the spec.pxc.replicationChannels subsection in the deploy/cr.yaml file:

spec.pxc.replicationChannels - The replication channel configuration. The name of the channel must match the name on the primary site.
spec.pxc.replicationChannels[].isSource - Set the value to false to indicate that the replica site is not the source of the data.
spec.pxc.replicationChannels[].sourcesList - The list of sources. Specify the external IP addresses of the database Pods from the primary site.

Run the following command to apply a patch to the replica site’s configuration with the required information. Don’t forget to replace the <placeholders> with your values:

$ kubectl patch pxc cluster1 -n <namespace> --type=merge --patch '{
"spec": {
  "pxc": {
    "replicationChannels": [
      {
        "name": "pxc1_to_pxc2",
        "isSource": false,
        "sourcesList": [
          { "host": "34.118.227.242", "port": 3306, "weight": 100 },
          { "host": "34.118.227.242", "port": 3306, "weight": 100 },
          { "host": "34.118.227.242", "port": 3306, "weight": 100 }
        ]
      }
    ]
  }
}}'

Verify the replication¶

To verify that the replication is working, do the following:

Connect to Percona XtraDB Cluster on the primary site.

Create a database and a table.

mysql> CREATE DATABASE demo;

Expected output

Query OK, 1 row affected (0.02 sec)

mysql> CREATE TABLE demo.users(user_id INT PRIMARY KEY, user_name VARCHAR(30));

Expected output

Query OK, 0 rows affected (0.03 sec)

Insert some data into the database:

mysql> INSERT INTO demo.users VALUES (1, 'percona');

Connect to Percona XtraDB Cluster on the replica site.

Retrieve the data from the database:

mysql> SELECT * FROM demo.users;

Expected output

+---------+-----------+
| user_id | user_name |
+---------+-----------+
|       1 | percona   |
+---------+-----------+
1 row in set (0.00 sec)

Last update: 2025-05-19