Skip to content

Cloud Native PostgreSQL: An Operator to run full featured PostgreSQL in Kubernetes

Introduction

CloudNativePG is the Kubernetes operator that covers the full lifecycle of a highly available PostgreSQL database cluster with a primary/standby architecture, using native streaming replication.

Installation

To install the CloudNativePG operator, you can use the helm chart:

1
2
3
helm repo add cloudnative-pg https://cloudnative-pg.io/charts
helm repo update
helm install cloudnative-pg cloudnative-pg/cloudnative-pg

Configuration

The CloudNativePG operator can be configured using the following parameters:

1
2
3
4
5
replicaCount: 1 # Number of replicas to run
resources: {}
# If you're using Prometheus, you can enable the pod monitor
monitoring:
  podMonitorEnabled: false

Deploy a cluster

First of all, set the login/password of the superuser. Let's define a secret for this:

1
2
3
4
5
6
7
8
apiVersion: v1
kind: Secret
type: kubernetes.io/basic-auth
metadata:
  name: pg-cluster-superuser
data:
  password: cG9zdGdyZXM
  username: cG9zdGdyZXM

As you can see, the password and username are base64 encoded. You can use the following command to encode them:

1
echo -n "postgres" | base64

And then set the cluster configuration with MetalLB service:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
apiVersion: postgresql.cnpg.io/v1
kind: Cluster
metadata:
  name: pg-cluster
spec:
  # Number of replicas to run
  instances: 2
  # Storage class to use
  storage:
    storageClass: openebs-lvm
    # Size of the storage
    size: 2Gi

  # Enable superuser access
  enableSuperuserAccess: true
  # Secret containing the superuser credentials we previously created
  superuserSecret:
    name: cluster-example-superuser

  # Resources for those PostgreSQL pods
  resources:
    requests:
      memory: "512Mi"
      cpu: "256m"
    limits:
      memory: "512Mi"
      cpu: "1"

  # Managed services
  managed:
    services:
      additional:
        # Point to the read/write load balancer service
        - selectorType: rw
          serviceTemplate:
            metadata:
              name: pg-cluster-rw-lb
              annotations:
                # Set the IP address of the load balancer if you're using MetalLB.
                metallb.universe.tf/loadBalancerIPs: 192.168.0.206
            spec:
              type: LoadBalancer

  # Optional: Set the affinity for the pods
  affinity:
    enablePodAntiAffinity: true
    topologyKey: kubernetes.io/hostname
    podAntiAffinityType: required
    tolerations:
      - key: xxx
        operator: "Equal"
        value: "yyy"
        effect: NoSchedule

Then, deploy the cluster and the secret:

1
kubectl apply -f secret.yaml -f cluster.yaml

You can validate the deployment pods and services this way:

1
2
3
4
5
6
$ kubectl get services
NAME               TYPE           CLUSTER-IP       EXTERNAL-IP   PORT(S)        AGE
pg-cluster-r       ClusterIP      10.43.71.9       <none>        5432/TCP       2d18h
pg-cluster-ro      ClusterIP      10.43.186.251    <none>        5432/TCP       2d18h
pg-cluster-rw      ClusterIP      10.43.65.15      <none>        5432/TCP       2d18h
pg-cluster-rw-lb   LoadBalancer   10.43.169.20     192.168.0.206 5432:32088/TCP 2d17h

1
2
3
4
$ kubectl get pods
NAME            READY   STATUS    RESTARTS   AGE
pg-cluster-1    1/1     Running   0          2d17h
pg-cluster-2    1/1     Running   0          2d17h

As you can see, multiple services:

  • pg-cluster-r: Points to any PostgreSQL instance in the cluster (read).
  • pg-cluster-ro Points to the replicas, where available (read-only).
  • pg-cluster-rw Points to the primary instance of the cluster (read/write).
  • pg-cluster-rw-lb is the load balancer service we've set earlier