Skip to content

Latest commit

 

History

History
113 lines (100 loc) · 4.7 KB

postgres.md

File metadata and controls

113 lines (100 loc) · 4.7 KB
Raw

Postgres-Operator 安装

依赖条件

开启 Postgres-Operator 组件

  1. 编辑 /etc/kubez/globals.yml

  2. 取消 enable_postgres: "no" 的注释，并设置为 "yes"

    ##################
# Postgres Options
##################
enable_postgres: "yes"

postgres_name: postgres
postgress_namespace: operators

  3. 执行安装命令（根据实际情况选择）

    # 单节点集群场景
kubez-ansible apply

# 高可用集群场景
kubez-ansible -i multinode apply

  4. 部署完验证

    # postgres 已注册至集群中
[root@VM-16-5-centos ~]# kubectl get deploy,csv -n operators
NAME                  READY   UP-TO-DATE   AVAILABLE   AGE
deployment.apps/pgo   1/1     1            1           116m

NAME                                                                 DISPLAY                           VERSION   REPLACES                  PHASE
clusterserviceversion.operators.coreos.com/postgresoperator.v5.3.0   Crunchy Postgres for Kubernetes   5.3.0     postgresoperator.v5.2.0   Succeeded

至此 Postgres Operator 已安装至集群中, 接下来展示 Postgres 实例的创建。

创建 Postgres CR 实例

  1. 修改 yaml 文件（根据实际情况选择具体参数） 
    apiVersion: postgres-operator.crunchydata.com/v1beta1
kind: PostgresCluster
metadata:
  name: hippo
spec:
  image: registry.developers.crunchydata.com/crunchydata/crunchy-postgres:ubi8-14.5-1
  postgresVersion: 14
  instances:
    - name: instance1
      replicas: 1
      dataVolumeClaimSpec:
        accessModes:
          - "ReadWriteOnce"
        storageClassName: "managed-nfs-storage"
        resources:
          requests:
            storage: 1Gi
  backups:
    pgbackrest:
      image: registry.developers.crunchydata.com/crunchydata/crunchy-pgbackrest:ubi8-2.40-1
      repos:
        - name: repo1
          volume:
            volumeClaimSpec:
              accessModes:
                - "ReadWriteOnce"
              storageClassName: "managed-nfs-storage"
              resources:
                requests:
                  storage: 1Gi
  • 修改 storageClassName 为实际存在的 storageClass
  • 修改 storage 为实际需要的大小

  1. 执行 kubectl apply 进行实例安装

    # create-postgres-cluster.yaml 为步骤1展示的内容
[root@VM-16-5-centos manifests]# kubectl apply -f create-postgres-cluster.yaml
postgrescluster.postgres-operator.crunchydata.com/hippo created

  2. 部署完验证

    # pod 均运行正常
[root@VM-16-5-centos manifests]# kubectl get po,pv,pvc
NAME                          READY   STATUS      RESTARTS   AGE
pod/hippo-backup-7gcg-zx684   0/1     Completed   0          3m26s
pod/hippo-instance1-crq2-0    4/4     Running     0          7m37s
pod/hippo-repo-host-0         2/2     Running     0          7m37s

NAME                                                        CAPACITY    ACCESS MODES   RECLAIM POLICY   STATUS   CLAIM                                 STORAGECLASS          REASON   AGE
persistentvolume/pvc-ba6958d7-3fc7-4a9c-bd69-4cb9d8d7c291   1Gi        RWO            Delete           Bound    default/hippo-repo1                   managed-nfs-storage            7m37s
persistentvolume/pvc-d2bcb0b3-9d7d-4e11-8bae-a74e62b3ad64   1Gi        RWO            Delete           Bound    default/hippo-instance1-crq2-pgdata   managed-nfs-storage            7m37s

NAME                                                STATUS   VOLUME                                     CAPACITY   ACCESS MODES   STORAGECLASS          AGE
persistentvolumeclaim/hippo-instance1-crq2-pgdata   Bound    pvc-d2bcb0b3-9d7d-4e11-8bae-a74e62b3ad64   1Gi        RWO            managed-nfs-storage   7m37s
persistentvolumeclaim/hippo-repo1                   Bound    pvc-ba6958d7-3fc7-4a9c-bd69-4cb9d8d7c291   1Gi        RWO            managed-nfs-storage   7m37s

# 进入 pod 验证
[root@VM-4-3-centos ~]# kubectl exec -it hippo-instance1-zw2j-0 -- /bin/bash
Defaulted container "database"out of: database, replication-cert-copy, pgbackrest, pgbackrest-config, postgres-startup (init), nss-wrapper-init (init)
psql (14.5)
Type "help" for help.
postgres=#

  3. 详细文档

    https://github.com/chenghongxi/kubernetes-learning/blob/master/olm/postgres-Operators/README.md