TiDB Operator备份TiDB集群到NFS持久卷

时间:2023-02-21 17:04:33

Kubernetes 上 TiDB 集群的数据备份到​​持久卷​​​上。本文描述的持久卷,指任何 ​​Kubernetes 支持的持久卷类型​​。本文以备份数据到网络文件系统 (NFS) 存储为例。

使用场景

如果你对数据备份有以下要求,可考虑使用 BR 将 TiDB 集群数据以 ​​Ad-hoc 备份​​(只执行一次)或​​定时快照备份​​(执行多次)的方式备份至持久卷:

  • 需要备份的数据量较大,而且要求备份速度较快
  • 需要直接备份数据的 SST 文件(键值对)

注意

  • BR 只支持 TiDB v3.1 及以上版本。
  • 使用 BR 备份出的数据只能恢复到 TiDB 数据库中,无法恢复到其他数据库中。

一、Ad-hoc 备份

Ad-hoc 备份支持快照备份与增量备份。Ad-hoc 备份通过创建一个自定义的 ​​Backup​​ custom resource (CR) 对象来描述一次备份。TiDB Operator 根据这个 ​​Backup​​ 对象来完成具体的备份过程。如果备份过程中出现错误,程序不会自动重试,此时需要手动处理。

本文档对K8S集群中命名空间为tidb下的lqb数据库集群进行备份,具体操作如下:

准备Ad-hoc备份环境

下载​backup-rbac.yaml

---
kind: Role
apiVersion: rbac.authorization.k8s.io/v1
metadata:
name: tidb-backup-manager
labels:
app.kubernetes.io/component: tidb-backup-manager
rules:
- apiGroups: [""]
resources: ["events"]
verbs: ["*"]
- apiGroups: ["pingcap.com"]
resources: ["backups", "restores"]
verbs: ["get", "watch", "list", "update"]

---
kind: ServiceAccount
apiVersion: v1
metadata:
name: tidb-backup-manager

---
kind: RoleBinding
apiVersion: rbac.authorization.k8s.io/v1
metadata:
name: tidb-backup-manager
labels:
app.kubernetes.io/component: tidb-backup-manager
subjects:
- kind: ServiceAccount
name: tidb-backup-manager
roleRef:
apiGroup: rbac.authorization.k8s.io
kind: Role
name: tidb-backup-manager

创建备份需要的RBAC相关资源在tidb命名空间下

[root@k8s-master backup]# kubectl apply -f backup-rbac.yaml -ntidb
role.rbac.authorization.k8s.io/tidb-backup-manager created
serviceaccount/tidb-backup-manager created
rolebinding.rbac.authorization.k8s.io/tidb-backup-manager created

确保可以从K8S集群中访问用于存储备份数据的NFS服务器,并且配置了TiKV挂载跟备份任务相同的NFS共享目录到相同的本地目录。tikv挂载NFS的具体配置如下:

spec:
tikv:
baseImage: pingcap/tikv
replicas: 4
maxFailoverCount: 6

#下边是配置tikv挂载nfs
additionalVolumes:
- name: nfs
nfs:
server: 172.16.5.194
path: /home/k8s-nfs
additionalVolumeMounts:
- name: nfs
mountPath: /nfs

requests:
cpu: "100m"
storage: 12Gi
memory: "400Mi"
limits:
cpu: "2000m"
memory: "4Gi"
mountClusterClientSecret: false
storageClassName: "local-hostpath"

如果TiDB版本低于V4.0.8则执行如下操作,若高于该版本可以省略。

创建secret用于存放Tidb集群的用户名和密码,由于使用v6.5省略该步骤
kubectl create secret generic backup-demo1-tidb-secret --from-literal=password=${password} --namespace=tidb

备份数据到NFS持久卷

创建备份的自定义资源CR,将数据备份到NFS

##以下以备份yz数据库为例
[root@k8s-master backup]# cat backup-nfs.yaml
---
apiVersion: pingcap.com/v1alpha1
kind: Backup
metadata:
name: demo1-backup-nfs
namespace: tidb
spec:
# backupType: full
# from:
# host:
# port:
# user:
# secretName: backup-yz

tableFilter:
- "yz.*"

br:
cluster: yz
clusterNamespace: tidb

local:
prefix: backup-nfs
volume:
name: nfs
nfs:
server: 172.16.5.194
path: /home/k8s-nfs
volumeMount:
name: nfs
mountPath: /nfs

创建应用配置

[root@k8s-master backup]# kubectl apply -f backup-nfs.yaml
backup.pingcap.com/demo1-backup-nfs created

查看备份状态和备份文件

[root@k8s-master backup]# kubectl get bk -n tidb -owide -w
NAME TYPE MODE STATUS BACKUPPATH BACKUPSIZE COMMITTS LOGTRUNCATEUNTIL STARTED COMPLETED AGE
demo1-backup-nfs snapshot Running local:///nfs/backup-nfs 57s
demo1-backup-nfs snapshot Complete local:///nfs/backup-nfs 2.8 GB 439178019024666631 68s 0s 72s
^C[root@k8s-master backup]ls /home/k8s-nfs/backup-nfs/de
1 20001 39461 4 backup.lock backupmeta checkpoint.meta checkpoints

备份实例如下:

备份全部集群数据
---
apiVersion: pingcap.com/v1alpha1
kind: Backup
metadata:
name: demo1-backup-nfs
namespace: test1
spec:
# # backupType: full
# # Only needed for TiDB Operator < v1.1.10 or TiDB < v4.0.8
# from:
# host: ${tidb-host}
# port: ${tidb-port}
# user: ${tidb-user}
# secretName: backup-demo1-tidb-secret
br:
cluster: demo1
clusterNamespace: test1
local:
prefix: backup-nfs
volume:
name: nfs
nfs:
server: ${nfs_server_ip}
path: /home/k8s-nfs
volumeMount:
name: nfs
mountPath: /nfs


备份单个数据库的数据
---
apiVersion: pingcap.com/v1alpha1
kind: Backup
metadata:
name: demo1-backup-nfs
namespace: test1
spec:
# # backupType: full
# # Only needed for TiDB Operator < v1.1.10 or TiDB < v4.0.8
# from:
# host: ${tidb-host}
# port: ${tidb-port}
# user: ${tidb-user}
# secretName: backup-demo1-tidb-secret
tableFilter:
- "db1.*"
br:
cluster: demo1
clusterNamespace: test1
local:
prefix: backup-nfs
volume:
name: nfs
nfs:
server: ${nfs_server_ip}
path: /home/k8s-nfs
volumeMount:
name: nfs
mountPath: /nfs
备份单张表的数据
---
apiVersion: pingcap.com/v1alpha1
kind: Backup
metadata:
name: demo1-backup-nfs
namespace: test1
spec:
# # backupType: full
# # Only needed for TiDB Operator < v1.1.10 or TiDB < v4.0.8
# from:
# host: ${tidb-host}
# port: ${tidb-port}
# user: ${tidb-user}
# secretName: backup-demo1-tidb-secret
tableFilter:
- "db1.table1"
br:
cluster: demo1
clusterNamespace: test1
local:
prefix: backup-nfs
volume:
name: nfs
nfs:
server: ${nfs_server_ip}
path: /home/k8s-nfs
volumeMount:
name: nfs
mountPath: /nfs
使用表库过滤功能备份多张表的数据
---
apiVersion: pingcap.com/v1alpha1
kind: Backup
metadata:
name: demo1-backup-nfs
namespace: test1
spec:
# # backupType: full
# # Only needed for TiDB Operator < v1.1.10 or TiDB < v4.0.8
# from:
# host: ${tidb-host}
# port: ${tidb-port}
# user: ${tidb-user}
# secretName: backup-demo1-tidb-secret
tableFilter:
- "db1.table1"
- "db1.table2"
br:
cluster: demo1
clusterNamespace: test1
local:
prefix: backup-nfs
volume:
name: nfs
nfs:
server: ${nfs_server_ip}
path: /home/k8s-nfs
volumeMount:
name: nfs
mountPath: /nfs

二、定时快照备份

用户通过设置备份策略来对 TiDB 集群进行定时备份,同时设置备份的保留策略以避免产生过多的备份。定时快照备份通过自定义的 ​​BackupSchedule​​ CR 对象来描述。每到备份时间点会触发一次快照备份,定时快照备份底层通过 Ad-hoc 快照备份来实现。下面是创建定时快照备份的具体步骤:

准备定时快照备份环境(和Ad-hoc备份的第一步一样)略

备份数据到NFS持久卷中

创建备份自定义资源BackupSchedule,将数据备份到NFS中

[root@k8s-master backup]# cat backup-schedule-nfs.yaml
apiVersion: pingcap.com/v1alpha1
kind: BackupSchedule
metadata:
name: demo1-backup-schedule-nfs
namespace: tidb
spec:

maxReservedTime: "3h"
schedule: "*/2 * * * *"
backupTemplate:

br:
cluster: yz
clusterNamespace: tidb

local:
prefix: backup-nfs
volume:
name: nfs
nfs:
server: 172.16.5.194
path: /home/k8s-nfs
volumeMount:
name: nfs
mountPath: /nfs

创建应用配置

[root@k8s-master backup]# kubectl apply -f backup-schedule-nfs.yaml
backupschedule.pingcap.com/demo1-backup-schedule-nfs created

查看备份状态和备份文件

[root@k8s-master backup]# kubectl get bks -n tidb -owide
NAME SCHEDULE MAXBACKUPS LASTBACKUP LASTBACKUPTIME AGE
demo1-backup-schedule-nfs */2 * * * * 9s
NAME SCHEDULE MAXBACKUPS LASTBACKUP LASTBACKUPTIME AGE
demo1-backup-schedule-nfs */2 * * * * 46s


demo1-backup-schedule-nfs */2 * * * * demo1-backup-schedule-nfs-2023-02-02t10-16-00 6s 89s
demo1-backup-schedule-nfs */2 * * * * demo1-backup-schedule-nfs-2023-02-02t10-18-00 6s 3m29s
demo1-backup-schedule-nfs */2 * * * * demo1-backup-schedule-nfs-2023-02-02t10-20-00 6s 5m29s
demo1-backup-schedule-nfs */2 * * * * demo1-backup-schedule-nfs-2023-02-02t10-22-00 6s 7m29s
demo1-backup-schedule-nfs */2 * * * * demo1-backup-schedule-nfs-2023-02-02t10-24-00 6s 9m29s

[root@k8s-master ~]# ls /home/k8s-nfs/backup-nfs/yz-pd.tidb-2379-2023-02-02t10-16-00/
1 20001 39461 4 backup.lock backupmeta checkpoint.meta checkpoints

三、删除备份的 Backup CR

kubectl delete backup ${name} -n ${namespace}
kubectl delete backupschedule ${name} -n ${namespace}