使用 BR 备份 TiDB 集群数据到兼容 S3 的存储

本文详细描述了如何将运行在 AWS Kubernetes 环境中的 TiDB 集群数据备份到 AWS 的存储上。BR 会在底层获取集群的逻辑备份,然后再将备份数据上传到 AWS 的存储上。

本文使用的备份方式基于 TiDB Operator v1.1 及以上版本的 Custom Resource Definition(CRD) 实现。

Ad-hoc 备份

Ad-hoc 备份支持全量备份与增量备份。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。

为了更好地描述备份的使用方式,本文档提供如下备份示例。示例假设对部署在 Kubernetes test1 这个 namespace 中的 TiDB 集群 demo1 进行数据备份,下面是具体操作过程。

Ad-hoc 备份环境准备

  1. 下载文件 backup-rbac.yaml,并执行以下命令在 test1 这个 namespace 中创建备份需要的 RBAC 相关资源:

    kubectl apply -f backup-rbac.yaml -n test1
  2. 远程存储访问授权。

    如果使用 Amazon S3 来备份集群,可以使用三种权限授予方式授予权限,参考 AWS 账号授权授权访问兼容 S3 的远程存储;使用 Ceph 作为后端存储测试备份时,是通过 AccessKey 和 SecretKey 模式授权,设置方式可参考通过 AccessKey 和 SecretKey 授权

  3. 创建 backup-demo1-tidb-secret secret。该 secret 存放用于访问 TiDB 集群的用户所对应的密码。

    kubectl create secret generic backup-demo1-tidb-secret --from-literal=password=${password} --namespace=test1

数据库账户权限

  • mysql.tidb 表的 SELECTUPDATE 权限:备份前后,Backup CR 需要一个拥有该权限的数据库账户,用于调整 GC 时间

使用 BR 备份数据到 Amazon S3 的存储

  • 创建 Backup CR,通过 accessKey 和 secretKey 授权的方式备份集群:

    kubectl apply -f backup-aws-s3.yaml

    backup-aws-s3.yaml 文件内容如下:

    --- apiVersion: pingcap.com/v1alpha1 kind: Backup metadata: name: demo1-backup-s3 namespace: test1 spec: backupType: full br: cluster: demo1 clusterNamespace: test1 # logLevel: info # statusAddr: ${status_addr} # concurrency: 4 # rateLimit: 0 # timeAgo: ${time} # checksum: true # sendCredToTikv: true # options: # - --lastbackupts=420134118382108673 # 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 s3: provider: aws secretName: s3-secret region: us-west-1 bucket: my-bucket prefix: my-folder
  • 创建 Backup CR,通过 IAM 绑定 Pod 授权的方式备份集群:

    kubectl apply -f backup-aws-s3.yaml

    backup-aws-s3.yaml 文件内容如下:

    --- apiVersion: pingcap.com/v1alpha1 kind: Backup metadata: name: demo1-backup-s3 namespace: test1 annotations: iam.amazonaws.com/role: arn:aws:iam::123456789012:role/user spec: backupType: full br: cluster: demo1 sendCredToTikv: false clusterNamespace: test1 # logLevel: info # statusAddr: ${status_addr} # concurrency: 4 # rateLimit: 0 # timeAgo: ${time} # checksum: true # options: # - --lastbackupts=420134118382108673 # 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 s3: provider: aws region: us-west-1 bucket: my-bucket prefix: my-folder
  • 创建 Backup CR,通过 IAM 绑定 ServiceAccount 授权的方式备份集群:

    kubectl apply -f backup-aws-s3.yaml

    backup-aws-s3.yaml 文件内容如下:

    --- apiVersion: pingcap.com/v1alpha1 kind: Backup metadata: name: demo1-backup-s3 namespace: test1 spec: backupType: full serviceAccount: tidb-backup-manager br: cluster: demo1 sendCredToTikv: false clusterNamespace: test1 # logLevel: info # statusAddr: ${status_addr} # concurrency: 4 # rateLimit: 0 # timeAgo: ${time} # checksum: true # options: # - --lastbackupts=420134118382108673 # 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 s3: provider: aws region: us-west-1 bucket: my-bucket prefix: my-folder

以上三个示例分别使用三种授权模式将数据导出到 Amazon S3 存储上。Amazon S3 的 aclendpointstorageClass 配置项均可以省略。兼容 S3 的存储相关配置参考 S3 存储字段介绍

以上示例中,.spec.br 中的一些参数项均可省略,如 logLevelstatusAddrconcurrencyrateLimitchecksumtimeAgosendCredToTikv。更多 .spec.br 字段的详细解释参考 BR 字段介绍

自 v1.1.6 版本起,如果需要增量备份,只需要在 spec.br.options 中指定上一次的备份时间戳 --lastbackupts 即可。有关增量备份的限制,可参考使用 BR 进行备份与恢复

更多 Backup CR 字段的详细解释参考 Backup CR 字段介绍

创建好 Backup CR 后,可通过如下命令查看备份状态:

kubectl get bk -n test1 -o wide

备份示例

备份全部集群数据
--- apiVersion: pingcap.com/v1alpha1 kind: Backup metadata: name: demo1-backup-s3 namespace: test1 spec: backupType: full serviceAccount: tidb-backup-manager br: cluster: demo1 sendCredToTikv: false clusterNamespace: test1 # 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 s3: provider: aws region: us-west-1 bucket: my-bucket prefix: my-folder
备份单个数据库的数据

以下示例中,备份 db1 数据库的数据。

--- apiVersion: pingcap.com/v1alpha1 kind: Backup metadata: name: demo1-backup-s3 namespace: test1 spec: backupType: full serviceAccount: tidb-backup-manager tableFilter: - "db1.*" br: cluster: demo1 sendCredToTikv: false clusterNamespace: test1 # 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 s3: provider: aws region: us-west-1 bucket: my-bucket prefix: my-folder
备份单张表的数据

以下示例中,备份 db1.table1 表的数据。

--- apiVersion: pingcap.com/v1alpha1 kind: Backup metadata: name: demo1-backup-s3 namespace: test1 spec: backupType: full serviceAccount: tidb-backup-manager tableFilter: - "db1.table1" br: cluster: demo1 sendCredToTikv: false clusterNamespace: test1 # 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 s3: provider: aws region: us-west-1 bucket: my-bucket prefix: my-folder
使用表库过滤功能备份多张表的数据

以下示例中,备份 db1.table1 表 和 db1.table2 表的数据。

--- apiVersion: pingcap.com/v1alpha1 kind: Backup metadata: name: demo1-backup-s3 namespace: test1 spec: backupType: full serviceAccount: tidb-backup-manager tableFilter: - "db1.table1" - "db1.table2" # ... br: cluster: demo1 sendCredToTikv: false clusterNamespace: test1 # 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 s3: provider: aws region: us-west-1 bucket: my-bucket prefix: my-folder

定时全量备份

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。定时全量备份通过自定义的 BackupSchedule CR 对象来描述。每到备份时间点会触发一次全量备份,定时全量备份底层通过 Ad-hoc 全量备份来实现。下面是创建定时全量备份的具体步骤:

定时全量备份环境准备

Ad-hoc 备份环境准备

使用 BR 定时备份数据到 Amazon S3 的存储

  • 创建 BackupSchedule CR,开启 TiDB 集群定时全量备份,通过 accessKey 和 secretKey 授权的方式备份集群:

    kubectl apply -f backup-scheduler-aws-s3.yaml

    backup-scheduler-aws-s3.yaml 文件内容如下:

    --- apiVersion: pingcap.com/v1alpha1 kind: BackupSchedule metadata: name: demo1-backup-schedule-s3 namespace: test1 spec: #maxBackups: 5 #pause: true maxReservedTime: "3h" schedule: "*/2 * * * *" backupTemplate: backupType: full # Clean outdated backup data based on maxBackups or maxReservedTime. If not configured, the default policy is Retain # cleanPolicy: Delete br: cluster: demo1 clusterNamespace: test1 # logLevel: info # statusAddr: ${status_addr} # concurrency: 4 # rateLimit: 0 # timeAgo: ${time} # checksum: true # sendCredToTikv: true # 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 s3: provider: aws secretName: s3-secret region: us-west-1 bucket: my-bucket prefix: my-folder
  • 创建 BackupSchedule CR,开启 TiDB 集群定时全量备份,通过 IAM 绑定 Pod 授权的方式备份集群:

    kubectl apply -f backup-scheduler-aws-s3.yaml

    backup-scheduler-aws-s3.yaml 文件内容如下:

    --- apiVersion: pingcap.com/v1alpha1 kind: BackupSchedule metadata: name: demo1-backup-schedule-s3 namespace: test1 annotations: iam.amazonaws.com/role: arn:aws:iam::123456789012:role/user spec: #maxBackups: 5 #pause: true maxReservedTime: "3h" schedule: "*/2 * * * *" backupTemplate: backupType: full # Clean outdated backup data based on maxBackups or maxReservedTime. If not configured, the default policy is Retain # cleanPolicy: Delete br: cluster: demo1 sendCredToTikv: false clusterNamespace: test1 # logLevel: info # statusAddr: ${status_addr} # concurrency: 4 # rateLimit: 0 # timeAgo: ${time} # checksum: true # 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 s3: provider: aws region: us-west-1 bucket: my-bucket prefix: my-folder
  • 创建 BackupSchedule CR,开启 TiDB 集群定时全量备份, 通过 IAM 绑定 ServiceAccount 授权的方式备份集群:

    kubectl apply -f backup-scheduler-aws-s3.yaml

    backup-scheduler-aws-s3.yaml 文件内容如下:

    --- apiVersion: pingcap.com/v1alpha1 kind: BackupSchedule metadata: name: demo1-backup-schedule-s3 namespace: test1 spec: #maxBackups: 5 #pause: true maxReservedTime: "3h" schedule: "*/2 * * * *" backupTemplate: backupType: full serviceAccount: tidb-backup-manager # Clean outdated backup data based on maxBackups or maxReservedTime. If not configured, the default policy is Retain # cleanPolicy: Delete br: cluster: demo1 sendCredToTikv: false clusterNamespace: test1 # logLevel: info # statusAddr: ${status_addr} # concurrency: 4 # rateLimit: 0 # timeAgo: ${time} # checksum: true # 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 s3: provider: aws region: us-west-1 bucket: my-bucket prefix: my-folder

定时全量备份创建完成后,可以通过以下命令查看定时全量备份的状态:

kubectl get bks -n test1 -o wide

在进行集群恢复时,需要指定备份的路径,可以通过如下命令查看定时快照备份下面所有的备份条目,这些备份的名称以定时快照备份名称为前缀:

kubectl get bk -l tidb.pingcap.com/backup-schedule=demo1-backup-schedule-s3 -n test1

从以上示例可知,backupSchedule 的配置由两部分组成。一部分是 backupSchedule 独有的配置,另一部分是 backupTemplatebackupTemplate 指定集群及远程存储相关的配置,字段和 Backup CR 中的 spec 一样,详细介绍可参考 Backup CR 字段介绍backupSchedule 独有的配置项具体介绍可参考 BackupSchedule CR 字段介绍

删除备份的 Backup CR

删除备份的 Backup CR 可参考删除备份的 Backup CR

故障诊断

在使用过程中如果遇到问题,可以参考故障诊断