Backup Storages

TiDB supports storing backup data to Amazon S3, Google Cloud Storage (GCS), Azure Blob Storage, and NFS. Specifically, you can specify the URI of backup storage in the --storage or -s parameter of br commands. This document introduces the URI format and authentication of different external storage services, and server-side encryption.

Send credentials to TiKV

CLI parameterDescriptionDefault value
—send-credentials-to-tikvControls whether to send credentials obtained by BR to TiKV.true

By default, BR sends a credential to each TiKV node when using Amazon S3, GCS, or Azure Blob Storage as the storage system. This behavior simplifies the configuration and is controlled by the parameter --send-credentials-to-tikv(or -c in short).

Note that this operation is not applicable to cloud environments. If you use IAM Role authorization, each node has its own role and permissions. In this case, you need to configure --send-credentials-to-tikv=false (or -c=0 in short) to disable sending credentials:

  1. tiup br backup full -c=0 -u pd-service:2379 --storage 's3://bucket-name/prefix'

If you back up or restore data using the BACKUP and RESTORE statements, you can add the SEND_CREDENTIALS_TO_TIKV = FALSE option:

  1. BACKUP DATABASE * TO 's3://bucket-name/prefix' SEND_CREDENTIALS_TO_TIKV = FALSE;

URI format

URI format description

The URI format of the external storage service is as follows:

  1. [scheme]://[host]/[path]?[parameters]

For more information about the URI format, see URI Formats of External Storage Services.

URI examples

This section provides some URI examples by using external as the host parameter (bucket name or container name in the preceding sections).

  • Amazon S3
  • GCS
  • Azure Blob Storage

Back up snapshot data to Amazon S3

  1. tiup br backup full -u "${PD_IP}:2379" \
  2. --storage "s3://external/backup-20220915?access-key=${access-key}&secret-access-key=${secret-access-key}"

Restore snapshot data from Amazon S3

  1. tiup br restore full -u "${PD_IP}:2379" \
  2. --storage "s3://external/backup-20220915?access-key=${access-key}&secret-access-key=${secret-access-key}"

Back up snapshot data to GCS

  1. tiup br backup full --pd "${PD_IP}:2379" \
  2. --storage "gcs://external/backup-20220915?credentials-file=${credentials-file-path}"

Restore snapshot data from GCS

  1. tiup br restore full --pd "${PD_IP}:2379" \
  2. --storage "gcs://external/backup-20220915?credentials-file=${credentials-file-path}"

Back up snapshot data to Azure Blob Storage

  1. tiup br backup full -u "${PD_IP}:2379" \
  2. --storage "azure://external/backup-20220915?account-name=${account-name}&account-key=${account-key}"

Restore the test database from snapshot backup data in Azure Blob Storage

  1. tiup br restore db --db test -u "${PD_IP}:2379" \
  2. --storage "azure://external/backup-20220915account-name=${account-name}&account-key=${account-key}"

Authentication

When storing backup data in a cloud storage system, you need to configure authentication parameters depending on the specific cloud service provider. This section describes the authentication methods used by Amazon S3, GCS, and Azure Blob Storage, and how to configure the accounts used to access the corresponding storage service.

  • Amazon S3
  • GCS
  • Azure Blob Storage

Before backup, configure the following privileges to access the backup directory on S3.

  • Minimum privileges for TiKV and Backup & Restore (BR) to access the backup directories during backup: s3:ListBucket, s3:GetObject, s3:DeleteObject, s3:PutObject, and s3:AbortMultipartUpload
  • Minimum privileges for TiKV and BR to access the backup directories during restore: s3:ListBucket, s3:GetObject, s3:DeleteObject, and s3:PutObject. BR writes checkpoint information to the ./checkpoints subdirectory under the backup directory. When restoring log backup data, BR writes the table ID mapping relationship of the restored cluster to the ./pitr_id_maps subdirectory under the backup directory.

If you have not yet created a backup directory, refer to Create a bucket to create an S3 bucket in the specified region. If necessary, you can also create a folder in the bucket by referring to Create a folder.

It is recommended that you configure access to S3 using either of the following ways:

  • Method 1: Specify the access key

    If you specify an access key and a secret access key in the URI, authentication is performed using the specified access key and secret access key. Besides specifying the key in the URI, the following methods are also supported:

    • BR reads the environment variables $AWS_ACCESS_KEY_ID and $AWS_SECRET_ACCESS_KEY.
    • BR reads the environment variables $AWS_ACCESS_KEY and $AWS_SECRET_KEY.
    • BR reads the shared credentials file in the path specified by the environment variable $AWS_SHARED_CREDENTIALS_FILE.
    • BR reads the shared credentials file in the ~/.aws/credentials path.

  • Method 2: Access based on the IAM role

    Associate an IAM role that can access S3 with EC2 instances where the TiKV and BR nodes run. After the association, BR can directly access the backup directories in S3 without additional settings.

    1. tiup br backup full --pd "${PD_IP}:2379" \
    2. --storage "s3://${host}/${path}"

You can configure the account used to access GCS by specifying the access key. If you specify the credentials-file parameter, the authentication is performed using the specified credentials-file. Besides specifying the key in the URI, the following methods are also supported:

  • BR reads the file in the path specified by the environment variable $GOOGLE_APPLICATION_CREDENTIALS
  • BR reads the file ~/.config/gcloud/application_default_credentials.json.

  • BR obtains the credentials from the metadata server when the cluster is running in GCE or GAE.

  • Method 1: Specify the shared access signature

    If you specify account-name and sas-token in the URI, the authentication is performed using the specified account name and shared access signature (SAS) token. Note that the SAS token contains the & character. You need to encode it as %26 before appending it to the URI. You can also directly encode the entire sas-token using percent-encoding.

  • Method 2: Specify the access key

    If you specify account-name and account-key in the URI, the authentication is performed using the specified account name and account key. Besides the method of specifying the key in the URI, BR can also read the key from the environment variable $AZURE_STORAGE_KEY.

  • Method 3: Use Azure AD for backup and restore

    Configure the environment variables $AZURE_CLIENT_ID, $AZURE_TENANT_ID, and $AZURE_CLIENT_SECRET on the node where BR is running.

    • When the cluster is started using TiUP, TiKV uses the systemd service. The following example shows how to configure the preceding three environment variables for TiKV:

      Backup Storages - 图1

      Note

      If this method is used, you need to restart TiKV in step 3. If your cluster cannot be restarted, use Method 1: Specify the access key for backup and restore.

      1. Suppose that the TiKV port on this node is 24000, that is, the name of the systemd service is tikv-24000:

        1. systemctl edit tikv-24000

      2. Edit the TiKV configuration file to configure the three environment variables:

        1. [Service]
        2. Environment="AZURE_CLIENT_ID=aaaaaaaa-aaaa-aaaa-aaaa-aaaaaaaaaaaa"
        3. Environment="AZURE_TENANT_ID=aaaaaaaa-aaaa-aaaa-aaaa-aaaaaaaaaaaa"
        4. Environment="AZURE_CLIENT_SECRET=aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa"

      3. Reload the configuration and restart TiKV:

        1. systemctl daemon-reload
        2. systemctl restart tikv-24000

    • To configure the Azure AD information for TiKV and BR started with command lines, you only need to check whether the environment variables $AZURE_CLIENT_ID, $AZURE_TENANT_ID, and $AZURE_CLIENT_SECRET are configured in the operating environment by running the following commands:

      1. echo $AZURE_CLIENT_ID
      2. echo $AZURE_TENANT_ID
      3. echo $AZURE_CLIENT_SECRET

    • Use BR to back up data to Azure Blob Storage:

      1. tiup br backup full -u "${PD_IP}:2379" \
      2. --storage "azure://external/backup-20220915?account-name=${account-name}"

Server-side encryption

Amazon S3 server-side encryption

BR supports server-side encryption when backing up data to Amazon S3. You can also use an AWS KMS key you create for S3 server-side encryption using BR. For details, see BR S3 server-side encryption.

Azure Blob Storage server-side encryption

BR supports specifying the Azure server-side encryption scope or providing the encryption key when backing up data to Azure Blob Storage. This feature lets you establish a security boundary for different backup data of the same storage account. For details, see BR Azure Blob Storage server-side encryption.

Other features supported by the storage service

Amazon S3 Object Lock can help prevent backup data from accidental or intentional deletion during a specified retention period, enhancing the security and integrity of data. Starting from v6.3.0, BR supports Amazon S3 Object Lock for snapshot backups, adding an additional layer of security for full backups. Starting from v8.0.0, PITR also supports Amazon S3 Object Lock. Whether for full backups or log data backups, the Object Lock feature ensures more reliable data protection, further strengthening the security of data backup and recovery and meeting regulatory requirements.

BR and PITR automatically detect whether the Amazon S3 Object Lock feature is enabled or disabled. You do not need to perform any additional operations.

Backup Storages - 图2

Warning

If the Object Lock feature is enabled during the snapshot backup or PITR log backup process, the snapshot backup or log backup might fail. You need to restart the snapshot backup or PITR log backup task to continue the backup.