- Creation of repositories, relying on existing brokers or accessing cloud storage directly through AWS s3 protocol, or accessing HDFS directly.
- If it is a read-only repository, restores can only be done on the repository. If not, backup and restore operations are available.
- PROPERTIES are different according to different types of broker or S3 or hdfs, see the example for details.
- ON LOCATION : if it is S3 , here followed by the Bucket Name.
- Create a warehouse named bos_repo, rely on BOS broker “bos_broker”, and the data root directory is: bos://palo_backup
WITH BROKER `bos_broker`
ON LOCATION "bos://palo_backup"
PROPERTIES
"bos_endpoint" = "http://gz.bcebos.com",
"bos_accesskey" = "bos_accesskey",
"bos_secret_accesskey"="bos_secret_accesskey"
);
- Create the same repository as Example 1, but with read-only properties:
- Create a warehouse named hdfs_repo, rely on Baidu hdfs broker “hdfs_broker”, the data root directory is: hdfs://hadoop-name-node:54310/path/to/repo/
WITH BROKER `hdfs_broker`
ON LOCATION "hdfs://hadoop-name-node:54310/path/to/repo/"
PROPERTIES
(
"username" = "user",
"password" = "password"
);
- Create a repository named hdfs_repo to link HDFS directly without going through the broker.
CREATE REPOSITORY `hdfs_repo`
ON LOCATION "hdfs://hadoop-name-node:54310/path/to/repo/"
PROPERTIES
(
"fs.defaultFS"="hdfs://hadoop-name-node:54310",
"hadoop.username"="user"
);
### Keywords
- Create a repository named minio_repo to link minio storage directly through the s3 protocol.
- A cluster can create multiple warehouses. Only users with ADMIN privileges can create repositories.
- When performing data migration operations, it is necessary to create the exact same warehouse in the source cluster and the destination cluster, so that the destination cluster can view the data snapshots backed up by the source cluster through this warehouse.