Versions Compared

Key

  • This line was added.
  • This line was removed.
  • Formatting was changed.

...

Due to the different characteristics of the underlying storage providers, permissions are configured differently depending on the dataset type.

FileSet Permissions

FileSets store their data as files and directories in the file system. The typical access control used in file systems is by assigning different permissions to the file's owner, the file's group, and all others. For example, in a Posix file system, the permission string:

...

This will override any permissions configured for output datasets. Be aware that the umask is the inverse of a file permission: the above umask of 027 translates into file permissions of 750 or rwxr-x---.

Table Permissions

Table-based datasets are typically backed by HBase as the storage provider. HBase controls access by granting privileges explicitly to users and groups. This is more flexible than Unix-style file system permissions, as it allows a fine-grained control over multiple groups or individual users.

...

This gives user joe read, write, and administration rights, while all members of the group subscribers can only read. Other privileges you can assign are c (for create) and x (for execute rights); see the Apache HBase™ documentation for more details.

PartitionedFileSet Permissions

Partitioned file sets and Time-partitioned file sets are a hybrid of a Table to store partition metadata and a FileSet to store the partition data (that is, files). You can set both Table permissions and FileSet permissions in the dataset properties for these dataset types, and they will be applied to the partition table and the file system data, respectively.