Manage fileset metadata using Gravitino

This page introduces how to manage fileset metadata in Gravitino. Filesets are a collection of files and directories. Users can leverage filesets to manage non-tabular data like training datasets and other raw data.

Typically, a fileset is mapped to a directory on a file system like HDFS, S3, ADLS, GCS, etc. With the fileset managed by Gravitino, the non-tabular data can be managed as assets together with tabular data in Gravitino in a unified way.

After a fileset is created, users can easily access, manage the files/directories through the fileset's identifier, without needing to know the physical path of the managed dataset. Also, with unified access control mechanism, filesets can be managed via the same role based access control mechanism without needing to set access controls across different storage systems.

To use fileset, please make sure that:

• Gravitino server has started, and the host and port is http://localhost:8090.
• A metalake has been created.

Catalog operations

Create a catalog

tip

For a fileset catalog, you must specify the catalog type as FILESET when creating the catalog.

You can create a catalog by sending a POST request to the /api/metalakes/{metalake_name}/catalogs endpoint or just use the Gravitino Java client. The following is an example of creating a catalog:

Shell

curl -X POST -H "Accept: application/vnd.gravitino.v1+json" \
-H "Content-Type: application/json" -d '{
  "name": "catalog",
  "type": "FILESET",
  "comment": "comment",
  "provider": "hadoop",
  "properties": {
    "location": "file:/tmp/root"
  }
}' http://localhost:8090/api/metalakes/metalake/catalogs

Java

GravitinoClient gravitinoClient = GravitinoClient
    .builder("http://localhost:8090")
    .withMetalake("metalake")
    .build();

Map<String, String> properties = ImmutableMap.<String, String>builder()
    .put("location", "file:/tmp/root")
    // Property "location" is optional. If specified, a managed fileset without
    // a storage location will be stored under this location.
    .build();

Catalog catalog = gravitinoClient.createCatalog(
    NameIdentifier.of("metalake", "catalog"),
    Type.FILESET,
    "hadoop", // provider, Gravitino only supports "hadoop" for now.
    "This is a Hadoop fileset catalog",
    properties);
// ...

Python

gravitino_client: GravitinoClient = GravitinoClient(uri="http://localhost:8090", metalake_name="metalake")
catalog = gravitino_client.create_catalog(ident=NameIdentifier.of("metalake", "catalog"),
                                          type=Catalog.Type.FILESET,
                                          provider="hadoop", 
                                          comment="This is a Hadoop fileset catalog",
                                          properties={"location": "/tmp/test1"})

Currently, Gravitino supports the following catalog providers:

Catalog provider	Catalog property
hadoop	Hadoop catalog property

Load a catalog

Refer to Load a catalog in relational catalog for more details. For a fileset catalog, the load operation is the same.

Alter a catalog

Refer to Alter a catalog in relational catalog for more details. For a fileset catalog, the alter operation is the same.

Drop a catalog

Refer to Drop a catalog in relational catalog for more details. For a fileset catalog, the drop operation is the same.

note

Currently, Gravitino doesn't support dropping a catalog with schemas and filesets under it. You have to drop all the schemas and filesets under the catalog before dropping the catalog.

List all catalogs in a metalake

Please refer to List all catalogs in a metalake in relational catalog for more details. For a fileset catalog, the list operation is the same.

List all catalogs' information in a metalake

Please refer to List all catalogs' information in a metalake in relational catalog for more details. For a fileset catalog, the list operation is the same.

Schema operations

Schema is a virtual namespace in a fileset catalog, which is used to organize the fileset. It is similar to the concept of schema in relational catalog.

tip

Users should create a metalake and a catalog before creating a schema.

Create a schema

You can create a schema by sending a POST request to the /api/metalakes/{metalake_name}/catalogs/{catalog_name}/schemas endpoint or just use the Gravitino Java client. The following is an example of creating a schema:

Shell

curl -X POST -H "Accept: application/vnd.gravitino.v1+json" \
-H "Content-Type: application/json" -d '{
  "name": "schema",
  "comment": "comment",
  "properties": {
    "location": "file:/tmp/root/schema"
  }
}' http://localhost:8090/api/metalakes/metalake/catalogs/catalog/schemas

Java

GravitinoClient gravitinoClient = GravitinoClient
    .builder("http://localhost:8090")
    .withMetalake("metalake")
    .build();

// Assuming you have just created a Hadoop catalog named `catalog`
Catalog catalog = gravitinoClient.loadCatalog(NameIdentifier.of("metalake", "catalog"));

SupportsSchemas supportsSchemas = catalog.asSchemas();

Map<String, String> schemaProperties = ImmutableMap.<String, String>builder()
    // Property "location" is optional, if specified all the managed fileset without
    // specifying storage location will be stored under this location.
    .put("location", "file:/tmp/root/schema")
    .build();
Schema schema = supportsSchemas.createSchema(
    NameIdentifier.of("metalake", "catalog", "schema"),
    "This is a schema",
    schemaProperties
);
// ...

Python

gravitino_client: GravitinoClient = GravitinoClient(uri="http://localhost:8090", metalake_name="metalake")

catalog: Catalog = gravitino_client.load_catalog(ident=NameIdentifier.of("metalake", "catalog"))
catalog.as_schemas().create_schema(ident=NameIdentifier.of("metalake", "catalog", "schema"), 
                                   comment="This is a schema",
                                   properties={"location": "/tmp/root/schema"})

Currently, Gravitino supports the following schema property:

Catalog provider	Schema property
hadoop	Hadoop schema property

Load a schema

Please refer to Load a schema in relational catalog for more details. For a fileset catalog, the schema load operation is the same.

Alter a schema

Please refer to Alter a schema in relational catalog for more details. For a fileset catalog, the schema alter operation is the same.

Drop a schema

Please refer to Drop a schema in relational catalog for more details. For a fileset catalog, the schema drop operation is the same.

Note that the drop operation will also remove all of the filesets as well as the managed files under this schema path if cascade is set to true.

List all schemas under a catalog

Please refer to List all schemas under a catalog in relational catalog for more details. For a fileset catalog, the schema list operation is the same.

Fileset operations

tip

• Users should create a metalake, a catalog, and a schema before creating a fileset.
• Currently, Gravitino only supports managing Hadoop Compatible File System (HCFS) locations.

Create a fileset

You can create a fileset by sending a POST request to the /api/metalakes/{metalake_name} /catalogs/{catalog_name}/schemas/{schema_name}/filesets endpoint or just use the Gravitino Java client. The following is an example of creating a fileset:

Shell

curl -X POST -H "Accept: application/vnd.gravitino.v1+json" \
-H "Content-Type: application/json" -d '{
  "name": "example_fileset",
  "comment": "This is an example fileset",
  "type": "MANAGED",
  "storageLocation": "file:/tmp/root/schema/example_fileset",
  "properties": {
    "k1": "v1"
  }
}' http://localhost:8090/api/metalakes/metalake/catalogs/catalog/schemas/schema/filesets

Java

GravitinoClient gravitinoClient = GravitinoClient
    .builder("http://localhost:8090")
    .withMetalake("metalake")
    .build();

Catalog catalog = gravitinoClient.loadCatalog(NameIdentifier.of("metalake", "catalog"));
FilesetCatalog filesetCatalog = catalog.asFilesetCatalog();

Map<String, String> propertiesMap = ImmutableMap.<String, String>builder()
        .put("k1", "v1")
        .build();

filesetCatalog.createFileset(
  NameIdentifier.of("metalake", "catalog", "schema", "example_fileset"),
  "This is an example fileset",
  Fileset.Type.MANAGED,
  "file:/tmp/root/schema/example_fileset",
  propertiesMap,
);

Python

gravitino_client: GravitinoClient = GravitinoClient(uri="http://localhost:8090", metalake_name="metalake")

catalog: Catalog = gravitino_client.load_catalog(ident=NameIdentifier.of("metalake", "catalog"))
catalog.as_fileset_catalog().create_fileset(ident=NameIdentifier.of("metalake", "catalog", "schema", "example_fileset"),
                                            type=Fileset.Type.MANAGED,
                                            comment="This is an example fileset",
                                            storage_location="/tmp/root/schema/example_fileset",
                                            properties={"k1": "v1"})

Currently, Gravitino supports two types of filesets:

• MANAGED: The storage location of the fileset is managed by Gravitino when specified as MANAGED, the physical location of the fileset will be deleted when this fileset is dropped.
• EXTERNAL: The storage location of the fileset is not managed by Gravitino, when specified as EXTERNAL, the files of the fileset will not be deleted when the fileset is dropped.

storageLocation

The storageLocation is the physical location of the fileset. Users can specify this location when creating a fileset, or follow the rules of the catalog/schema location if not specified.

For a MANAGED fileset, the storage location is:

1. The one specified by the user during the fileset creation.
2. When the catalog property location is specified but the schema property location isn't specified, the storage location is catalog location/schema name/fileset name.
3. When the catalog property location isn't specified but the schema property location is specified, the storage location is schema location/fileset name.
4. When both the catalog property location and the schema property location are specified, the storage location is schema location/fileset name.
5. When both the catalog property location and schema property location isn't specified, the user should specify the storageLocation in the fileset creation.

For EXTERNAL fileset, users should specify storageLocation during the fileset creation, otherwise, Gravitino will throw an exception.

Alter a fileset

You can modify a fileset by sending a PUT request to the /api/metalakes/{metalake_name} /catalogs/{catalog_name}/schemas/{schema_name}/filesets/{fileset_name} endpoint or just use the Gravitino Java client. The following is an example of modifying a fileset:

Shell

curl -X PUT -H "Accept: application/vnd.gravitino.v1+json" \
-H "Content-Type: application/json" -d '{
  "updates": [
    {
      "@type": "removeProperty",
      "property": "key2"
    }, {
      "@type": "setProperty",
      "property": "key3",
      "value": "value3"
    }
  ]
}' http://localhost:8090/api/metalakes/metalake/catalogs/catalog/schemas/schema/filesets/fileset

Java

// ...
// Assuming you have just created a Fileset catalog named `catalog`
Catalog catalog = gravitinoClient.loadCatalog(NameIdentifier.of("metalake", "catalog"));

FilesetCatalog filesetCatalog = catalog.asFilesetCatalog();

Fileset f = filesetCatalog.alterFileset(NameIdentifier.of("metalake", "catalog", "schema", "fileset"),
    FilesetChange.rename("fileset_renamed"), FilesetChange.updateComment("xxx"));
// ...

Python

gravitino_client: GravitinoClient = GravitinoClient(uri="http://localhost:8090", metalake_name="metalake")

catalog: Catalog = gravitino_client.load_catalog(ident=NameIdentifier.of("metalake", "catalog"))
changes = (
   FilesetChange.remove_property("fileset_properties_key1"),
   FilesetChange.set_property("fileset_properties_key2", "fileset_propertie_new_value"),
)
fileset_new = catalog.as_fileset_catalog().alter_fileset(NameIdentifier.of("metalake", "catalog", "schema", "fileset"), 
                                                         *changes)

Currently, Gravitino supports the following changes to a fileset:

Supported modification	JSON	Java
Rename a fileset	{"@type":"rename","newName":"fileset_renamed"}	FilesetChange.rename("fileset_renamed")
Update a comment	{"@type":"updateComment","newComment":"new_comment"}	FilesetChange.updateComment("new_comment")
Set a fileset property	{"@type":"setProperty","property":"key1","value":"value1"}	FilesetChange.setProperty("key1", "value1")
Remove a fileset property	{"@type":"removeProperty","property":"key1"}	FilesetChange.removeProperty("key1")

Drop a fileset

You can remove a fileset by sending a DELETE request to the /api/metalakes/{metalake_name} /catalogs/{catalog_name}/schemas/{schema_name}/filesets/{fileset_name} endpoint or by using the Gravitino Java client. The following is an example of dropping a fileset:

Shell

curl -X DELETE -H "Accept: application/vnd.gravitino.v1+json" \
-H "Content-Type: application/json" \
http://localhost:8090/api/metalakes/metalake/catalogs/catalog/schemas/schema/filesets/fileset

Java

// ...
// Assuming you have just created a Fileset catalog named `catalog`
Catalog catalog = gravitinoClient.loadCatalog(NameIdentifier.of("metalake", "catalog"));

FilesetCatalog filesetCatalog = catalog.asFilesetCatalog();

// Drop a fileset
filesetCatalog.dropFileset(NameIdentifier.of("metalake", "catalog", "schema", "fileset"));
// ...

Python

gravitino_client: GravitinoClient = GravitinoClient(uri="http://localhost:8090", metalake_name="metalake")

catalog: Catalog = gravitino_client.load_catalog(ident=NameIdentifier.of("metalake", "catalog"))
catalog.as_fileset_catalog().drop_fileset(ident=NameIdentifier.of("metalake", "catalog", "schema", "fileset"))

For a MANAGED fileset, the physical location of the fileset will be deleted when this fileset is dropped. For EXTERNAL fileset, only the metadata of the fileset will be removed.

List filesets

You can list all filesets in a schema by sending a GET request to the /api/metalakes/ {metalake_name}/catalogs/{catalog_name}/schemas/{schema_name}/filesets endpoint or by using the Gravitino Java client. The following is an example of listing all the filesets in a schema:

Shell

curl -X GET -H "Accept: application/vnd.gravitino.v1+json" \
-H "Content-Type: application/json" \
http://localhost:8090/api/metalakes/metalake/catalogs/catalog/schemas/schema/filesets

Java

// ...
Catalog catalog = gravitinoClient.loadCatalog(NameIdentifier.of("metalake", "catalog"));

FilesetCatalog filesetCatalog = catalog.asFilesetCatalog();
NameIdentifier[] identifiers =
    filesetCatalog.listFilesets(Namespace.ofFileset("metalake", "catalog", "schema"));
// ...

Python

gravitino_client: GravitinoClient = GravitinoClient(uri="http://localhost:8090", metalake_name="metalake")

catalog: Catalog = gravitino_client.load_catalog(ident=NameIdentifier.of("metalake", "catalog"))
fileset_list: List[NameIdentifier] = catalog.as_fileset_catalog().list_filesets(namespace=Namespace.of_fileset("metalake", "catalog", "schema")))