Login API#

[source]

login#

hopsworks.login(
    host=None,
    port=443,
    project=None,
    api_key_value=None,
    api_key_file=None,
    hostname_verification=False,
    trust_store_path=None,
    engine=None,
)

Connect to Serverless Hopsworks by calling the hopsworks.login() function with no arguments.

import hopsworks

project = hopsworks.login()

Alternatively, connect to your own Hopsworks installation by specifying the host, port and api key.

import hopsworks

project = hopsworks.login(host="my.hopsworks.server",
                          port=8181,
                          api_key_value="DKN8DndwaAjdf98FFNSxwdVKx")

In addition to setting function arguments directly, hopsworks.login() also reads the environment variables: HOPSWORKS_HOST, HOPSWORKS_PORT, HOPSWORKS_PROJECT, HOPSWORKS_API_KEY, HOPSWORKS_HOSTNAME_VERIFICATION and HOPSWORKS_TRUST_STORE_PATH.

The function arguments do however take precedence over the environment variables in case both are set.

Arguments

• host str | None: The hostname of the Hopsworks instance, defaults to None.
• port int: The port on which the Hopsworks instance can be reached, defaults to 443.
• project str | None: Name of the project to access. If used inside a Hopsworks environment it always gets the current project. If not provided you will be prompted to enter it.
• api_key_value str | None: Value of the Api Key
• api_key_file str | None: Path to file wih Api Key
• hostname_verification bool: Whether to verify Hopsworks' certificate
• trust_store_path str | None: Path on the file system containing the Hopsworks certificates
• engine None | Literal['spark'] | Literal['python'] | Literal['training'] | Literal['spark-no-metastore'] | Literal['spark-delta']: Specifies the engine to use. Possible options are "spark", "python", "training", "spark-no-metastore", or "spark-delta". The default value is None, which automatically selects the engine based on the environment: "spark": Used if Spark is available, such as in Hopsworks or Databricks environments. "python": Used in local Python environments or AWS SageMaker when Spark is not available. "training": Used when only feature store metadata is needed, such as for obtaining training dataset locations and label information during Hopsworks training experiments. "spark-no-metastore": Functions like "spark" but does not rely on the Hive metastore. "spark-delta": Minimizes dependencies further by avoiding both Hive metastore and HopsFS.

Returns

Project: The Project object to perform operations on

Raises

• RestAPIError: If unable to connect to Hopsworks
• HopsworksSSLClientError: If SSLError is raised from underlying requests library

[source]

get_current_project#

hopsworks.get_current_project()

Get a reference to the current logged in project.

import hopsworks

hopsworks.login()

project = hopsworks.get_current_project()

Returns

Project. The Project object to perform operations on

Feature Store API#

[source]

get_feature_store#

Project.get_feature_store(name=None)

Connect to Project's Feature Store.

Defaulting to the project name of default feature store. To get a shared feature store, the project name of the feature store is required.

import hopsworks

project = hopsworks.login()

fs = project.get_feature_store()

Arguments

• name str | None: Project name of the feature store.

Returns

hsfs.feature_store.FeatureStore: The Feature Store API

Raises

• RestAPIError: If unable to connect

Model Registry API#

[source]

get_model_registry#

Project.get_model_registry()

Connect to Project's Model Registry API.

import hopsworks

project = hopsworks.login()

mr = project.get_model_registry()

Returns

hsml.model_registry.ModelRegistry: The Model Registry API

Raises

• RestAPIError: If unable to connect

Model Serving API#

[source]

get_model_serving#

Project.get_model_serving()

Connect to Project's Model Serving API.

import hopsworks

project = hopsworks.login()

ms = project.get_model_serving()

Returns

hsml.model_serving.ModelServing: The Model Serving API

Raises

• RestAPIError: If unable to connect