Azure (dagster_azure)

Utilities for using Azure Storage Accounts with Dagster. This is mostly aimed at Azure Data Lake Storage Gen 2 (ADLS2) but also contains some utilities for Azure Blob Storage.


NOTE: This package is incompatible with dagster-snowflake! This is due to a version mismatch between the underlying azure-storage-blob package; dagster-snowflake has a transitive dependency on an old version, via snowflake-connector-python.

dagster_azure.adls2.adls2_resource ResourceDefinition[source]

Resource that gives solids access to Azure Data Lake Storage Gen2.

The underlying client is a DataLakeServiceClient.

Attach this resource definition to a ModeDefinition in order to make it available to your solids.

Example

from dagster import ModeDefinition, execute_solid, solid
from dagster_azure.adls2 import adls2_resource

@solid(required_resource_keys={'adls2'})
def example_adls2_solid(context):
    return list(context.resources.adls2.list_file_systems())

result = execute_solid(
    example_adls2_solid,
    run_config={
        'resources': {
            'adls2': {
                'config': {
                    'storage_account': 'my_storage_account'
                }
            }
        }
    },
    mode_def=ModeDefinition(resource_defs={'adls2': adls2_resource}),
)

Note that your solids must also declare that they require this resource with required_resource_keys, or it will not be initialized for the execution of their compute functions.

You may pass credentials to this resource using either a SAS token or a key, using environment variables if desired:

resources:
  adls2:
    config:
      storage_account: my_storage_account
      # str: The storage account name.
      credential:
        sas: my_sas_token
        # str: the SAS token for the account.
        key:
          env: AZURE_DATA_LAKE_STORAGE_KEY
        # str: The shared access key for the account.
class dagster_azure.adls2.FakeADLS2Resource(account_name, credential='fake-creds')[source]

Stateful mock of an ADLS2Resource for testing.

Wraps a mock.MagicMock. Containers are implemented using an in-memory dict.

dagster_azure.adls2.adls2_file_cache ResourceDefinition[source]
dagster_azure.adls2.adls2_intermediate_storage IntermediateStorageDefinition[source]

Persistent intermediate storage using Azure Data Lake Storage Gen2 for storage.

Suitable for intermediates storage for distributed executors, so long as each execution node has network connectivity and credentials for ADLS and the backing container.

Attach this intermediate storage definition, as well as the adls2_resource it requires, to a ModeDefinition in order to make it available to your pipeline:

pipeline_def = PipelineDefinition(
    mode_defs=[
        ModeDefinition(
            resource_defs={'adls2': adls2_resource, ...},
            intermediate_storage_defs=[adls2_intermediate_storage],
            ...
        ), ...
    ], ...
)

You may configure this storage as follows:

intermediate_storage:
  adls2:
    config:
      adls2_sa: my-best-storage-account
      adls2_file_system: my-cool-file-system
      adls2_prefix: good/prefix-for-files-
class dagster_azure.blob.AzureBlobComputeLogManager(storage_account, container, secret_key, local_dir=None, inst_data=None, prefix='dagster')[source]

Logs solid compute function stdout and stderr to Azure Blob Storage.

This is also compatible with Azure Data Lake Storage.

Users should not instantiate this class directly. Instead, use a YAML block in dagster.yaml such as the following:

compute_logs:
  module: dagster_azure.blob.compute_log_manager
  class: AzureBlobComputeLogManager
  config:
    storage_account: my-storage-account
    container: my-container
    credential: sas-token-or-secret-key
    prefix: "dagster-test-"
    local_dir: "/tmp/cool"
Parameters
  • storage_account (str) – The storage account name to which to log.

  • container (str) – The container (or ADLS2 filesystem) to which to log.

  • secret_key (str) – Secret key for the storage account. SAS tokens are not supported because we need a secret key to generate a SAS token for a download URL.

  • local_dir (Optional[str]) – Path to the local directory in which to stage logs. Default: dagster.seven.get_system_temp_directory().

  • prefix (Optional[str]) – Prefix for the log file keys.

  • inst_data (Optional[ConfigurableClassData]) – Serializable representation of the compute log manager when newed up from config.

dagster_azure.adls2.adls2_pickle_io_manager IOManagerDefinition[source]

Persistent IO manager using Azure Data Lake Storage Gen2 for storage.

Serializes objects via pickling. Suitable for objects storage for distributed executors, so long as each execution node has network connectivity and credentials for ADLS and the backing container.

Attach this resource definition to a ModeDefinition in order to make it available to your pipeline:

pipeline_def = PipelineDefinition(
    mode_defs=[
        ModeDefinition(
            resource_defs={
                'io_manager': adls2_pickle_io_manager,
                'adls2': adls2_resource, ...},
        ), ...
    ], ...
)

You may configure this storage as follows:

resources:
    io_manager:
        config:
            adls2_file_system: my-cool-file-system
            adls2_prefix: good/prefix-for-files-