Skip to content

Flask configuration

After modifying any variable in config.py described in this document, be sure to rebuild your application with these changes.

NOTE: This document is a work in progress and does not include 100% of features. We welcome PRs to complete this document

Custom Routes

In order to add any custom Flask endpoints to Amundsen’s frontend application, configure a function on the INIT_CUSTOM_ROUTES variable. This function takes the created Flask application and can leverage Flask’s add_url_rule method to add custom routes.

Example: Setting INIT_CUSTOM_ROUTES to the init_custom_routes method below will expose a /custom_route endpoint on the frontend application.

def init_custom_routes(app: Flask) -> None:
  app.add_url_rule('/custom_route', 'custom_route', custom_route)

def custom_route():
  pass

Dashboard Preview

This service provides an API to download preview image of dashboard resources, which currently only supports Mode.

The dashboard preview image is cached in user’s browser up to a day. In order to adjust this you can change the value of DASHBOARD_PREVIEW_IMAGE_CACHE_MAX_AGE_SECONDS

How to configure Mode dashboard preview

Add the following environment variables:

    CREDENTIALS_MODE_ADMIN_TOKEN
    CREDENTIALS_MODE_ADMIN_PASSWORD
    MODE_ORGANIZATION

How to enable authorization on Mode dashboard

By default, Amundsen does not do any authorization on showing preview. By registering name of the Mode preview class in configuration, you can enable authorization.

    ACL_ENABLED_DASHBOARD_PREVIEW = {'ModePreview'}
Amundsen ingests Mode dashboards only from the shared space, which all registered Mode users are able to view. Therefore our authorization first validates if the current user is registered in Mode. This feature is dependent on Amundsen also ingesting Mode user information via the ModeDashboardUserExtractor and the metadata service version must be at least v2.5.2.

How to support preview of different product?

You can add preview support for different products by adding its preview class to DefaultPreviewMethodFactory In order to develop new preview class, you need to implement the class that inherits BasePreview class and ModePreview would be a great example.

Mail Client Features

Amundsen has two features that leverage the custom mail client – the feedback tool and notifications. For these features a custom implementation of base_mail_client must be mapped to the MAIL_CLIENT configuration variable.

To fully enable these features in the UI, the application configuration variables for these features must also be set to true. Please see this entry in our application configuration doc for further information.

Issue Tracking Integration Features

Amundsen has a feature to allow display of associated tickets within the table detail view. The feature both displays open tickets and allows users to report new tickets associated with the table. These tickets must contain the table_uri within the ticket text in order to be displayed; the table_uri is automatically added to tickets created via the feature. Tickets are displayed from most recent to oldest, and currently only open tickets are displayed. Currently only JIRA is supported. The UI must also be enabled to use this feature, please see configuration notes here.

There are several configuration settings in config.py that should be set in order to use this feature.

Here are the settings and what they should be set to

    ISSUE_LABELS = []  # type: List[str] (Optional labels to be set on the created tickets)
    ISSUE_TRACKER_URL = None  # type: str (Your JIRA environment, IE 'https://jira.net')
    ISSUE_TRACKER_USER = None  # type: str (Recommended to be a service account)
    ISSUE_TRACKER_PASSWORD = None  # type: str
    ISSUE_TRACKER_PROJECT_ID = None  # type: int (Project ID for the project you would like JIRA tickets to be created in)
    ISSUE_TRACKER_CLIENT = None  # type: str (Fully qualified class name and path)
    ISSUE_TRACKER_CLIENT_ENABLED = False  # type: bool (Enabling the feature, must be set to True)
    ISSUE_TRACKER_MAX_RESULTS = None  # type: int (Max issues to display at a time)
    ISSUE_TRACKER_ISSUE_TYPE_ID = None # type: int (Jira only: Override default issue tracker ID whenever needed for cloud/hosted deployments)

Programmatic Descriptions

Amundsen supports configuring other mark down supported non-editable description boxes on the table page. This can be useful if you have multiple writers which want to write different pieces of information to Amundsen that are either very company specific and thus would never be directly integrated into Amundsen or require long form text to properly convey the information.

What are some more specific examples of what could be used for this? - You have an existing process that generates quality reports for a dataset that you want to embed in the table page. - You have a process that detects pii information (also adding the appropriate tag/badge) but also generates a simple report to provide context. - You have extended table information that is applicable to your datastore which you want to scrape and provide in the table page

Programmatic descriptions are referred to by a “description source” which is a unique identifier. In the UI, they will appear on the table page under structured metadata.

In config.py you can then configure the descriptions to have a custom order, as well as whether or not they should exist in the left column or right column.

PROGRAMMATIC_DISPLAY = {
    'RIGHT': {
      "test3" : {},
      "test2" : { "display_order": 0 }
    },
    'LEFT': {
      "test1" : { "display_order": 1 },
      "test0" : { "display_order": 0 },
    },
    'test4': {"display_order": 0},
}
Description sources not mentioned in the configuration will be alphabetically placed at the bottom of the list. If PROGRAMMATIC_DISPLAY is left at None all added fields will show up in the order in which they were returned from the backend. Here is a screenshot of what it would look like in the bottom left:

Uneditable Table Descriptions

Amundsen supports configuring table and column description to be non-editable for selective tables. You may want to make table descriptions non-editable due to various reasons such as table already has table description from source of truth. You can define matching rules in config.py for selecting tables. This configuration is useful as table selection criteria can be company specific which will not directly integrated with Amundsen. You can use different combinations of schema and table name for selecting tables.

Here are some examples when this feature can be used: 1. You want to set ALL tables in your application as un-editable 2. You want to set all tables with a given schema or schema pattern as un-editable. 3. You want to set all tables with a specific table name pattern in a given schema pattern as un-editable. 4. You want to set all tables with a given table name pattern as un-editable.

Amundsen has two variables in config.py file which can be used to define match rules: 1. ALL_UNEDITABLE_SCHEMAS : boolean on/off switch for ability to edit tables. This can also be set using the environment variable ‘ALL_UNEDITABLE_SCHEMAS’ 2. UNEDITABLE_SCHEMAS : Set of schemas where all tables should be un-editable. It takes exact schema name. 3. UNEDITABLE_TABLE_DESCRIPTION_MATCH_RULES : List of MatchRuleObject, where each MatchRuleObject consists of regex for schema name or regex for table name or both.

Purpose of ALL_UNEDITABLE_SCHEMAS is to provide a blanket on/off switch for disabling editing schemas. This should be used if you plan to up all of your schemas via databuilder rather then allow users to do it via the UI.

Purpose of UNEDITABLE_SCHEMAS can be fulfilled by UNEDITABLE_TABLE_DESCRIPTION_MATCH_RULES but we are keeping both variables for backward compatibility. If you want to restrict tables from a given schemas then you can use UNEDITABLE_SCHEMAS as follows:

UNEDITABLE_SCHEMAS = set(['schema1', 'schema2'])
After above configuration, all tables in ‘schema1’ and ‘schema2’ will have non-editable table and column descriptions.

If you have more complex matching rules you can use UNEDITABLE_TABLE_DESCRIPTION_MATCH_RULES. It provides you more flexibility and control as you can create multiple match rules and use regex for matching schema and table names.

You can configure your match rules in config.py as follow:

UNEDITABLE_TABLE_DESCRIPTION_MATCH_RULES = [
        # match rule for all table in schema1
        MatchRuleObject(schema_regex=r"^(schema1)"),
        # macth rule for all tables in schema2 and schema3
        MatchRuleObject(schema_regex=r"^(schema2|schema3)"),
        # match rule for tables in schema4 with table name pattern 'noedit_*'
        MatchRuleObject(schema_regex=r"^(schema4)", table_name_regex=r"^noedit_([a-zA-Z_0-9]+)"),
        # match rule for tables in schema5, schema6 and schema7 with table name pattern 'noedit_*'
        MatchRuleObject(schema_regex=r"^(schema5|schema6|schema7)", table_name_regex=r"^noedit_([a-zA-Z_0-9]+)"),
        # match rule for all tables with table name pattern 'others_*'
        MatchRuleObject(table_name_regex=r"^others_([a-zA-Z_0-9]+)")
    ]

After configuring this, users will not be able to edit table and column descriptions of any table matching above match rules from UI.