Adding an Atlas Connection
Prerequisites
In order to establish a connection to an Atlas data catalog, the user must have sufficient permissions.
Zeenea traffic flows towards Atlas must be open.
Supported Versions
This connector may be used on an Atlas catalog linked through its v2 API to a Hadoop cluster.
Installing the Plugin
Since scanner version 26.0, the Atlas plugin can be downloaded here: Connectors: download links
Declaring the Connection
Creating and configuring connectors is done through a dedicated configuration file located in the /connections folder of the relevant scanner.
In order to establish a connection with Atlas, specifying the following parameters in the dedicated file is required:
Parameter | Expected value |
|---|
name | The name that will be displayed to catalog users for this connection. |
code | The unique identifier of the connection on the Zeenea platform. Once registered on the platform, this code must not be modified or the connection will be considered as new and the old one removed from the scanner. |
connector_id | The type of connector to be used for the connection. Here, the value must be Atlas and this value must not be modified. |
connection.url | URL used to connect to the application |
connection.username | Atlas user used to log into Zeenea. Must have sufficient permissions |
connection.password | User password |
connection.authentication_scheme | Method to be used for authentication with the Atlas server. Possible values: Basic, Digest, Ntlm, Spnego, Kerberos. Default value: Basic. |
connection.max_concurrent_request | Maximum concurrent requests to Atlas (default: 16) |
connection.request_timeout | Request timeout in seconds (default: 60 s) |
connection.request_page_size | Request page size (default: 100) |
api.min_ext_info.atlas_types | Allows adding the minExtInfo parameter when calling the Atlas entity retrieval service. Allowed values : hive_table, hbase_table, hive_process, hdfs_path, spark_process, sqoop_process, or sqoop_dbdatastore. |
cache.path | Directory where caches are stored |
cache.inventory.enable | Enable Inventory cache |
cache.inventory.expiration_delay | Inventory Cache expiration Delay (format ISO-8601: PnDTnHnMn.nS) |
Some additional settings, dedicated to improving performance, are available in the Atlas connector configuration file.
Finally, the connector can be configured to filter items (by type) during inventory and import. Without specific configuration, all items are taken into account. Both parameters must have as their value a list of types, separated by a comma, between double-quotes. Example: "hdfs_path, sqoop_process, sqoop_dbdatastore".
The available types are listed below:
inventoried types
hive_process
spark_process
sqoop_process
sqoop_dbdatastore
Parameter | Value |
|---|
filter.atlas_type.includes | Inclusion list. All types included if not defined. |
filter.atlas_type.excludes | Exclusion list. No exclusion if not defined. |
Extracting Metadata
In order to gather metadata from Atlas, the connector first references datasets and fields, but it will also retrieve the Data Process used in Atlas to build a Dataset. This allows Zeenea to build the entirety of the process, and the lineage.
Important: The Atlas connector behavior is very specific, and must be taken into account when deploying the connector. When a dataset is imported or updated, the connector will also import:
The selected Dataset
Any Data Process of which the selected dataset is an output
Any Data Process of which the selected Dataset is an input
Note that, in the case of updates, they are by default scheduled to be done daily.
Collected Metadata
Inventory
The inventory collects the list of data sources that the user can access.
Dataset
The Datasets are data sources.
Name
Source Description
Technical Data:
GUID: Atlas technical id
Qualified Name: Atlas technical name
Created: creation date
Created By
Data Source Type: Data source type
Owner
Updated: date of last update
Updated By
Database: optional
Table: optional
Display Name: optional
Field
Name
Source Description
Type
Can be null: false (default)
Multivalued: true when the field represents a table, false otherwise
Primary Key: false (default)
Technical Metadata: none
Data Process
All documented lineages in Atlas.
Object Identification Keys
An identification key is associated with each object in the catalog. In the case of the object being created by a connector, the connector builds it.
Object | Identification Key | Description |
|---|
Dataset | code/type/dataset identifier | - code: Unique identifier of the connection noted in the configuration file - type: Object type, possible values: hbase_table, hdfs_path, hive_table - dataset identifier: Atlas attribute qualifiedName |
Field | code/type/dataset identifier/field name | - code: Unique identifier of the connection noted in the configuration file - type: Object type, possible values: hbase_table, hdfs_path, hive_table - dataset identifier: Atlas attribute qualifiedName - field name |
Data process | code/type/data process identifier | - code: Unique identifier of the connection noted in the configuration file - type: Object type, possible values: hive_process, spark_process, sqoop_process - data process identifier: Technical data process identifier calculated from data process outputs |
Last modified date: 11/28/2025