Configuring Authorization with Apache Ranger

DataHub integration with Apache Ranger allows DataHub Authorization policies to be controlled inside Apache Ranger. Admins can create users, groups and roles on Apache Ranger, and then assign them to Ranger policies to control the authorization of requests to DataHub.

We'll break down configuration of the DataHub Apache Ranger Plugin into two parts:

1. Configuring your Apache Ranger Deployment
2. Configuring your DataHub Deployment

Disclaimer: All configurations shown in this documented were tested against Privacera Platform v6.3.0.1.

Prerequisites

• User identifier present in CorpRole URN should be the name of the AD/LDAP user. For example in URN urn:li:corpuser:datahub, the datahub should present as name of user in AD/LDAP
• Apache Ranger and DataHub are configured for authentication via same IDP (either LDAP + JaaS or OIDC SSO)
• Apache Ranger service available via HTTP
• Basic authentication is enabled on Apache Ranger Service

Configuration

Configuring your Apache Ranger Deployment

Perform the following steps to configure an Apache Ranger deployment to support creating access policies compatible with DataHub. For kubernetes example command, please replace the <ranger-pod-name> and <namespace> as per your environment.

1. Download the datahub-ranger-plugin from Maven

2. Create a "datahub" directory inside the "ranger-plugins" directory where Apache Ranger is deployed. For example, to do this in a Privacera container

   Docker command:

   bashCopy

   docker exec privacera_ranger_1 mkdir ews/webapp/WEB-INF/classes/ranger-plugins/datahub
   

   Kubernetes command:

   bashCopy

   kubectl exec <ranger-pod-name> mkdir ews/webapp/WEB-INF/classes/ranger-plugins/datahub -n <namespace>
   

3. Copy the downloaded datahub-ranger-plugin jar into the newly created "datahub" directory. For example, to do this in a Privacera container

   Docker command:

   bashCopy

   docker cp datahub-ranger-plugin-<version>.jar privacera_ranger_1:/opt/ranger/ranger-2.1.0-admin/ews/webapp/WEB-INF/classes/ranger-plugins/datahub/
   

   Kubernetes command:

   bashCopy

   kubectl cp datahub-ranger-plugin-<version>.jar <ranger-pod-name>:/opt/ranger/ranger-2.1.0-admin/ews/webapp/WEB-INF/classes/ranger-plugins/datahub/ -n <namespace>
   

4. Download the service definition file. This service definition is the ranger service definition JSON file for datahub-ranger-plugin-<version>.jar

5. Register the downloaded service definition file with Apache Ranger Service. To do this executes the below curl command

   Replace variables with corresponding values in curl command

   • <ranger-admin-username>

   • <ranger-admin-password>

   • <ranger-host>

     bashCopy

     curl -u <ranger-admin-username>:<ranger-admin-password> -X POST -H "Accept: application/json" -H "Content-Type: application/json" --data @servicedef.json http://<ranger-host>:6080/service/public/v2/api/servicedef
     

Defining a Ranger Policy

Now, you should have the DataHub plugin registered with Apache Ranger. Next, we'll create a sample user and add them to our first resource policy.

1. Login into the Apache Ranger UI (Privacera Portal) to performs below steps.

2. Verify datahub-ranger-plugin is registered successfully: The datahub-ranger-plugin should be visible as DATAHUB in Access Management -> Resource Policies.

3. Create a service under the plugin DATAHUB with name ranger_datahub

   DATAHUB plugin and ranger_datahub service is shown in below screenshot:

<p align="center"> </p>

4. Create a new policy under service ranger_datahub - this will be used to control DataHub authorization.

5. Create a test user & assign them to a policy. We'll use the datahub user, which is the default root user inside DataHub.

   To do this performs below steps

   • Create a user datahub
   • Create a policy under ranger_datahub service. To assign Platform Privileges (e.g. Admin privileges), simply use the "platform" resource type which is defined. To test the flow, we can simply assign the datahub user all platform privileges that are available through the Ranger UI. This will enable the "datahub" to have full platform admin privileges.

   To define fine-grained resource privileges, e.g. for DataHub Datasets, Dashboards, Charts, and more, you can simply select the appropriate Resource Type in the Ranger policy builder. You should also see a list of privileges that are supported for each resource type, which correspond to the actions that you can perform. To learn more about supported privileges, check out the DataHub Policies Guide.

   DataHub platform access policy screenshot:

<p align="center"> </p>

Once we've created our first policy, we can set up DataHub to start authorizing requests using Ranger policies.

Configuring your DataHub Deployment

Perform the following steps to configure DataHub to send incoming requests to Apache Ranger for authorization.

1. Download Apache Ranger security xml ranger-datahub-security.xml
2. In ranger-datahub-security.xml edit the value of property ranger.plugin.datahub.policy.rest.url. Sample snippet is shown below
   xmlCopy

       <property>
           <name>ranger.plugin.datahub.policy.rest.url</name>
           <value>http://199.209.9.70:6080</value>
           <description>
               URL to Ranger Admin
           </description>
       </property>
   

As per your deployment follow either Docker or Kubernetes section below

Docker

Build Ranger Authorizer Plugin

1. Clone DataHub Repo: Clone the DataHub repository

   shellCopy

       cd ~/
       git clone https://github.com/acryldata/datahub-ranger-auth-plugin.git
   

2. Go inside the datahub directory: You should be inside the datahub-ranger-auth-plugin directory to execute build command

   shellCopy

       cd ~/datahub-ranger-auth-plugin/
   

3. Build plugin: Execute below gradle command to build Ranger Authorizer Plugin jar

   shellCopy

     ./gradlew apache-ranger-plugin:shadowJar
   

   This step will generate a jar file i.e. ./apache-ranger-plugin/build/libs/apache-ranger-plugin-<version>-SNAPSHOT.jar.

   Let's call this jar as ranger-plugin-jar. We need this jar in below step (Configure Ranger Authorizer Plugin)

Configure Ranger Authorizer Plugin

On the host where datahub-gms is deployed, follow these steps:

1. Create directory ~/.datahub/plugins/auth/apache-ranger-authorizer/: Executes below command
   bashCopy

   mkdir -p ~/.datahub/plugins/auth/apache-ranger-authorizer/
   

2. Copy ranger-datahub-security.xml file to ~/.datahub/plugins/auth/apache-ranger-authorizer/
3. Copy ranger-plugin-jar: Copy the apache-ranger-plugin-<version>-SNAPSHOT.jar
   bashCopy

   cp ./apache-ranger-plugin/build/libs/apache-ranger-plugin-<version>-SNAPSHOT.jar ~/.datahub/plugins/auth/apache-ranger-authorizer/apache-ranger-authorizer.jar
   

4. Create config.yml: Create config.yml if not exist
   shellCopy

       touch ~/.datahub/plugins/auth/config.yml
   

5. Set Apache Ranger Plugin config: Add below entry in config.yml file. Set username and password to Apache Ranger user credentials
   yamlCopy

   plugins:
     - name: "apache-ranger-authorizer"
       type: "authorizer"
       enabled: "true"
       params:
         className: "com.datahub.authorization.ranger.RangerAuthorizer"
         configs:
           username: "<Apache Ranger username>"
           password: "<Apache Ranger password>"
   

6. Restart DataHub GMS container (i.e. datahub-gms)

Kubernetes

Helm support is coming soon.

That's it! Now we can test out the integration.

Validating your Setup

To verify that things are working as expected, we can test that the root datahub user has all Platform Privileges and is able to perform all operations: managing users & groups, creating domains, and more. To do this, simply log into your DataHub deployment via the root DataHub user.

Revert the Configuration

If you want to revert your deployment configuration and don't want Apache Ranger to control the authorization of your DataHub deployment then follow the below sections to undo the configuration steps you have performed in section Configuring Authorization with Apache Ranger

1. Revert Configuration of your Apache Ranger Deployment
2. Revert Configuration of your DataHub Deployment

Revert Configuration of your Apache Ranger Deployment

For kubernetes example command, please replace the <ranger-pod-name> and <namespace> as per your environment.

1. Delete ranger_datahub service: Login into the Privacera Portal and delete service ranger_datahub

   ranger_datahub service is shown in below screenshot:

<p align="center"> </p>

2. Delete datahub plugin: Execute below curl command to delete datahub plugin Replace variables with corresponding values in curl command

   • <ranger-admin-username>
   • <ranger-admin-password>
   • <ranger-host>
   bashCopy

   curl -u <ranger-admin-username>:<ranger-admin-password> -X DELETE -H "Accept: application/json" -H "Content-Type: application/json" http://<ranger-host>:6080/service/public/v2/api/servicedef/name/datahub
   

3. Delete datahub plugin directory: Execute below command to delete the datahub plugin directory from Apache Ranger

   Docker command:

   bashCopy

   docker exec privacera_ranger_1 rm -rf ews/webapp/WEB-INF/classes/ranger-plugins/datahub
   

   Kubernetes command:

   bashCopy

   kubectl exec <ranger-pod-name> -n <namespace> -- sh -c 'rm -rf ews/webapp/WEB-INF/classes/ranger-plugins/datahub'
   

Revert Configuration of your DataHub Deployment

Docker

1. Remove Apache Ranger Plugin entry: From config.yml file remove the entry which was added for Apache Ranger Plugin
2. Redeploy DataHub (datahub-gms)

Kubernetes

For kubernetes example command, please replace the <namespace> as per your environment.

1. Open deployment editor: Execute below command
   bashCopy

     kubectl edit deployment datahub-datahub-gms -n <namespace>
   

2. Remove below environments variables
   1. AUTH_POLICIES_ENABLED
   2. RANGER_AUTHORIZER_ENABLED
   3. RANGER_USERNAME
   4. RANGER_PASSWORD
3. Remove below volumes related settings
   1. volumes
   2. volumeMounts
4. Save and quit the editor and use below command to check status of datahub-datahub-gms deployment rollout
   bashCopy

   kubectl rollout status deployment/datahub-datahub-gms -n <namespace>
   

   On successful rollout you should see a message deployment "datahub-datahub-gms" successfully rolled out

Validating your Setup

To verify that things are working as expected, we can test that the root datahub user has all Platform Privileges and is able to perform all operations: managing users & groups, creating domains, and more. To do this, simply log into your DataHub deployment via the root DataHub user.