Self-managed installation and maintenance > Installation and configuration guide > DataRobot advanced configuration section > Distributed Feature Discovery

Distributed Feature Discovery¶

Distributed feature discovery is an optional feature and can be enabled on-demand. Datarobot currently supports the feature only on AWS installations. Datarobot uses AWS EMR Serverless 7.3.0 base docker image that currently exposed to multiple CVEs: https://docs.aws.amazon.com/emr/latest/ReleaseGuide/emr-730-common-vulnerabilities-exposures.html . Please, make sure that it complies with company policies.

Dependencies¶

All requirements are satisfied to enable Spark Batch Feature in compute spart service
The image is presented in ECR as described here
Configure env as shown in the table below

Configuration Values¶

To configure these options, refer to the Tuning Datarobot Environment Variables section of this guide.

Config	Description	Default
`CSP_SPARK_SAFER_CUSTOM_IMAGE_HOST`	ECR registry used to store spark-batch-image. Example: 11122233334444.dkr.ecr.us-east-1.amazonaws.com	None
`CSP_SPARK_SAFER_CUSTOM_IMAGE_REPO`	Repository name in ECR	spark-batch-image
`ENABLE_SAFER_DISTRIBUTED_MODE`	Flag to enable distributed feature discovery	false

Enable feature¶

Feature can be enabled per user:

Login to application
Click on Settings icon
Select Feature Access
Search for Enable Feature Discovery in Distributed Mode and make sure it's enabled.