Search code examples
azureazure-data-factoryazure-synapseazure-mapping-data-flow

Which Spark version is being used by Mapping Data Flows?


I ran into some issues recently which lead me to believe that the Mapping Data Flows used in my Synapse Analytics Workspace are still running Spark 2.4, which has reached EOL. I tried to find out which Spark version is being used and how to change it, without any success.

Is there a way to migrate to a newer version or to at least verify which Spark version is being used under the covers?

The concrete reasons for me caring revolve around the breaking changes regarding date/time handling in Spark 3.x.


Solution

  • You could figure out the version of spark through Monitoring output. Migration to Spark 3.x is already underway. And your subscription should be migrated to 3.x soon (if its not already migrated)

    You can find the information in the output details of a data flow activity. Spark Activity Output