Hi @stephen_obrien As discussed via Cloudera case, there is a performance bottleneck when connecting via knox ( tracked in internal jira ) than directly from phoenix-sqlline from the edge node.   You can test the same when the runtime version 7.2.17 is released. ... View more

To provide a module with custom Python functions that are declared as UDFs, one must specify: spark_session.sparkContext.addPyFile("/app/mount/python_utils.py")   This file should be included in a resource attached to the job.   See this post for further examples: https://blog.cloudera.com/managing-python-dependencies-for-spark-workloads-in-cloudera-data-engineering/   ... View more