Ok, Single Node didn't work because of Azure's laggy backend.
Single Node works, also with Spark; creating a zero node "cluster" will not execute Spark functions (as is mentioned on Create Cluster page).
Huh, Single Node "cluster" did not work, but creating a "cluster" with 0 to 1 nodes apparently creates a driver and no nodes(?). I also defined to following to make sure only one core is being requested:
spark.executor.cores 1
spark.executor.memory ...
Ok, got it to work using a single node "cluster" that doesn't require a driver. This does not, however, conform to the training material, and it will of course not enable users to test multiple node processing.
Requesting quota changes is not possible with Azure Trial subscription. We need 2 core cluster nodes and driver in order to run Databricks trial and complete Databricks training within the hard 4 core Azure Trial limit.
@Databricks_Support How can w...