I haven't had much luck when pipelining the format and mode options.   I've been doing it like this instead. I'm using python though not scala.   dataFrame.write.saveAsTable("tableName", format="parquet", mode="overwrite") The issue I'm having isn't that it won't create the table or write the data using saveAsTable, its that spark doesn't see any data in the the table if I go back and try to read it later. I can do queries on it using Hive without an issue. If I use hive and remove the spark.sql.sources.provider option from tblproperties then I can read the table with spark properly.  ... View more