I wonder if anyone ever tried rcongiu's Json Serde to define an external table on a partitioned Hive table.
If I define a non-partitioned external table setting the location to one of the partition folders, then it returns records as expected (for that single partition, of course).
However, if I set the parent folder for all partitions as the location for the partitioned external table, zero rows are returned.
By setting enabling DEBUG on the hive shell command line, I can see that the job reports a single path to be processed, which in turn results in zero map tasks.
If I define a partitioned table that sees the entire record as a single string column, then it fetches records from the partitions without a problem.
I am currently trying to accomplish the same, but with little success so far. Did you find a solution?