I have a spark job(scala) which writes timeseries data onto hadoop over which there is a external table in hive. Table is partitioned on mulitple columns and one of the colu