Hi all, I'm using hive-12. I have a file that contains 10 integer columns stored in ORC format. The ORC file is zlib compressed and indexing is enabled. I'm running a simple select count(*) with a predicate of the form (Col1 =0 OR col2 = 0 etc). The predicate touches all 10 columns but its selectivity is 0 (none of the records in the file satisfy the predicate). I have set the hive.optimize.ppd property to true. For this query basically none of the stripes should be read. However after the MR job finishes, I can see that the whole file has been read. Why is that? Is there any other parameter that I should enable for predicate pushdown/indexing to work?
Thanks, Avrilia