@suresh krish
Here issue with cache in partitions, Clear your partitions and MSCK to repair the table. dropping partitions: alter table dbschemaname.tableanme DROP partition (source <> ''); Run MSCK for table repair, MSCK REPAIR TABLE <tablename>; run count command. Basically when you run select script, your request directly goes to hdfs file to get the requested information, here request wouldn't run mr job. But when you run count(*) it will trigger mr job to pull requested info.
... View more
