My ultimate goal is to spill each reducer's orc output in its own folder, making several external tables stored as orc. I'm trying to reproduce this http://hadoopathome.logdown.com/posts/277986-using-multipleoutputs-with-orc-in-mapreduce and it doesn't seem to work. The output neither splits to folders nor produces the correct file.
I've tested MultipleOutputs with text output and OrcNewOutputFormat with context.write(), it works like a charm. Did anyone have any luck with that combination?
Hive is 0.14.0. Hadoop 2.6.0