Hey Piotr, The feature that Crunch needs from later versions of Hadoop is the ability to write multiple output files from a single mapper/reducer; it's baked pretty deeply into the design and would be tough to pull out w/o a bunch of work. If the version of Hadoop on the cluster supports multiple outputs and the mapreduce.* APIs, Crunch should run-- if it doesn't, I'd like to know more.
Josh On Tue, Dec 18, 2012 at 7:56 PM, Peter Knap <[email protected]> wrote: > Hi, > > I was wondering why Hadoop 0.20 is not supported by crunch. I like crunch > and would like to use it but we have a hadoop cluster on the 0.20 release > level and no immediate plans to upgrade it. Are you guys planning to > support older version of hadoop? > > Thanks, > Piotr > -- Director of Data Science Cloudera <http://www.cloudera.com> Twitter: @josh_wills <http://twitter.com/josh_wills>
