If you need to use your own version of a jar file in your map reduce job, you can do so through following steps:
1. Bundle the jar file into your map reduce jar or pass the jar file through distributed cache.
2. From your map reduce client, update mapred-site.xml file to put following section into it:
mapreduce.job.user.classpath.first=true
(of course, you need to call conf.addResource() to add mapred-site.xml on your client code)
3. Restart your client.
