In Hadoop 2, the components are split into modules and so we now have multiple jars. For writing mapreduce programs, it should be enough to simply depend on the hadoop-client module/jar. The rest of the transitive dependencies should be pulled in automatically assuming you are using maven or ant/ivy.
about Yarn‘s programming
when i use Java language to program MapReduce,i should import hadoop’s Jar to use hadoop’API,
but in hadoop 2.2.0 I can not find hadoop jar,so i can not use java to run it
Support from the Experts
A HDP Support Subscription connects you experts with deep experience running Apache Hadoop in production, at-scale on the most demanding workloads.
Become HDP Certified
Real world training designed by the core architects of Hadoop. Scenario-based training courses are available in-classroom or online from anywhere in the world