We have docs on how to implement a MapReduce, but we do not document any of the following:
- avoid using org.apache.hadoop.mapred (old APIs)
- what needs to be included and excluded in the pom.xml to use MapReduce
- what versions of what dependencies are needed
The pom.xml from the MapReduce examples also does not make it clear, it inherits from the parent pom, which few people will read. The pom.xml in the MapReduce guide has it, but it explains nowhere what needs to be in the pom, and why it is important to exclude some transitive dependencies.