I was trying to use pentaho to parse and generate a flat text file from a avro input file. Writing custom type converters solved the problem of mapping AvroWrapper type keys to Kettle's String type. Also I was able to plugin my custom Reducer to the Mapreduce job. But the inconvenience I got was that I had to use the old Mapred API to write my custom Reducer class, i.e. I had to implement the MapReduceBase interface instead of extending the Mapper class. When I tried to use the MapReduce API to write a custom output format I got a kettle exception like "com.achyut.mapreduce.test.MyCustomInputFormat" is not "org.apache.hadoop.mapred.OutputFormat". Can anyone tell me whether Pentaho supports the new MapReduce API? Or is there any workaround to make this work? I am using CDH4 hadoop distro and pentaho-4.4.0 stable community edition. Any response is appreciated. Thanks in advance