Just ran into this myself. Turns out that your Split object gets created on the job master node, and needs to get serialized around to workerbee nodes. Thus, your split type needs to implement Writable. See .../src/mapred/org/apache/hadoop/mapreduce/lib/input/FileSplit.java for an example.
-- View this message in context: http://n3.nabble.com/org-apache-hadoop-io-serializer-SerializationFactory-getSerializer-SerializationFactory-java-73-tp641003p728499.html Sent from the Mahout User List mailing list archive at Nabble.com.
