This is the documentation for CDH 4.7.0.
Documentation for other versions is available at Cloudera Documentation.

What's New in CDH4.6.0

Apache HBase

For CDH4.6.0, HBase has been rebased to 0.94.15, differing from Apache HBase 0.94.15 in the following ways:
  • Reverted HBASE-8521 (cell cannot be overwritten with bulk loaded HFiles) due to the semantics change it introduced with prior-to-CDH4.6.0 HBase.
  • Reverted HBASE-9097 (set HBASE_CLASSPATH before rest of classpath) due to incompatibilities with prior-to-CDH4.6.0 HBase.
  • Reverted HBASE-8352 (change .snapshot to .hbase-snapshot dir) due to incompatibilities with prior-to-CDH4.6.0 HBase.
New Feature:
  • HBASE-9047 - Added ReplicationSyncUp tool to finish replication when cluster is offline.

Apache HDFS

Upstream fixes:
  • HADOOP-10326 - MapReduce jobs cannot access S3 if Kerberos is enabled
  • HDFS-5031 - BlockScanner scans the block multiple times and on restart scans

Apache Flume

New Features:
  • FLUME-2155 - File Channel is now indexed during replay, improving replay speed
  • FLUME-2130 - Syslog UDP source can now handle larger messages
  • FLUME-2217 - Syslog headers can now be optionally added to the message body

Apache Oozie

New Feature:
  • Secure HBase Table Copy between two HBase servers from Oozie now works.

Apache MapReduce v1 (MRv1)

New Features:
  • Fair Scheduler placement policies allow placing jobs into pools based on the secondary group.
  • Combiners allow custom grouping comparators.

Apache MapReduce v2 (YARN)

New Feature:
  • Combiners allow custom grouping comparators.