Categories

Versions

You are viewing the RapidMiner Radoop documentation for version 7.6 - Check here for latest version

What’s New in RapidMiner Radoop 2.4?

This page describes the enhancement being delivered by RapidMiner Radoop 2.4.

Improved connection setup experience, new tests and error messages

Connection tests to the Hadoop cluster have improved and now provide more detail on potential issues. Tests include checking for networking problems, missing High Availability configurations, and more. The connection dialog now provides more details for determining the root cause of an integration problem.

The new release supports connection to the latest Hortonworks distribution (version 2.3) and continues to support the latest Cloudera and Amazon releases. Additionally, it automates a large part of the connection setup so that no manual steps are required to connect to a cluster with default settings and default networking. Connection to VMs that some of the vendors provide is also now possible.

RapidMiner Radoop 2.4 additionally provides several fixes and workarounds for Hadoop bugs in different distribution versions. The cleanup of various temporary objects is now more reliable.

Enhancements and bug fixes

The following improvements are part of RapidMiner Radoop 2.4.

Enhancements

  • Added new networking, security, and other connection tests

  • Improved connection testing framework

  • Automated connection setup for HDP advanced properties

  • Dropped support for CDH3 and Apache 0.20.x distributions

  • Updated Spark assembly default locations for latest distributions

  • Improved Spark quick fixes

  • Repackaged Hive UDF jars to avoid Hive classloading issues

  • Added normalization model output to Normalize operator

  • Added option to override sample size in Radoop Nest

  • Added option to clean temporary data from Hive Management perspective

Bug fixes

  • BUGFIX: Range normalization formula corrected if minimum is not zero

  • BUGFIX: Fixed potential failure of Pig Script operator on HDP

  • BUGFIX: K-Means no longer fails on special numerical attributes

  • BUGFIX: Decision Tree now handles special characters in nominal attributes