This is the documentation for Cloudera Manager 4.8.3.
Documentation for other versions is available at Cloudera Documentation.

What's New in Cloudera Manager 4.1.0

Cloudera Manager 4.1.0 includes the following new features:

  • Support for CDH 4.1 new features. CDH4.1 has introduced a number of new features. Cloudera Manager supports these as follows:

  • High Availability using Quorum-based Storage. CDH 4.1 provides Quorum-based Storage (also know as the Quorum Journal Manager or QJM) as an alternative to NFS-mounted storage as the location for keeping the shared edits in a High Availability configuration. You can now select this as the shared edits mechanism when you configure HDFS High Availability.

  • JobTracker Recoverability. Provides a configurable option to allow job recovery upon JobTracker restart.

  • Support for the Cloudera Impala beta release. Install, manage, and monitor Impala using Cloudera Manager. See Installing Impala with Cloudera Manager

  • Support for configuring and managing Flume NG. With CDH4.1 or CDH3u5 refresh 2, Flume NG can now be added as a service, its agents configured through Cloudera Manager, and its health and performance monitored with Cloudera Manager.

  • Monitoring enhancements:

  • Cloudera Impala monitoring (separately-licensed)

  • ZooKeeper monitoring

  • Flume NG monitoring

  • Directory Disk-Space Monitoring: Monitoring of available space in log and data directories used by CDH components and Cloudera Manager.

  • Improvements in monitoring latency.

  • Host-level Decommissioning. You can now decommission all roles on a host in a single step. HDFS DataNodes, MapReduce TaskTrackers, YARN NodeManagers, and HBase RegionServers can be decommissioned; other roles running on the host are stopped. You can also recommission a host; however, roles are not restarted automatically.

  • Maintenance Mode. You can set monitored components (hosts, services, individual roles, or even an entire cluster) into maintenance mode, which will suppress all alerting for those components as long as the elements remain in that mode. Note that this does not suppress the logging of events, just the alerts that would otherwise be generated by those events.

  • Customizable service names. You can now provide your own display names for your services. These will then appear in the Cloudera Manager Admin Console UI. Note, however, that the original (internal) service name may appear or be required in certain circumstances, such as in log messages or in the API.

  • Usability enhancements:

  • Configuration properties that take values in units such as bytes or seconds now allow you to specify the units (e.g. Bytes, Seconds, KiBytes, MiBytes, MSecs, GiBytes and so on).

  • More readable installation logs: in case of failure, package installation logs now highlight the step of failure and other errors or warnings.

  • Various performance improvements in the UI.

  • The automatic restart of roles is now configurable on a role-by-role basis. Previously, every Cloudera Manager-managed process would automatically restart on unexpected failures. Now, when this property is set "off", a role that fails unexpectedly will be left Stopped and Cloudera Manager will not transparently restart it.

  • Health checks are now actionable: the health check messages provide links to information that includes suggestions for actions you can take. Links are also provided to health check threshold configurations where applicable.

  • Health check results are now listed by priority, with Bad/Critical ("red") checks at the top. The list is now collapsible to save space.

  • Chart views, pagination, and Activity Monitor filter settings now persist across user sessions in most cases.