What's New in Apache Hadoop YARN

Learn about the new features of Hadoop YARN in Cloudera Runtime 7.2.10.

Migrating database configuration to a new location

The operations performed on queues in Queue Manager UI are stored as Queue Manager versions. You can either store these versions in the default database location on the host or configure a new location using Cloudera Manager UI. For security reasons, if you do not want to allow users to access the default database, you can move the database file to an alternative location. During an upgrade, you can move the database file to some other location and then restore this file to the default location after the upgrade.

For more information, see Migrating database configuration to a new location.

Configuring Node Attribute for Application Master Placement (Technical Preview)

You can use the Node Attribute property to describe the attributes of a Node. The placement preference assigns nodes as worker nodes or compute nodes using the Node Attribute property. Application Master (AM) container is placed to run on worker nodes instead of compute nodes. The worker group is more stable because YARN ResourceManager and HDFS NameNode run in it. Also, the worker group nodes are less likely to be shut down due to autoscaling.

For more information, see Configuring Node Attribute for Application Master Placement.

YARN Ranger authorization support

Before this feature a single cm_yarn service had to be shared across multiple Data Hub clusters which was not ideal in a multi-tenant setup. That is because a single admin could update and change queue permission in all clusters.

From 7.2.10 if Cloudera Manager 7.4.2 or higher is used, each Data Hub cluster’s YARN cluster can have a dedicated Ranger YARN repository. That enables admins to set different YARN policies for different DataHub clusters.

This feature is enabled by default in a new Data Hub cluster installation. Cloudera Manager automatically creates a Ranger YARN repository for each cluster. If you are migrating from a lower CDP Public Cloud version to 7.2.10, this feature is disabled. In such cases, the cm_yarn repository is used until the cluster is deleted and a new DataHub cluster with Cloudera Manager 7.4.2 or higher is created.

For more information, see YARN Ranger authorization support.