Yarn NodeManager and ResourceManager in the same node

Question

(By default) Is there a "node manager" in the same node with "resource manager" in Hadoop Yarn? If not, is it possible to run them on the same node?

Answer 1

Depends on whether you want to run other containers (for ApplicationMasters or tasks) on the RM node. If you want your RM node to be dedicated to only resource-management, scheduling, etc., then do not start NM on the RM node. If you think you'll have spare cpu/mem/disk on the RM node (especially for a small cluster, or a cluster with few jobs to schedule), then run a NM on the RM node.

Answer 2

The script sbin/start-yarn.sh does just that

# start resourceManager

"$bin"/yarn-daemon.sh --config $YARN_CONF_DIR start resourcemanager

# start nodeManager

"$bin"/yarn-daemons.sh --config $YARN_CONF_DIR start nodemanager

in the yarn-site.xml you can set the IP:PORT of the resourcemanager as well as the node manager

yarn.resourcemanager.address

yarn.nodemanager.address

The node manager is the one that locate the resource manager (see http://hadoop.apache.org/docs/current/hadoop-yarn/hadoop-yarn-site/YARN.html)