HADOOP / YARN - Are ResourceManager and hdfs NameNode always installed on the same host?

Are the "resource manager" and the "hdfs namenode" always installed on the same host?

1) When I want to send an HTTP request (YARN REST API) to get a new application identifier I use this web uri:

http://<rm http address:port>/ws/v1/cluster/apps/new-application 

This port is the http port of the resource manager web interface, its default value is 8088, as shown in img1: source for img1 Yarn ports

2) When I want to send an HTTP request (WebHDFS REST API) for hdfs commands, for example, to get the file status I use this web uri:

 http://<HOST>:<PORT>/webhdfs/v1/<PATH>?op=GETFILESTATUS 

This <PORT> is the http port of the NameNode HTTP node, its default value is 50070, as shown in img2: source for img2 HDFS Ports

Are the hosts of these components (ResourceManager and NaneNode) always installed on the same host?

Any help would be appreciated, thanks!

+8
rest apache hadoop yarn webhdfs
source share
1 answer

There is no need to run resourceManager and nameNode on the same machine, we have this configuration when working in the same node / small cluster --- with multiple nodes --- configurations. When there is a large cluster, the main nodes usually run resourceManager, nameNode and secondaryNamenode on different machines.

cluster topology

Take a look at these links:

Master Nodes in Hadoop Clusters

HortonWorks: a typical Hadoop cluster

+9
source share

All Articles