Sparks job hangs in multi-node yarn cluster

Question

I have setup multi-node yarn-cluster

Node-1 : Datanode and master Node-2 : Datanode

When i submit my job, job works fine when first container is launched on Node-1 but job remains in accepted state when first container is launched on node-2 .Why this happens

On yarn web ui, i can see both nodes registered

Answer 1

Most probably there lack off available required resources to run job on cluster. To verify it check overall resources (cores, RAM) and required/used resources for your single job. The second one would be able to change state running after first job finished.