You are viewing a plain text version of this content. The canonical link for it is here.
Posted to user@spark.apache.org by Praveen Sripati <pr...@gmail.com> on 2014/11/26 03:45:46 UTC

Spark on YARN - master role

Hi,

In the Spark on YARN, the AM (driver) will ask the RM for resources. Once
the resources are allocated by the RM, the AM will start the executors
through the NM. This is my understanding.

But, according to the Spark documentation (1), the
`spark.yarn.applicationMaster.waitTries` properties specifies `Set the
number of times the ApplicationMaster waits for the the Spark master and
then also the number of tries it waits for the SparkContext to be
initialized.`.

What is the significance of master in the Spark on YARN configuration?

(1) - http://spark.apache.org/docs/latest/running-on-yarn.html

Thanks,
Praveen