How to set Spark job's log

One: If you want spark to log your job, you must set two values

spark.eventLog.enabled=true
spark.eventLog.dir=/tmp/spark-events

this will storage the job info to spark.eventLog.dir as json data.

Tow: If you want to see the job's info, you can

start-history-server.sh
http://localhost:18080/

Three: Principle

Spark have a template for every job, but the data is null by default,
if you set spark.eventLog.enabled=true, Spark will filling the data and store it
in spark.eventLog.dir, and historyserver will remember the dir.