Spark Job failed due to OOM while reading huge files

Question

This is my spark configuration

I am trying to process 3 xml files with the data 12GB+1GB+1GB

Sprak properties are

  spark:
    default: