Spark with elasticsearch

What is the best way to integrate elasticsearch with spark?

truly expecting offline indexing -> building index by spark cluster, we do have a much bigger spark cluster but still suffering from online indexing.