apache spark with hadoop