Spark query on Hue 4.3.0 with EMR 5.20

Whenever we run query in hue using set Hive.execution.engine =spark we see yarn manager shows only 1-3 % of cluster resources getting utilised and query remains in execution state for hours. However when we execute same query from hive-cli can can see yarn-manager GUI shows resource utilisation around 80 to 90 % and query gets completed in few seconds
Hue 4.3.0 with EMR 5.20