//
val rdd = Global.sparkContext.loadFromMongoDB(...).withPipeline(...)

rdd.count()

it always stuck at the last task. it may take 30 minutes to finish this last task, or maybe hange foreaver. No exception or error is found.

If it just reads few records, for example, 2000 records, it could finish the last task quickly. If it reads above 100000 records, it will hange there.

I have set 

System.setProperty("spark.mongodb.keep_alive_ms", "1000000"), but it does not help.