My mapPartition code as given below outputs one record for each input record.
So, the output object has equal number of records as input. I am loading the
output data into a listbuffer object. This object is turning out to be too
huge for memory leading to Out Of Memory exception.
To be more clear my logic of partition is as below:
*Iterator(Iter1) -> Processing -> ListBuffer(list1)
iter1.size() = list1.size()
list1 goes out of memory*
*I cannot change the partition size.* My parition is based on input key and
all the records corresponding to a key need to go into same partition. Is
there a workaround to this?
/ tempRDD = iterateRDD.mapPartitions(p => {
var outputList = ListBuffer[String]()
var minVal = 0L
while (p.hasNext) {
val tpl = p.next()
val key = tpl._1
val value = tpl._2
if(key != prevKey){
if(value < key){
minVal = value;
outputList.add(minVal.toString() + "\t" +key.toString())
}else{
minVal = key;
outputList.add(minVal.toString() + "\t" +value.toString())
}
}else{
outputList.add(minVal.toString() + "\t" +value.toString())
}
prevKey = key;
}
outputList.iterator
})/
--
View this message in context:
http://apache-spark-user-list.1001560.n3.nabble.com/Out-of-Memory-error-caused-by-output-object-in-mapPartitions-tp26229.html
Sent from the Apache Spark User List mailing list archive at Nabble.com.
---------------------------------------------------------------------
To unsubscribe, e-mail: user-unsubscr...@spark.apache.org
For additional commands, e-mail: user-h...@spark.apache.org
