Hi
Why dont you check if you can just process the large file standalone and
then do the outer loop next.
sqlContext.read.json(jsonFile) .select($"some", $"fields") .withColumn(
"new_col", some_transformations($"col")) .rdd.map( x: Row => (k, v) )
.combineByKey()
Deenar
On 14 October 2015 at
Yes it went wrong when processing a large file only. I removed
transformations on DF, and it worked just fine. But doing a simple filter
operation on the DF became the last straw that breaks the camel’s back.
That’s confusing.
On Wed, Oct 14, 2015 at 2:11 PM Deenar Toraskar
Hey Spark Users,
I kept getting java.lang.OutOfMemoryError: Java heap space as I read a
massive amount of json files, iteratively via read.json(). Even the result
RDD is rather small, I still get the OOM Error. The brief structure of my
program reads as following, in psuedo-code: