Github user MJFND commented on the issue:
https://github.com/apache/spark/pull/14658
Okay, but even if not then increasing the number of shuffle partition
should fix it, but its not.
On Dec 26, 2017 8:51 PM, "Guoqiang Li" wrote:
> Spark 2.2 has fix
Github user MJFND commented on the issue:
https://github.com/apache/spark/pull/14658
If "Remote Shuffle Blocks cannot be more than 2 GB" then setting up
spark.sql.shuffle.partitions=value, where value should be such that it has 2gb
per executor, like for 200GB of data, w