In the documentation is I found something like this.
· Local mode: number of cores on the local machine
· Mesos fine grained mode: 8
· Others: total number of cores on all executor nodes or 2, whichever is larger
I am using 2 node cluster with 48 cores(24+24). As per above no of data sets should be 1000/48=20.83, can be around 20 or 21.
But it is dividing into 2 sets of each 500 size.
I have used the function sc.parallelize(data, 10). But 10 datasets of size 100. 8 datasets executing on one node and 2 datasets on another node.
How to check how many cores are running to complete task of 8 datasets?(Is there any commands or UI to check that)
So by default when we call parallelize it will be parallelized by the default number (which we can control with the property spark.default.parallelism) or if we just want a specific instance of parallelize to have a different number of partitions, we can instead call sc.parallelize(data, numpartitions). The default value of this is documented in http://spark.apache.org/docs/latest/configuration.html#spark-properties
On Thu, Nov 6, 2014 at 10:43 PM, Naveen Kumar Pokala <firstname.lastname@example.org> wrote:
JavaRDD<Integer> distData = sc.parallelize(data);
On what basis parallelize splits the data into multiple datasets. How to handle if we want these many datasets to be executed per executor?
For example, my data is of 1000 integers list and I am having 2 node yarn cluster. It is diving into 2 batches of 500 size.
Cell : 425-233-8271