Hi all, Since the data I want to process is not on HDFS, I try to use sc.makeRDD() to ensure all items of a partition is located on one node, then the task can be launched on that node.
Now comes the problem, sometimes, the task is already assigned to some executors, then other executors are launched. Thus, when deal with partition on node A, spark may launched a executor on node B, which is not my expect. How can I ensure only until all executors are launched, then assign task? If anyone have some idea, please let me know. Thanks so much! Best Regards, Qian --------------------------------------------------------------------- To unsubscribe, e-mail: user-unsubscr...@spark.apache.org For additional commands, e-mail: user-h...@spark.apache.org
