customer use queue "compute" to submit yarn jobs, yarn have limited computer resources, only one yarn job is allowed to run at one time, so, the second yarn job will not run, until the first one is removed from queue. Here the the question. suppose, a long-running yarn job is running on cluster, if user want to the second yarn job run , without killing the first yarn job. what is the best method? One possible solution is moving first application to another queue. then submit second yarn job, is there any other method?
... View more