I am launching 7 sqoop jobs in parallel which launches 14 containers. Even after the jobs are Finished and Final Status is Succeeded in RM, I see that there are still 7 more containers running holding up the resources. They stop only after 10 minutes. Is there a configuration which is affecting this? I see the part_success files getting generated before itself. I am not able to figure out what extra work it is doing in these 10 minutes.