I think if your job is running and you want to deploy a new jar which is the new version for the other, spark will think the new jar is another job , they distinguish job by Job ID , so if you want to replace the jar ,you have to kil job every time;
------------------ ???????? ------------------ ??????: "Mina Aslani"<aslanim...@gmail.com>; ????????: 2018??11??29??(??????)????2:44 ??????: "user @spark"<user@spark.apache.org>; ????: Do we need to kill a spark job every time we change and deploy it? Hi, I have a question for you. Do we need to kill a spark job every time we change and deploy it to cluster? Or, is there a way for Spark to automatically pick up the recent jar version? Best regards, Mina