Did you by any chances left a sparkSession.setMaster("local") lurking in
your code?
Last time i checked, to run on yarn you have to package a 'fat jar'. could
you make sure the spark depedencies in your jar matches the version you are
running on Yarn?
alternatively please share code including
I have a simple program that works fine in the local mode. But I am having
issues when I try to run the program in yarn-cluster mode. I know usually
no such method happens when compile and run version mismatch but I made sure
I took the same version.
205 [main] INFO