• copy past the application Id from the spark scheduler, for instance application142848729615225597
  • connect to the server that have launch the job
  • yarn application -kill application_1428487296152_25597