spark-summit 中出现Initial job has not accepted any resources; check your cluster UI to ensure that wor

#

Initial job has not accepted any resources; check your cluster UI to ensure that workers are registered and have sufficient memory

15/03/23 22:15:07 ERROR SparkDeploySchedulerBackend: Application has been killed. Reason: All masters are unresponsive! Giving up.

实际上，我这个问题时akka的一个bug，即它不能解析域名。

• 问题描述：

  这是我的启动命令： ./spark-submit –master spark://10.4.21.220:7077 –class RemoteDebug –executor-memory 256m –driver-memory 500m /home/hadoop/spark/SparkRemoteDebug.jar

• 下图是我的master的情况

  
  问题来了，akka不能将10.4.21.220解析为master ！，这导致前面问题的报错，即ip错误情况。

• 解决方法：

  spark-submit的提交代码的master的地址（即–master spark://10.4.21.220:7077）应该跟master UI（见上图）上的地址一致，所以我把提交代码修改为 ./spark-submit –master spark://master:7077 –class RemoteDebug –executor-memory 256m –driver-memory 500m /home/hadoop/spark/SparkRemoteDebug.jar 就能解决问题！