How did you resolve the issue? I am facing a similar problem with datanode not starting   ... View more

   Kindly check log detail on Resource manager:   <clusterip>:8088    Open terminal and checl actual log problem:   yarn logs -applicationId <APP_ID>    Example: APP_ID = application_1535002188113_0001      In mine case it was showing permission issue to directory '/user/history' so i gave it   sudo -u hdfs hadoop fs -chmod 775 /user/history or sudo -u hdfs hadoop fs -chmod 777 /user/history   Then it works.   ... View more

@mbigelow thats wright! the app used an older API from MRv1. The app uses almost the same amount of data each day plus aprox 30GB so the expected time could be a little bit more each day, as the app runs daily, but in the last days past from 9 hours to more than a day to finish. The solution I've founded was to decrease the amount of memory instead of increase it. Now the app can allocate more containers so the app is runing pretty fast. Thanks for your help, just in case I'll let you know. Guido.   ... View more

@mbigelowTry to search for "How to install CM over an existing CDH Cluster" ... View more

Finally I can get my cluster up and running! As msbigelow said two of my three JNs were up and running but bad rdeclared in hdfs-site.xml dfs.namenode.shared.edits.dir property. After change it the namenode service starts! Now everything apperars to be in order. I hope my problem could help in this community. Thanks @saranvisa and @mbigelow!       ... View more