It appears that cloudera-scm still cannot access files in the java directories.  Check that the file is read/execute by other. This is mine:   -rwxr-xr-x 1 root root 106366 Sep 2 2014 libjli.so   I recommend just doing the following as again it appears there are permission problems in your JDK   chmod -R 777 /usr/java/jdk1.8.0_72   I'm not sure why permissions are like they are, but normally java is accessible by "other" I believe.   Regards,   Ben ... View more

  Looking back at your permissions, it appears that you have the following as reported by "stat /usr/java/jdk1.8.0_72/bin/java":   Access: (0776/-rwxrwxrw-) Uid: ( 504/ hduser) Gid: ( 504/ hadoop)   Since the shell used to start Clouder Manager is owned by cloudera-scm, this would mean that unless you altered your init script to start as user "hduser" then "other" users don't have access to run java.   Try adding the execute permission for other on your java binary and try starting Cloudera Manager.   Regards,   Ben ... View more

It appears you may not be starting Cloudera Manager when root or sudo. This is required and is likely the reason you are not allowed permission to java.   Try "service cloudera-scm-server start" as root or with sudo. ... View more

Hello,   When you mention you are seeing the same problem, does the solution not work for you? If it does not, it is likely you are seeing a new problem and you may want to show what is happening via screen shot in a new thread so we can work to diagnose the issue.   When using parcels with Cloudera Manager, it is possible to have many different parcels distributed in the cluster.  Are you using parcels or packages for CDH?  Could you explain what steps you attempted to use to upgrade or what guide you followed?   -Ben ... View more

When installing, click the More Options page and add a new URL to specify the version of CDH you need.   If you want 5.3.8, add this URL:   https://archive.cloudera.com/cdh5/parcels/5.3.8/ ... View more

Oops... disregard my last update.  It was off topic!     ... View more

I took a brief look at your issue.  If Host Monitor is not running, then you can check the stderr.log and stdout.log for clues if it is running out of heap (in which case the default behavior is to shutdown the service).   In Cloudera Manager, go to Clusters --> Cloudera Management Service --> (under Status Summary) Host Monitor --> Processes   Click on the "stdout" link to open the stdout for the process.  See if there are OOMEs there.   Regardless of how much RAM you have on this host, if the Host Monitor's max heap is tuned too low, it can happen that it will throw an OOME exception. ... View more