How to dump JVM memory for Private Cloud to troubleshoot "Out Of Memory" exceptions

Document created by Adam Arrowsmith Employee on Nov 14, 2011
Version 1Show Document
  • View in full screen mode
The instructions below describe how to temporarily configure your Private Cloud to dump the contents of its Java JVM memory to file to assist the Dell Boomi engineering team in researching "out of memory"-related errors. This information is useful to our engineering team to assess whether there are any code-level issues with memory management in the Cloud run time engine.

Note: These steps should only be attempted after going through the general High Volume Troubleshooting tips.
  1.           In your Cloud’s ../<Atom root>/bin directory, make a backup of the current procrunner.ps1 file.
  2.           Overwrite that file with the one attached to the end of this article.           
    1.                     Note: This is a "hot" fix. You do NOT need to restart any of the Cloud nodes.
  3.           Execute the process in question to get the out of memory error. It is ok if other processes are running concurrently.
  4.           When the process errors with the memory exception, it will generate a very large file named *.hprof in the appropriate ../<Atom root>/accounts/<sub-account> directory.
  5.           Zip this file and contact Dell Boomi support for instructions on how to upload this file to our FTP site.
  6.           You can swap the procrunner.ps1 file back to the original at your convenience.