You are viewing a plain text version of this content. The canonical link for it is here.

Posted to issues@drill.apache.org by "Chris Westin (JIRA)" <ji...@apache.org> on 2015/10/13 20:19:05 UTC

[jira] [Created] (DRILL-3928) OutOfMemoryException should not be derived from FragmentSetupException

Chris Westin created DRILL-3928:
-----------------------------------

             Summary: OutOfMemoryException should not be derived from FragmentSetupException
                 Key: DRILL-3928
                 URL: https://issues.apache.org/jira/browse/DRILL-3928
             Project: Apache Drill
          Issue Type: Bug
          Components: Execution - Flow
    Affects Versions: 1.2.0
            Reporter: Chris Westin


Discovered while working on DRILL-3927.

The client and server both use the same direct memory allocator code. But the allocator's OutOfMemoryException is derived from FragmentSetupException (which is derived from ForemanException).

Firstly, OOM situations don't only happen during setup.

Secondly, Fragment and Foreman classes shouldn't exist on the client side. (This is causing unnecessary dependencies on the jdbc-all jar on server-only code).

There's nothing special in those base classes that OutOfMemoryException depends on. This looks like it was just a cheap way to avoid extra catch clauses in Foreman and FragmentExecutor by catching the baser classes only.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)