Issue Details (XML | Word | Printable)

Key: NUCRDBMS-686
Type: Bug Bug
Status: Closed Closed
Resolution: Fixed
Priority: No Testcase No Testcase
Assignee: Unassigned
Reporter: Ramya Nimmagadda
Votes: 0
Watchers: 0
Operations

If you were logged in you would be able to see more operations.
DataNucleus Store RDBMS

org.datanucleus.store.rdbms.ConnectionFactoryImpl.ManagedConnectionImpl.close() leaks connections if commit on the connection throws exception

Created: 16/Aug/13 02:29 AM   Updated: 23/Oct/13 11:56 AM   Resolved: 08/Oct/13 04:38 PM
Component/s: Connection
Affects Version/s: 3.0.8
Fix Version/s: None

File Attachments: 1. Text File NUCRDBMS-686.2.patch (2 kB)
2. Text File NUCRDBMS-686.patch (2 kB)


Datastore: Microsoft SQL Server
Severity: Production


 Description  « Hide
The close() method does not handle all the exception cases well while working with the connection. It should ensure that close() on the underlying connection object is called to let the connection pool manager do the necessary cleanup


Sort Order: Ascending order - Click to sort in descending order
Andy Jefferson added a comment - 16/Aug/13 06:49 AM
Provided patch will not apply on current codebase. Please generate any patch against current (supported) codebase ... SVN trunk

Shuaishuai Nie added a comment - 17/Aug/13 12:41 AM
Thanks Andy. Attached the re-based patch.

Andy Jefferson added a comment - 17/Aug/13 09:17 AM
That doesn't apply either. That "patch" has a conn.isClosed around line 506, just like the previous "patch". There is no such isClosed near there, and in fact method "public void close()" starts at line 506. https://sourceforge.net/p/datanucleus/code/18082/tree/platform/store.rdbms/trunk/src/java/org/datanucleus/store/rdbms/ConnectionFactoryImpl.java

Shuaishuai Nie added a comment - 12/Sep/13 07:45 PM
Hi Andy, I see the fix is already checked in to the trunk code base. Can we close this JIRA as resolved?

Andy Jefferson added a comment - 08/Oct/13 04:38 PM
Since you never provided a test for this, nor a valid patch I couldn't say anything was "fixed". If you say this doesn't happen with latest code then I can mark as fixed.