[
https://issues.apache.org/jira/browse/AXIS2-3641?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Mike Rheinheimer resolved AXIS2-3641.
-------------------------------------
Resolution: Fixed
Fixed in rev 639025. Thanks David Strite for the fix!
> memory leaks related to unmarshaller pooling
> --------------------------------------------
>
> Key: AXIS2-3641
> URL: https://issues.apache.org/jira/browse/AXIS2-3641
> Project: Axis 2.0 (Axis2)
> Issue Type: Bug
> Components: jaxws
> Reporter: Mike Rheinheimer
> Assignee: Mike Rheinheimer
> Fix For: 1.4
>
> Attachments: patch.txt
>
>
> David Strite provides the following comments and attached patch to fix:
> Enabling the pooling of the unmarshallers surfaced a few small memory leaks.
> 1. The first is caused by a race condition creating the innerMap and
> JAXBContext in JAXBUtils. Currently multiple equivalent JAXBContexts can be
> created. When the unmarshallers are pooled, the JAXBContext is the key,
> meaning the JAXBContext and Unmarshaller are kept around. Since only one of
> these will be in the jaxbMap, the rest are never used but never get cleaned
> up. Synchronizing around the puts solves this problem.
> 2. The second leak is caused by the AttachmentUnmarshallers on the JAXB
> Unmarshaller. The AttachmentUnmarshallers hold on to the MessageContext.
> These will not be cleaned up until the next request sets a new
> AttachmentUnmarshaller on that Unmarshaller. Setting the
> AttachmentUnmarshaller to null before putting the Unmarshaller in the pool
> resolves this.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
