[jira] [Commented] (FLUME-1981) Rpc client expiration can be done in a more thread-safe way

["Hari Shreedharan (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/FLUME-1981?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13626809#comment-13626809 ] 

Hari Shreedharan commented on FLUME-1981:
-----------------------------------------

Though this is not a data loss or stability issue, it causes an unnecessary exception to logged, which can be a red herring during debugging.
                
> Rpc client expiration can be done in a more thread-safe way
> -----------------------------------------------------------
>
>                 Key: FLUME-1981
>                 URL: https://issues.apache.org/jira/browse/FLUME-1981
>             Project: Flume
>          Issue Type: Bug
>            Reporter: Hari Shreedharan
>
> If the destroyConnection() call happens just before a client.append() or client.getBatchSize(), it is possible that an EventDeliveryException is thrown which is caused by an NPE (since destroyConnection() will set client to null). 

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira