[jira] [Commented] (DIRMINA-884) Apache Mina byte buffer shrink affecting badly

["Emmanuel Lecharny (Commented) (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/DIRMINA-884?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13196451#comment-13196451 ] 

Emmanuel Lecharny commented on DIRMINA-884:
-------------------------------------------

I've looked at MINA code, and the only possible way for the buffer to shrink is that the buffer are created and the setAutoShrink() is called. Otherwise, once the buffer's size is set, it should remain the same.

Can you check that ?
                
> Apache Mina byte buffer shrink affecting badly
> ----------------------------------------------
>
>                 Key: DIRMINA-884
>                 URL: https://issues.apache.org/jira/browse/DIRMINA-884
>             Project: MINA
>          Issue Type: Bug
>          Components: Core
>    Affects Versions: 2.0.2
>         Environment: Linux, Cent OS
>            Reporter: Kasun Dilunika
>            Priority: Critical
>
> I have developed server application to process charging messages which are coming in a
> binary format. In this application I have used mina core libraries to develop the server
> following the state machine pattern.
> In my scenario, when client initiated a connection for the first time, it is not closed.
> Client maintains the connection with server by sending some echo messages periodically.
> This echo message is 24 bytes in size.When a message hits server first time, it always
> allocates a byte buffer with 30000 bytes and read the content in to the byte buffer.
> Every 30 seconds echo message comes to the server. Mina framework watches this behavior
> and starts to reduce the buffer in to a half. [30000 -> 15000 -> 7500 -> 3750 ??-> 58].
> And it remains in the size 58 as the incoming message is 24 bytes long.
> But the problem is starting when the actual message is received, which is 704 bytes
> long. As the server buffer is allocated only to 58 bytes, it only reads first 58 bytes
> from the message. After that only framework realize that this buffer is not enough and
> starting to double the byte buffer. But the time is too late to react, because decoder
> state machine reject the message as it is not tallying with the size stated in its
> header.
> So, is there a way to stop this optimization to this message, because I can?t go with
> this logic when it is considered the logic I am having?
> If there is a configuration to stop this please let me know, otherwise tell me where
> should I modify the source code in order to stop this optimization.

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira