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

Posted to dev@geronimo.apache.org by "David Jencks (JIRA)" <ji...@apache.org> on 2007/01/03 00:13:27 UTC

[jira] Created: (GERONIMO-2688) no-pool and xa-transaction with no caching result in connections being closed before the tx ends.

no-pool and xa-transaction with no caching result in connections being closed before the tx ends.
-------------------------------------------------------------------------------------------------

                 Key: GERONIMO-2688
                 URL: http://issues.apache.org/jira/browse/GERONIMO-2688
             Project: Geronimo
          Issue Type: Bug
      Security Level: public (Regular issues)
          Components: connector
    Affects Versions: 1.1.1, 1.2, 2.0
            Reporter: David Jencks
         Assigned To: David Jencks
             Fix For: 2.0


A user has discovered that a connection manager for an outbound resource adapter configured for no-pool and xa-transaction (with no caching) results in the physical connection being closed before the transaction can be completed.  This might possibly be a problem with pooling enabled if there's a long delay between the connection handle being closed and the transaction ending, during which time the idle remover might fire up and remove the connection.

This problem can be worked around by enabling transaction-caching which keeps the connection associated with the transaction until the transaction ends (preventing it from being closed) or by enabling thread-caching (which ties the connection to the thread more or less indefinitely, and has not really been tested ever).

I'll think about whether it is practical to fix this directly. 

-- 
This message is automatically generated by JIRA.
-
If you think it was sent incorrectly contact one of the administrators: http://issues.apache.org/jira/secure/Administrators.jspa
-
For more information on JIRA, see: http://www.atlassian.com/software/jira

[jira] Closed: (GERONIMO-2688) no-pool and xa-transaction with no caching result in connections being closed before the tx ends.

Posted by "David Jencks (JIRA)" <ji...@apache.org>.

     [ https://issues.apache.org/jira/browse/GERONIMO-2688?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

David Jencks closed GERONIMO-2688.
----------------------------------

    Resolution: Won't Fix

after thinking about this I think the current behavior is correct.  It's possible for a connector to supply an XAResource that still works even after the ManagedConnection it came from has been destroyed (for instance it could delay the actual connection shutdown till the tx is complete, it could open a new connection for the tx complete, or it could be multiplexing all work over a single connection).  So to allow for this possibility we should keep the current behavior.

With less strange connectors, you should configure transaction caching the purpose of which is to hold on to the ManagedConnection until the tx completes, and then "close" it.  Depending on the pool configuration that will put it back in the pool or destroy it.

> no-pool and xa-transaction with no caching result in connections being closed before the tx ends.
> -------------------------------------------------------------------------------------------------
>
>                 Key: GERONIMO-2688
>                 URL: https://issues.apache.org/jira/browse/GERONIMO-2688
>             Project: Geronimo
>          Issue Type: Bug
>      Security Level: public(Regular issues) 
>          Components: connector
>    Affects Versions: 1.1.1, 1.2, 2.0-M5
>            Reporter: David Jencks
>            Assignee: David Jencks
>             Fix For: 2.0-M5
>
>
> A user has discovered that a connection manager for an outbound resource adapter configured for no-pool and xa-transaction (with no caching) results in the physical connection being closed before the transaction can be completed.  This might possibly be a problem with pooling enabled if there's a long delay between the connection handle being closed and the transaction ending, during which time the idle remover might fire up and remove the connection.
> This problem can be worked around by enabling transaction-caching which keeps the connection associated with the transaction until the transaction ends (preventing it from being closed) or by enabling thread-caching (which ties the connection to the thread more or less indefinitely, and has not really been tested ever).
> I'll think about whether it is practical to fix this directly. 

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.