[
https://issues.apache.org/jira/browse/ODE-644?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12740763#action_12740763
]
William McCusker commented on ODE-644:
--------------------------------------
I have not been able to reproduce this issue using hibernate with the embedded
derby database, however the theory is the same as for jpa. Hibernate does not
guarantee when statements as flushed which is most likely why this is difficult
to reproduce. Explicitly calling Session.flush should ensure that old routes
are not reused when they are scheduled to be removed.
> INTERNAL ERROR: No ENTRY for RESPONSE CHANNEL [xy]
> --------------------------------------------------
>
> Key: ODE-644
> URL: https://issues.apache.org/jira/browse/ODE-644
> Project: ODE
> Issue Type: Bug
> Components: BPEL Runtime
> Affects Versions: 1.3.2, 2.0
> Reporter: William McCusker
> Attachments: responsechannel_hib_1x.diff,
> responsechannel_jpa_1x.diff, responsechannel_trunk_jpa.diff
>
>
> When a receive activity is repeated for the same partner link and operation
> (for example a pick activity inside of a while loop) the second message with
> often result inINTERNAL ERROR: No ENTRY for RESPONSE CHANNEL [xy] where xy
> ends up being the id of the old channel used for the first receive. In the
> JPA case this is because org.apache.ode.dao.jpa.CorrelatorDAOImpl
> removeLocalRoutes calls EntityManager remove but since it is using managed
> transactions these changes are not necessarily committed immediately.
> The mem dao should be unaffected by this, still need to test hibernate but it
> appears safe.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
