February 2009 - jboss-dev-forums - Jboss List Archives

[Design of Messaging on JBoss (Messaging/JBoss)] - Re: rolled back messages delivered twice.

by timfox

If it's the real onmessage then the order you specified is impossible. It's JBM that calls the real onMessage so how can you call rollback on the same thread that jbm called onmessage after onmessage is complete? You don't have control of that thread.... View the original post : http://www.jboss.org/index.html?module=bb&op=viewtopic&p=4213631#4213631 Reply to the post : http://www.jboss.org/index.html?module=bb&op=posting&mode=reply&p=4213631

15 years, 2 months

1
0
0 / 0

[Design of Messaging on JBoss (Messaging/JBoss)] - Re: rolled back messages delivered twice.

by ataylor

that is the real onmessage see JBMMessageHandler:onMessage() View the original post : http://www.jboss.org/index.html?module=bb&op=viewtopic&p=4213628#4213628 Reply to the post : http://www.jboss.org/index.html?module=bb&op=posting&mode=reply&p=4213628

15 years, 2 months

1
0
0 / 0

[Design of Messaging on JBoss (Messaging/JBoss)] - Re: rolled back messages delivered twice.

by timfox

So.. I'm interested in the following events: 1) The real onmessage where the jca layer receives the message from JBM 2) The transaction enlistment, delistment and rollback 3) The point where the JCA layer forwards the message to the delegate onmessage (the one that represents the endpoint). And the order in which they occur View the original post : http://www.jboss.org/index.html?module=bb&op=viewtopic&p=4213625#4213625 Reply to the post : http://www.jboss.org/index.html?module=bb&op=posting&mode=reply&p=4213625

15 years, 2 months

1
0
0 / 0

[Design of Messaging on JBoss (Messaging/JBoss)] - Re: rolled back messages delivered twice.

by timfox

But that onMessage is not the onMessage that JBM itself calls right. This is another delegate onmessage. The tx needs to be enlisted before the "real" onMessage is called. View the original post : http://www.jboss.org/index.html?module=bb&op=viewtopic&p=4213623#4213623 Reply to the post : http://www.jboss.org/index.html?module=bb&op=posting&mode=reply&p=4213623

15 years, 2 months

1
0
0 / 0

[Design of Messaging on JBoss (Messaging/JBoss)] - Re: rolled back messages delivered twice.

by ataylor

basically: tm.begin(); trans = tm.getTransaction(); trans.enlistResource(res)//res is xasession.getresource onmessage() tm.rollback(); all in the same thread View the original post : http://www.jboss.org/index.html?module=bb&op=viewtopic&p=4213620#4213620 Reply to the post : http://www.jboss.org/index.html?module=bb&op=posting&mode=reply&p=4213620

15 years, 2 months

1
0
0 / 0

[Design of Messaging on JBoss (Messaging/JBoss)] - Re: rolled back messages delivered twice.

by timfox

Are you sure all the tx operations from JCA are being performed on the bean in a single threaded fashion? Can you outline them here? View the original post : http://www.jboss.org/index.html?module=bb&op=viewtopic&p=4213617#4213617 Reply to the post : http://www.jboss.org/index.html?module=bb&op=posting&mode=reply&p=4213617

15 years, 2 months

1
0
0 / 0

[Design of Messaging on JBoss (Messaging/JBoss)] - rolled back messages delivered twice.

by ataylor

The scenario is this: I have a pool of MDB's, the MDB basically just calls ctx.setRollbackOnly(). This just calls rollback after onMessage. The rollback flushes any pending acks and then calls rollback on the server. The message is then obviously added back to the queue and delivered again. Heres some debug i caught: 1 adding ref 1295696329288188257 for session 10 | 2 ClientSessionImpl.acknowledge 1295696329288188257 | 3 acking 1295696329288188257 for session 10 | 4 polled ref 1295696329288188257 for session 10 | 5 adding ref 1295696329288188257 for session 21 | 6 clearing ref 1295696329288188257 for 21 | 7 adding ref 1295696329288188257 for session 15 | 8 ClientSessionImpl.acknowledge 1295696329288188257 | 9 acking 1295696329288188257 for session 15 | 10 ClientSessionImpl.acknowledge 1295696329288188257 | 11 acking 1295696329288188257 for session 21 | | | You can see the message being added, the client acknowledging it and then the server acking then the consumer polling it from deliveringRefs during the cancelRefs method. The problem occurs between steps 5 and 6 i think. After the message has been redelivered from session 10 it is added to session 21, in the meantime however the consumer from session 21 is also rolling back a different message (I havent shown this debug for clarity) and during this the message is also canceled from session 21 even tho the client hasn't received it yet. The message is the added to session 15 and handled correctly by it but the original client for session 21 still receives it and tries to acknowledge it. You can see the last 4 lines showing the message being delivered twice. View the original post : http://www.jboss.org/index.html?module=bb&op=viewtopic&p=4213615#4213615 Reply to the post : http://www.jboss.org/index.html?module=bb&op=posting&mode=reply&p=4213615

15 years, 2 months

1
0
0 / 0

[Design of Management Features on JBoss] - JBAS-6285, platform mbeans managed components

by scott.stark＠jboss.org

I finally got the platform mbeans integrated into the profile service management view and merged into the 5_x branch. Numerous issues in the jboss-managed project showed up that had to be addressed. These are all part of the 2.1.0.CR2 release which is now being used by jbossas 5_x. The biggest potential change is that the SimpleMetaType now has a notion of primitive types. The following final objects have been added: | public static final SimpleMetaType BOOLEAN_PRIMITIVE; | public static final SimpleMetaType BYTE_PRIMITIVE; | public static final SimpleMetaType CHARACTER_PRIMITIVE; | public static final SimpleMetaType DOUBLE_PRIMITIVE; | public static final SimpleMetaType FLOAT_PRIMITIVE; | public static final SimpleMetaType INTEGER_PRIMITIVE; | public static final SimpleMetaType LONG_PRIMITIVE; | public static final SimpleMetaType SHORT_PRIMITIVE; | These are created by using the corresponding primitive class name, for example, "int" for the int.class. The SimpleValueSupport class has new type safe wrap(...) static methods that allow one to create SimpleValues of the given primitive SimpleMetaTypes. Although the types are different, the java.lang.* and primitive version of a given SimpleValue do compare to be the same, even though their types are not equal. The addition of primitive types was needed to be able to properly resolve the signature of ManagedOpertions whose underlying java method used primitives as is the case for most of the platform mbean operations. Let me know if there are any issues with this change in the embedded jopr project. View the original post : http://www.jboss.org/index.html?module=bb&op=viewtopic&p=4213595#4213595 Reply to the post : http://www.jboss.org/index.html?module=bb&op=posting&mode=reply&p=4213595

15 years, 2 months

1
0
0 / 0

[Design of POJO Server] - Re: Integrating platform mbeans into kernel bus?

by scott.stark＠jboss.org

Ok, I finally finished this and the signature naturally falls out of the extended SimpleMetaTypes for the primitives, so no changes to the RuntimeComponentDispatcher or implementations are needed. These changes are in the 5_x branch which is now using the 2.1.0.CR2 jboss-managed release. View the original post : http://www.jboss.org/index.html?module=bb&op=viewtopic&p=4213592#4213592 Reply to the post : http://www.jboss.org/index.html?module=bb&op=posting&mode=reply&p=4213592

15 years, 2 months

1
0
0 / 0

[Design of JCA on JBoss] - Re: TCK testcases for JCA spec 10 chapter - WorkManager

by jeff.zhang

Do you think we can use mock object (like EasyMock) to test these cases? View the original post : http://www.jboss.org/index.html?module=bb&op=viewtopic&p=4213584#4213584 Reply to the post : http://www.jboss.org/index.html?module=bb&op=posting&mode=reply&p=4213584

15 years, 2 months

1
0
0 / 0

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

2006

jboss-dev-forums February 2009