Still problems with commit failures. help much appreciated.

View: New views
3 Messages — Rating Filter:   Alert me  

Still problems with commit failures. help much appreciated.

by James Richardson-3 :: Rate this Message:

Reply to Author | View Threaded | Show Only this Message


Hi,

I am still having considerable problems with the generic jms ra when
testing failure scenarios.

If anybody can help I would very much appreciate it.

I am using glassfish v1_ur1_p01_b02, and the RA v.17

I am connecting to Tibco EMS, and have a hibernate persistance manager.

What I would like is for Tibco to manage message redelivery.

When a transaction fails due to hibernate's inability to commit the
transaction (say due to a non-unique constraint failure) - the messages do
not seem to be placed back on the queue for redelivery. However the
actually message delivery will have completed OK ( the onMessage worked
fine )

I have the following activation spec:

                    <activation-config-property>
                    <activation-config-property-name>RedeliveryAttempts</activation-config-property-name>
                        <activation-config-property-value>0</activation-config-property-value>
                    </activation-config-property>
                    <activation-config-property>                      
<activation-config-property-name>SendBadMessagesToDMD</activation-config-property-name>
                        <activation-config-property-value>false</activation-config-property-value>
                    </activation-config-property>

What I would expect under these circumstances is that the entire
transaction will be rolled back (this is an XA transaction) as hibernate
will vote out of the commit.

However what exactly happens has been hard to determine. It looks like
messages are lost - but never redelivered.

Has anybody seen this behaviour? or does anybody have any ideas?

I will try to get some log files of this happening.

Thanks!!!

James


---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@...
For additional commands, e-mail: users-help@...


Re: Still problems with commit failures. help much appreciated.

by rampsarathy :: Rate this Message:

Reply to Author | View Threaded | Show Only this Message


Hi James,
        You want to do XA transaction where one resource is the TIBCO EMS
inbound JMS resource and the other resource is the outbound hibernate XA
resource Is this correct ?
Could you please post your genericra configuration (from domain.xml).

Redelivery comes into the picture only when an exception is thrown in
the onMessage method. In your case the XA transation should just be
rolled back when the  method ends. So, i dont think redelivery comes
into the picture here
Can you turn on jts, jta, connector and generic ra logs to finest and
send us the logs, we have to ensure if rollback is called on both the
resources.
       

Thanks
-Ramesh

James Richardson wrote On 10/29/07 12:37,:

> Hi,
>
> I am still having considerable problems with the generic jms ra when
> testing failure scenarios.
>
> If anybody can help I would very much appreciate it.
>
> I am using glassfish v1_ur1_p01_b02, and the RA v.17
>
> I am connecting to Tibco EMS, and have a hibernate persistance manager.
>
> What I would like is for Tibco to manage message redelivery.
>
> When a transaction fails due to hibernate's inability to commit the
> transaction (say due to a non-unique constraint failure) - the messages do
> not seem to be placed back on the queue for redelivery. However the
> actually message delivery will have completed OK ( the onMessage worked
> fine )
>
> I have the following activation spec:
>
>                     <activation-config-property>
>                     <activation-config-property-name>RedeliveryAttempts</activation-config-property-name>
>                         <activation-config-property-value>0</activation-config-property-value>
>                     </activation-config-property>
>                     <activation-config-property>                      
> <activation-config-property-name>SendBadMessagesToDMD</activation-config-property-name>
>                         <activation-config-property-value>false</activation-config-property-value>
>                     </activation-config-property>
>
> What I would expect under these circumstances is that the entire
> transaction will be rolled back (this is an XA transaction) as hibernate
> will vote out of the commit.
>
> However what exactly happens has been hard to determine. It looks like
> messages are lost - but never redelivered.
>
> Has anybody seen this behaviour? or does anybody have any ideas?
>
> I will try to get some log files of this happening.
>
> Thanks!!!
>
> James
>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: users-unsubscribe@...
> For additional commands, e-mail: users-help@...
>

---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@...
For additional commands, e-mail: users-help@...


Re: Still problems with commit failures. help much appreciated.

by James Richardson-3 :: Rate this Message:

Reply to Author | View Threaded | Show Only this Message

>
> Hi James,
> You want to do XA transaction where one resource is the TIBCO EMS
> inbound JMS resource and the other resource is the outbound hibernate XA
> resource Is this correct ?
> Could you please post your genericra configuration (from domain.xml).
>
> Redelivery comes into the picture only when an exception is thrown in
> the onMessage method. In your case the XA transation should just be
> rolled back when the  method ends. So, i dont think redelivery comes
> into the picture here
> Can you turn on jts, jta, connector and generic ra logs to finest and
> send us the logs, we have to ensure if rollback is called on both the
> resources.
>
>
> Thanks
> -Ramesh

Right - I have TIBCO as an inbound and outbound JMS resource and hibernate
as the other.

I just switched on all the log levels and ran my simple test. Rather
confusingly it looks it worked exactly as anticipated.

I think I need to be more specific in my tests differentiating between
failures in the commit phase and those causing 'application exceptions'.
Currently there is a belief in the team that 'it doesn't work' - but now i
need to tie that down a little.

Thanks for your help - especially clarifying when redelivery (as far as
the JMSRA is concerned) plays a part.

James




---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@...
For additional commands, e-mail: users-help@...

LightInTheBox - Buy quality products at wholesale price