MQSeries.net :: View topic - Aggregating Original Request message to a Reply Problem

ddg · Posted: Wed Aug 29, 2007 1:30 pm Post subject:

I am in WMB6 trying to do a Fan-out/Fan-In (Aggregation) flow setup. This is my first time creating
a aggregation flow setup so bear with me.

Error:
I always get this in User Trace for Fan-In:

2007-08-29 10:49:10.324821 5944 UserTrace BIP4412I: Corresponding reques
t record not found for the reply message.
An AggregateReply node has received a mes
sage at its 'in' terminal. No corresponding record of a request message being se
nt could be found. See subsequent messages to determine how this situation has b
een handled.
It is possible that extraneous messages a
re arriving at the AggregateReply node 'in' terminal. Check your flow to ensure
that the only messages arriving here are replies to request messages previously
sent out and passed through an AggregateRequest node. It is possible that this m
essage is a valid reply but part of an aggregation which previously timed out. I
t is possible that this is a reply to a message which has not yet been recorded
by an AggregateRequest node. This can happen if request messages are sent outsid
e of transactional control. Adjust your transaction settings to ensure that mess
ages are sent under transactional control.

2007-08-29 10:49:10.326069 5944 UserTrace BIP4413I: Storing unknown mess
age persistently for subsequent processing.
The message has been stored persistently
by the AggregateReply node. If it is subsequently discovered to be a valid reply
in an aggregation then it will be processed as such at that time. If a number o
f seconds passes as specified by the 'unknown timeout' attribute, then the messa
ge will be propagated to the 'unknown' terminal.
No user action required.

I have an XML message coming in to a

Fan-out flow:
- InputQ to AggControlNode,
- AggControlNode splits to Compute1 and Compute2,
- Each Compute* goes to MQOutputNodes (mqOUT1 and mqOUT2(same queue as fan-in InputQueue later below))
- and then to 2 respective AggRequestNodes

Compute 1:
Set OutputRoot = InputRoot;
SET OutputLocalEnvironment.ComIbmAggregateControlNode =
InputLocalEnvironment.ComIbmAggregateControlNode;
Set OutputRoot.BLOB.BLOB to <some XMLNS field's value>;
SET OutputRoot.XMLNS = NULL;
SET OutputRoot.MQMD.CorrelId = InputRoot.MQMD.MsgId;

Compute 2:
Set OutputRoot = InputRoot;
SET OutputLocalEnvironment.ComIbmAggregateControlNode =
InputLocalEnvironment.ComIbmAggregateControlNode;
SET OutputRoot.MQMD.CorrelId = InputRoot.MQMD.MsgId;

Now, mqOUT1 mentioned above is the same as the inputQ to another flow for processing.
This processing flow has:
inputQ(mqOUT1) ==> *OutputQ
(*OutputQ being the same as input queue to the Fan-In InputQueue later below)
The reason for this is I am trying to mimick/simulate another system getting a blob then doing something to it
and then sending it back again as blob.

Fan-In:
InputQueue ==> AggReplyNode
AggReplyNode's Unknown Terminal ==> Compute_1
AggReplyNode's Out Terminal ==> Compute_2
AggReplyNode's TimeOut Terminal ==> Compute_3

Compute_1, _2 and _3 format the message to send particular folders after aggregation to go to queues for logging.

In addition:
- I might have some Trace Nodes in different places just to keep track of ${Root} and/or ${LocalEnvironment}.
- AggCOntrolNode has Setting of 300 secs for now
- AggReplyNode, which has the same Aggregate Name as AggControlNode, has unknown messaeg Timeout of 5 for now.

Any help/suggestion is greatly appreciated. Sorry for the very long post. I wanted to give as much info as I can.