| Author |
Message
|
| pcelari |
 Posted: Mon Jan 28, 2008 7:13 am Post subject: 2nd parsing of same MRM fails occasionally. |
 |
|
Chevalier
Joined: 31 Mar 2006
Posts: 411
Location: New York
|
I have three msg flows with aggregation. the input is MRM msg.
The MQInput node in the fan-out flow is connected to a compute node who does nothing more than storing the original MsgId into MQRFH2.
The msg processing flow starts with MQInput node, with the same MRM msg set, format, as the one in fan-out flow.
However, occasionally, say, one in about every 5 msg, fails at the second MQINput node, with MRM parserException:
Text = XML Writing Errors have occurred
...
Text = No valid body of the document could be found.
I deployed all three msgflow in the same EG. set the fan-out/in to 4 instances, while the msg processing flow to 8 instances hoping to take advantage of paralelle processing.
Has anyone experienced this inconsistent behavior? Or any insight into what I should look into?
thanks a lot. 
_________________
pcelari
-----------------------------------------
- a master of always being a newbie |
|
| Back to top |
|
 |
| jefflowrey |
| Posted: Mon Jan 28, 2008 7:42 am Post subject: |
|
|
Grand Poobah
Joined: 16 Oct 2002
Posts: 19981
|
I'd say that, occasionally, about 1 in 5 messages, you don't receive a valid RHF2 header, and so your code to add one mangles the body of the message.
_________________
I am *not* the model of the modern major general. |
|
| Back to top |
|
|
| pcelari |
| Posted: Mon Jan 28, 2008 7:57 am Post subject: |
|
|
Chevalier
Joined: 31 Mar 2006
Posts: 411
Location: New York
|
No, the messages are exactly the same test msg, even the MRM msg body shown in the debug window is correct.
sth must be causing the parser to fail internally occasionally. but why does it behavior inconsistently?
_________________
pcelari
-----------------------------------------
- a master of always being a newbie |
|
| Back to top |
|
|
| jefflowrey |
| Posted: Mon Jan 28, 2008 8:02 am Post subject: |
|
|
Grand Poobah
Joined: 16 Oct 2002
Posts: 19981
|
Do you have a phantom EG? with an older version of the code?
Try undeploying the flow completely and then redeploy.
_________________
I am *not* the model of the modern major general. |
|
| Back to top |
|
|
| fjb_saper |
| Posted: Mon Jan 28, 2008 1:29 pm Post subject: |
|
|
Grand High Poobah
Joined: 18 Nov 2003
Posts: 20756
Location: LI,NY
|
Consistency is one of the reasons we set our flows all to output XML(NS) before the aggregation.
So when a message hits the aggregation node it is in xml.
In the same spirit we serialized the original MQMD and pass it as an XML field. This gets parsed back to an MQMD and used in the MQReply node in the fan in...
Enjoy 
_________________
MQ & Broker admin |
|
| Back to top |
|
|
| kimbert |
| Posted: Tue Jan 29, 2008 2:13 am Post subject: |
|
|
Jedi Council
Joined: 29 Jul 2003
Posts: 5542
Location: Southampton
|
| Quote: |
| I deployed all three msgflow in the same EG. set the fan-out/in to 4 instances |
Does it happen with a single instance? |
|
| Back to top |
|
|
| pcelari |
| Posted: Wed Jan 30, 2008 11:09 am Post subject: |
|
|
Chevalier
Joined: 31 Mar 2006
Posts: 411
Location: New York
|
thanks for the advice.
So it is not advisable to keep MRM after the aggregate control node?
I'll try it.
No, I never tried the same on a single instance configuration, as that's the reason of using aggregation - for paralelle processing.
But I'll try both of your recommendations.
many thanks for the insights.
_________________
pcelari
-----------------------------------------
- a master of always being a newbie
Last edited by pcelari on Wed Jan 30, 2008 11:15 am; edited 1 time in total |
|
| Back to top |
|
|
| jefflowrey |
| Posted: Wed Jan 30, 2008 11:11 am Post subject: |
|
|
Grand Poobah
Joined: 16 Oct 2002
Posts: 19981
|
Did you check for phantom EGs?
_________________
I am *not* the model of the modern major general. |
|
| Back to top |
|
|
| pcelari |
| Posted: Wed Jan 30, 2008 11:18 am Post subject: |
|
|
Chevalier
Joined: 31 Mar 2006
Posts: 411
Location: New York
|
I did. I removed all EGs except default. removed the deployed children, then deployed only to the default EG, but with multiple instances of processing flows. But still get one out of every 3 to 6 msg put into the failure queue connected to the FAILURE terminal of the second MQInput node.
_________________
pcelari
-----------------------------------------
- a master of always being a newbie |
|
| Back to top |
|
|
| jefflowrey |
| Posted: Wed Jan 30, 2008 11:24 am Post subject: |
|
|
Grand Poobah
Joined: 16 Oct 2002
Posts: 19981
|
Are you generating new MsgIds for each test message?
it's possibly you're doing something wrong with how you store the ID in the RFH2, which is causing it to include bad characters, which is causing the message to fail to parse.
_________________
I am *not* the model of the modern major general. |
|
| Back to top |
|
|
|
|
