| Author |
Message
|
| bdrummond |
 Posted: Tue Dec 04, 2012 3:47 am Post subject: Compute Node not passing message |
 |
|
Disciple
Joined: 06 May 2004
Posts: 164
|
We have a strange issue and are using v 8.0.0.1.
Subflow is receiving a message and the first compute node completes successfully and passes the message to it's 'out' terminal. Then nothing happens. ie, the second compute node never receives the message even though the appropriate terminals are connected.
We have placed a trace node after the 1st compute node but nothing is written to the trace, so it looks like the message never gets there.
If we propagate to a Label node and not a terminal, the Label node accepts the message, passes it to it's own 'out' terminal and then nothing happens again.
It appears that the SubFlow is not recognising the connections between nodes.
Very strange.
If anyone has any ideas on what the issue could be, it would be greatly appreciated. |
|
| Back to top |
|
 |
| kash3338 |
| Posted: Tue Dec 04, 2012 4:02 am Post subject: |
|
|
Shaman
Joined: 08 Feb 2009
Posts: 709
Location: Chennai, India
|
| Where are you setting your output message? Is it in Outputroot or some Environment? What is your code in the compute node? Did you take a User trace? |
|
| Back to top |
|
|
| bdrummond |
| Posted: Tue Dec 04, 2012 4:09 am Post subject: |
|
|
Disciple
Joined: 06 May 2004
Posts: 164
|
We are setting both OutputRoot and Environment and the 1st Compute node is successful and then passes the message to the 'out' terminal. The 2nd node in the SubFlow never receives the message.
We have done a User Trace and processing in this SubFlow stops as soon as the 1st compute node is completed. |
|
| Back to top |
|
|
| McueMart |
| Posted: Tue Dec 04, 2012 4:17 am Post subject: |
|
|
Chevalier
Joined: 29 Nov 2011
Posts: 490
Location: UK...somewhere
|
| Do you have any messages in the broker error log (syslog / windows event log)? |
|
| Back to top |
|
|
| bdrummond |
| Posted: Tue Dec 04, 2012 4:22 am Post subject: |
|
|
Disciple
Joined: 06 May 2004
Posts: 164
|
| No, nothing in there. No errors reported, just the second node in the subflow never receives the message. |
|
| Back to top |
|
|
| mqjeff |
| Posted: Tue Dec 04, 2012 4:35 am Post subject: |
|
|
Grand Master
Joined: 25 Jun 2008
Posts: 17447
|
Exactly what ESQL statement is used to pass control out of the compute node?
RETURN? PROPAGATE? |
|
| Back to top |
|
|
| bdrummond |
| Posted: Tue Dec 04, 2012 4:40 am Post subject: |
|
|
Disciple
Joined: 06 May 2004
Posts: 164
|
Both don't seem to work when passing control via a terminal.
If we PROPAGATE to a label node, the label node receives the message but the node after the label does not receive it.
If we PROPAGATE to a specific terminal or RETURN from the 1st compute node, no further nodes (of any type) receive the message. |
|
| Back to top |
|
|
| mqjeff |
| Posted: Tue Dec 04, 2012 4:48 am Post subject: |
|
|
Grand Master
Joined: 25 Jun 2008
Posts: 17447
|
So if you put in
you do not observe control being passed to the out terminal, or into the next node?
And you do not see any indication in the error logs of the EG restarting?
And you do not see any abend files created?
Does this occur if you deploy the same code to a brand new ExecutionGroup?
If it does, I would suggest deploying it to a brand new execution group and taking a service trace. |
|
| Back to top |
|
|
| bdrummond |
| Posted: Tue Dec 04, 2012 5:32 am Post subject: |
|
|
Disciple
Joined: 06 May 2004
Posts: 164
|
We can see the Compute node passing to the 'out' terminal after 'RETURN TRUE'. But that is where the SubFlow stops.
The message never arrives on hte next node.
The main flow has a FLlwOrder node (SubFlow above is First) and the Trace file shows that the message is then passed back to the Second terminal of the FlowOrder node and the rest of the flow is successful.
The EG does not restart and there are no abend files created.
We have not tried deploying to a new EG. |
|
| Back to top |
|
|
| mqjeff |
| Posted: Tue Dec 04, 2012 5:34 am Post subject: |
|
|
Grand Master
Joined: 25 Jun 2008
Posts: 17447
|
It almost sounds like you have an old version of code deployed.
Again, it's worth deploying to a new EG. That will verify that the deployment package is good or bad, and give you a clean environment free from interference or noise from other things in which to run traces for when you have to open a PMR. |
|
| Back to top |
|
|
| bdrummond |
| Posted: Tue Dec 04, 2012 5:48 am Post subject: |
|
|
Disciple
Joined: 06 May 2004
Posts: 164
|
The code is an old version as we are currently migrating to V8. However, the code works fine in v7 and it only appears to be this particular SubFlow that is causing an issue. Other SubFlows are fine
We have tried adding a new ComputeNode with simply 'CopyEntireMessage' in it and we see the same strange behaviour.
We'll try deploying to a new clean EG and see what happens there. A PMR will also be raised.
Cheers. |
|
| Back to top |
|
|
| mqjeff |
| Posted: Tue Dec 04, 2012 5:55 am Post subject: |
|
|
Grand Master
Joined: 25 Jun 2008
Posts: 17447
|
| What I meant was that it was possible that the deployed code did not match the code in the Toolkit. |
|
| Back to top |
|
|
| nathanw |
| Posted: Tue Dec 04, 2012 6:11 am Post subject: |
|
|
Knight
Joined: 14 Jul 2004
Posts: 550
|
Just out of curiosity does the subflow also have aMQ input node?
We have been facing issues that we cannot compile / deploy in V8001 with subflows like this because of the way sub flows are handled.
_________________
Who is General Failure and why is he reading my hard drive?
Artificial Intelligence stands no chance against Natural Stupidity.
Only the User Trace Speaks The Truth  |
|
| Back to top |
|
|
| bdrummond |
| Posted: Tue Dec 04, 2012 6:18 am Post subject: |
|
|
Disciple
Joined: 06 May 2004
Posts: 164
|
@mqjeff - We have removed everything from the EG, bounced the broker, redeployed the BAR file and still face the same issue so we are confident that the code we are looking at in the Toolkit is also the code that is deployed.
@nathanw - There is no MQInput node in this subflow, just the Input node.
There are no issues with compiling and/or deploying.
Thanks for your suggestions.  |
|
| Back to top |
|
|
| mqjeff |
| Posted: Tue Dec 04, 2012 7:01 am Post subject: |
|
|
Grand Master
Joined: 25 Jun 2008
Posts: 17447
|
| nathanw wrote: |
Just out of curiosity does the subflow also have aMQ input node?
We have been facing issues that we cannot compile / deploy in V8001 with subflows like this because of the way sub flows are handled. |
there's a difference between a '.subflow' subflow and a '.msgflow' subflow in v8001.
You should accept the fact that you have to convert all of your '.msgflow' subflows to '.subflow' subflows as of 8.0.0.1. It may be a pain, but it's better to feel it now than struggle with it further. |
|
| Back to top |
|
|
|
|
