| Author |
Message
|
| apa |
 Posted: Thu Nov 15, 2007 11:40 am Post subject: XMLTransformation incompatible with additionalInstances? |
 |
|
Newbie
Joined: 15 Nov 2007
Posts: 7
|
This is quite difficult to explain, so please bear with me...
I have the following flow set up:
MQInput -> Compute -> XMLTransformation -> MQOutput
I have 2 different stylesheets. The compute node contains code to determine what stylesheet to use and then sets the OutputLocalEnvironment.ComIbmXslXmltStylesheetname to the appropriate one. I did some testing and it runs fine.
Now I modify the "additionalInstances" parameter in the BAR file to have this flow run multiple times in parallell (for performance reasons). It seems that, sometimes, when one type of message comes in, the wrong stylesheet is used... I notice that this occurs when a message requiring the other stylesheet is processed in parallell.
I am wondering whether this might be due to the fact that there is just one thread processing XSL's per execution group. If so, the first message could set the ComIbmXslXmltStylesheetname and the other message would then modify the XSL used before the first message is transformed.
P.S.: I run WMB v6 (with FixPack5 installed). |
|
| Back to top |
|
 |
| jefflowrey |
| Posted: Thu Nov 15, 2007 11:53 am Post subject: |
|
|
Grand Poobah
Joined: 16 Oct 2002
Posts: 19981
|
Try setting XML Embedded Stylesheet Selection Priority to 0.
_________________
I am *not* the model of the modern major general. |
|
| Back to top |
|
|
| apa |
| Posted: Thu Nov 15, 2007 11:54 am Post subject: |
|
|
Newbie
Joined: 15 Nov 2007
Posts: 7
|
It is  |
|
| Back to top |
|
|
| jefflowrey |
| Posted: Thu Nov 15, 2007 12:00 pm Post subject: |
|
|
Grand Poobah
Joined: 16 Oct 2002
Posts: 19981
|
Are you using deployed style sheets, or non-deployed style sheets?
Are you caching the compiled style sheets - i.e., what's the Stylesheet Cache Level?
_________________
I am *not* the model of the modern major general. |
|
| Back to top |
|
|
| apa |
| Posted: Thu Nov 15, 2007 12:05 pm Post subject: |
|
|
Newbie
Joined: 15 Nov 2007
Posts: 7
|
| The stylesheets deployed are cached, cache-level is set to 5 on the XMLTransformation node (default setting). |
|
| Back to top |
|
|
| jefflowrey |
| Posted: Thu Nov 15, 2007 12:24 pm Post subject: |
|
|
Grand Poobah
Joined: 16 Oct 2002
Posts: 19981
|
You can try turning on user trace, and see if you see anything relevant.
But it sounds a lot like you need to open a PMR.
The other thing to try is making a copy of your flow, with a different name. Deploy that and the original to the same EG, each with 1 additional instance.
See if this repeats the same problem.
_________________
I am *not* the model of the modern major general. |
|
| Back to top |
|
|
| apa |
| Posted: Thu Nov 15, 2007 12:32 pm Post subject: |
|
|
Newbie
Joined: 15 Nov 2007
Posts: 7
|
That's more or less the way it used to work before and the problem never occurred then (but I had other issues then which forced me to rewrite parts of the flows).
I was afraid of that "PMR" reply. The flows are currently still in testing phase but production installation is approaching rapidly and the final large test is planned for next monday. That doesn't leave a lot of time for a PMR 
I guess I was hoping someone else ran across a similar issue here. Thanks your help so far!
Oh, and we are not allowed to use the user-trace where I work so that's not really an option. Tomorrow I'll try to re-produce the issue more clearly and perhaps we'll open a PMR then. |
|
| Back to top |
|
|
| jefflowrey |
| Posted: Thu Nov 15, 2007 12:50 pm Post subject: |
|
|
Grand Poobah
Joined: 16 Oct 2002
Posts: 19981
|
Not allowed to use user-trace?
Do you mean "not allowed to use Trace nodes"?
That's got some validity.
Or do you really mean "not allowed to run mqsichangetrace"?
That's a lot more unusual, in my opinion.
It'd be very hard to work a PMR, without it.
_________________
I am *not* the model of the modern major general. |
|
| Back to top |
|
|
| apa |
| Posted: Thu Nov 15, 2007 1:05 pm Post subject: |
|
|
Newbie
Joined: 15 Nov 2007
Posts: 7
|
I realize it sounds strange... In the company I work for, I am a developer and am not allowed to run user traces myself. IBM can request such a trace in the context of a PMR but it is a specialized team's responsibility to run that trace.
Anyway, there may be a couple of ways I can think of to verify what I believe is happening... |
|
| Back to top |
|
|
| jefflowrey |
| Posted: Thu Nov 15, 2007 1:11 pm Post subject: |
|
|
Grand Poobah
Joined: 16 Oct 2002
Posts: 19981
|
Well. Ask the team to run the trace.
I understand not allowing a developer to have that access on prod or UAT systems.
I typically advise admins to give that level of access to at least some developers on Integration/Dev systems - just to reduce workload on the admin teams.
_________________
I am *not* the model of the modern major general. |
|
| Back to top |
|
|
| apa |
| Posted: Thu Nov 15, 2007 2:01 pm Post subject: |
|
|
Newbie
Joined: 15 Nov 2007
Posts: 7
|
I will, thanks.
I just thought of something. The flow I described above is a simplified version of the actual flow. In reality, we do not set the ComIbmXslXmltStylesheetname field through a Compute node, but rather through a filter-node (SET LocalEnvironment.ComIbmXslXmltStylesheetname = ...). This way of working was suggested to me by the WMB specialist in the company I work for. Do you know whether that could cause this kind of problems? |
|
| Back to top |
|
|
| jefflowrey |
| Posted: Thu Nov 15, 2007 2:07 pm Post subject: |
|
|
Grand Poobah
Joined: 16 Oct 2002
Posts: 19981
|
LocalEnvironment is LocalEnvironment is LocalEnvironment.
It shouldn't matter how it's set. The only reason to use a Filter node is to avoid having to copy the message tree. This is a performance optimization technique.
_________________
I am *not* the model of the modern major general. |
|
| Back to top |
|
|
| XZ |
| Posted: Fri Nov 16, 2007 7:45 am Post subject: |
|
|
Apprentice
Joined: 22 May 2006
Posts: 45
Location: IBM Hursley
|
You are probably affected by a known bug and as far as I know a fix has been put in for V6. Is FixPack5 the latest FP?
_________________
Regards,
-XZ
WMQI Development |
|
| Back to top |
|
|
| apa |
| Posted: Sat Nov 17, 2007 6:01 am Post subject: |
|
|
Newbie
Joined: 15 Nov 2007
Posts: 7
|
| It's the latest I know of... IBM suggested we install it when we submitted our previous PMR and it is installed. |
|
| Back to top |
|
|
|
|
