NullPointer exceptions

[Tom McNeer]

Hi,

I recently moved an application to a new dedicated 64-bit server running Win2008 and CF9 Standard. Since doing so, I have begun to see NullPointerExceptions in which the stack trace points to Transfer objects, specifically the ActionAfterTransferDiscardEvent method.

Here's a portion of a sample stack trace:

"Error","jrpp-13","02/09/10","11:33:17",,""
java.lang.NullPointerException
    at coldfusion.runtime.NeoPageContext.searchScopes(NeoPageContext.java:744)
    at coldfusion.runtime.NeoPageContext.findAttribute(NeoPageContext.java:638)
    at coldfusion.runtime.CfJspPage._get(CfJspPage.java:296)
    at coldfusion.runtime.CfJspPage._get(CfJspPage.java:283)
    at coldfusion.runtime.CfJspPage._get(CfJspPage.java:271)
    at cfdocuments2edocument40F80DBE6087F1DF6736330761605BCD9A2etransfer1116507617$funcACTIONAFTERDISCARDTRANSFEREVENT.runFunction(C:\webapps\mlhealthcare\model\data\documents...@F80DBE6087F1DF6736330761605BCD9A.transfer:32)
    at coldfusion.runtime.UDFMethod.invoke(UDFMethod.java:472)
    at coldfusion.filter.SilentFilter.invoke(SilentFilter.java:47)
    at coldfusion.runtime.UDFMethod$ReturnTypeFilter.invoke(UDFMethod.java:405)
    at coldfusion.runtime.UDFMethod$ArgumentCollectionFilter.invoke(UDFMethod.java:368)
    at coldfusion.filter.FunctionAccessFilter.invoke(FunctionAccessFilter.java:55)
    at coldfusion.runtime.UDFMethod.runFilterChain(UDFMethod.java:321)
    at coldfusion.runtime.UDFMethod.invoke(UDFMethod.java:220)
    at coldfusion.runtime.TemplateProxy.invoke(TemplateProxy.java:490)
    at coldfusion.runtime.TemplateProxy.invoke(TemplateProxy.java:336)
    at coldfusion.runtime.CfJspPage._invoke(CfJspPage.java:2360)
    at cfSoftReferenceAdapter2ecfc1501711166$funcACTIONAFTERDISCARDTRANSFEREVENT.runFunction(C:\webapps\frameworks\transfer\com\events\adapter\SoftReferenceAdapter.cfc:148)

I am running the most recent SVN version of Transfer, and the generated Transfer data objects were deleted and regenerated on the server.

Has anyone experienced anything like this? The same application has been running in other environments (dev, etc.) without any similar errors.

--
Thanks,

Tom

Tom McNeer
MediumCool
http://www.mediumcool.com
1735 Johnson Road NE
Atlanta, GA 30306
404.589.0560

[Tom McNeer]

As an update:

As I move through more and more logs, I'm finding similar NullPointerExceptions that have nothing to do with Transfer.

So whatever's going on is evidently not Transfer-specific, so we can probably let this thread go.

On the other hand, if anyone is reading this who has seen a sudden flurry of such exceptions when accessing a Coldspring/Transfer app via Flash Remoting from a Flex app, I'd appreciate any insight.

I'm even seeing some entries in the JRun log that appear to reference BlazeDS, though I'm not using it (unless there's some BlazeDS stuff buried in the CF9 Flash Remoting functionality).

[Dennis Clark]

On Tue, Feb 9, 2010 at 12:04 PM, Tom McNeer <tmc...@gmail.com> wrote:

I'm even seeing some entries in the JRun log that appear to reference BlazeDS, though I'm not using it (unless there's some BlazeDS stuff buried in the CF9 Flash Remoting functionality).

Your "unless" is correct. With ColdFusion 9, Adobe changed the built-in Flash Remoting server code from a completely closed implementation to one based on BlazeDS 3.2.

Cheers,

-- Dennis

[Tom McNeer]

Your "unless" is correct. With ColdFusion 9, Adobe changed the built-in Flash Remoting server code from a completely closed implementation to one based on BlazeDS 3.2.

Thanks, Dennis. That's what I was guessing happened.

Now if I were just seeing the same errors on my CF9 dev machine, I'd know the source.

[Mark Mandel]

Sounds like a bug in CF9.  Have you checked the bug tracker / submitted a bug to see if it can be resolved / hot fixed?

Mark

--

Before posting questions to the group please read:
http://groups.google.com/group/transfer-dev/web/how-to-ask-support-questions-on-transfer
 
You received this message because you are subscribed to the Google Groups "transfer-dev" group.
To post to this group, send email to transf...@googlegroups.com
To unsubscribe from this group, send email to transfer-dev...@googlegroups.com
For more options, visit this group at http://groups.google.com/group/transfer-dev?hl=en

--
E: mark....@gmail.com
T: http://www.twitter.com/neurotic
W: www.compoundtheory.com

Hands-on ColdFusion ORM Training @ cf.Objective() 2010
www.ColdFusionOrmTraining.com/

[Tom McNeer]

Hi Mark,

On Tue, Feb 9, 2010 at 2:20 PM, Mark Mandel <mark....@gmail.com> wrote:

Sounds like a bug in CF9. 

While there seem to be some known issues in CF9 regarding serialization of aliased CFCs, I've found (since the last post) that I was wrong in pointing to CF9. Going back to the previous production environment, which was CF8, I also see a trail of NullPointerExceptions during January. They were in the JRun logs but not the CF logs.

They seem to relate to three particular method calls, but so far, I haven't found anything in those methods that could cause the error.

In fact, I can replicate one of the errors now (although it occurs inconsistently). But here's the odd part: when I view the transaction in Charles, I see: two requests go out with the same method call.

I don't believe the event is being fired twice in Flex, but I haven't yet been able to check that thoroughly. It could be that I'm inadvertently dispatching twice.

However - the two requests go out relatively simultaneously (same second, anyway). When there is an error, one returns a NullPointerException; the other returns the correct array of objects expected. I have checked the params being sent in each request; they are identical.

Additionally, while duplicating the call with the same params again seems to result again in two requests, I received the correct response from both requests.

So I'm guessing (and I really mean _guessing_) that the failure is somewhere in the serialization process. I do know that I recently had a different issue where the wrong data appeared to be returned to Flex, and I was able to prove (to myself, anyway) that the same call with the same params to the same proxy CFC produced different results when called from a plain .cfm page than when called from the Flex app.

I'm checking into the bug tracker, but if anyone has seen anything similar, I'd sure like to know. Since I can't replicate the error consistently, it'll be hard to submit a test case for a bug.