Humble request to validate my findings and confirming that it is a bug

On 06-Jan-2018 5:06 PM, "Bhuvan Gupta" <bhuva...@gmail.com> wrote:

Camel Users,

I have found the issue
First lets go over the code routines that are executed for a route like

from("file:stageDummy")
     .threads(1)
         .process(myProcess)

[1] In the above route there will be one polling thread (Let call it as "
*pollingThread*",) which poll the route directory and fetch all the files
in it. (link
<https://github.com/apache/camel/blob/20936b173f332090c6fc5903737152e7d26eb60b/camel-core/src/main/java/org/apache/camel/component/file/FileConsumer.java#L75>
)
[2] Then fo reach file *pollingThread *does a GenericFileConsumer.isValid
check , inside isValid it checks if the file is processed using LRUcache of
size 1000 (link)
<https://github.com/apache/camel/blob/bb39dc44d9603c0e6fce2460ddf35a7b94ca30d4/camel-core/src/main/java/org/apache/camel/component/file/GenericFileConsumer.java#L541>
[3] If file is Valid then *pollingThread *submit the task to
executorService having 1 thread(Lets Name is it CamelThread#1)
[4] Inside CamelThread#1 does following things
[1] Call the processor
[1] moving the file to .camel
[2] clear up the .camelLock file
[3] Remove the entry from LRU cache ----- IMPORTANT Point  (link
<https://github.com/apache/camel/blob/f6ac3d81420d3f53b029a0f0c31dd4610810e2c2/camel-core/src/main/java/org/apache/camel/component/file/GenericFileOnCompletion.java#L95>
)


Now if the two thread execution in the following timeline fashion THEN we
will end up with the same file been processed twice.

[1] *pollingThread*: poll and fetch all the files in it let say we get
[File1]
[2] *pollingThread*: send the file for isValid Check
[3] *pollingThread*: isValid Check passes for File1 as it is not present in
LRU cahce
[4] *pollingThread*: Submit the File1 exchange to CamelThread#1
[5] *pollingThread*: poll and fetch all the files in it let say we get
[File1, File1.camelLock]
[6] *CamelThread#1*: call processor for File1 and then remove File1 entry
from LRU cache
[7] *pollingThread*: send the file for isValid Check
[8] *pollingThread*: isValid check passes for the File1 as it is not is LRU
cache.
NOTE: isValid checked passed for File1 twice....HENCE THE EXCEPTION


Solution I see:
[1] Dont remove entry from cache after onCompletion (link
<https://github.com/apache/camel/blob/f6ac3d81420d3f53b029a0f0c31dd4610810e2c2/camel-core/src/main/java/org/apache/camel/component/file/GenericFileOnCompletion.java#L95>
)

*Humble request to validate my findings and confirming it as a bug :)*
Feel free to ask for more explanatory text if above text is not clean
enough.

Thanks
Bhuvan

On Wed, Jan 3, 2018 at 11:18 PM, <sim_yihts...@yahoo.com.invalid> wrote:

> I suspect using “File consumer + threads(…) + default readLock” is like
> having multiple File consumer watching the same folder - they’ll screw up
> each other.
>
>
> I offer this as a possible proof: if you add an `idempotent=true` to your
> URI, the problem should go away.
>
>
>
> p/s: I think you’re trying to process your files parallel-ly.  If that is
> your _actual_ question, I have no idea if the above is what you’re supposed
> to do even if it works.
>
>
>
>
>
>
> From: Claus Ibsen
> Sent: ‎Wednesday‎, ‎3‎ ‎January‎, ‎2018 ‎9‎:‎56‎ ‎PM
> To: users@camel.apache.org
>
>
>
>
>
> We cannot reproduce your problem.
>
> There can be several reasons why a file cannot be renamed, but
> unfortunately the file api just returns a boolean if it cannot rename
> a file.
> A typical problem is that your do some custom code where you read the
> file or do something with a input stream you do not close properly.
>
> Also using threads(1) does not really make sense.
>
>
>
> On Tue, Jan 2, 2018 at 6:02 AM, Bhuvan Gupta <bhuva...@gmail.com> wrote:
> > Hello nice people,
> >
> > We are using camel 2.17 with java 1.7 in our sample project and we
> defined
> > our route as:
> >
> > from("file:stageDummy")
> >     .threads(1)
> >     .process(myProcess);
> >
> >    We first add 1000 file in stageDummy folder, then we run the code and
> we
> > get exception as below
> >
> >
> > org.apache.camel.component.file.GenericFileOperationFailedException:
> Cannot
> > rename file: GenericFile[newFile319] to: GenericFile[.camel/newFile319]
> > at
> > org.apache.camel.component.file.strategy.GenericFileProcessS
> trategySupport.renameFile(GenericFileProcessStrategySupport.java:115)
> > at
> > org.apache.camel.component.file.strategy.GenericFileRenamePr
> ocessStrategy.commit(GenericFileRenameProcessStrategy.java:88)
> > at
> > org.apache.camel.component.file.GenericFileOnCompletion.proc
> essStrategyCommit(GenericFileOnCompletion.java:127)
> > at
> > org.apache.camel.component.file.GenericFileOnCompletion.onCo
> mpletion(GenericFileOnCompletion.java:83)
> > at
> > org.apache.camel.component.file.GenericFileOnCompletion.onCo
> mplete(GenericFileOnCompletion.java:57)
> > at
> > org.apache.camel.util.UnitOfWorkHelper.doneSynchronizations(
> UnitOfWorkHelper.java:104)
> > at org.apache.camel.impl.DefaultUnitOfWork.done(DefaultUnitOfWo
> rk.java:229)
> > at org.apache.camel.util.UnitOfWorkHelper.doneUow(UnitOfWorkHel
> per.java:65)
> > at
> > org.apache.camel.processor.CamelInternalProcessor$UnitOfWork
> ProcessorAdvice.after(CamelInternalProcessor.java:654)
> > at
> > org.apache.camel.processor.CamelInternalProcessor$UnitOfWork
> ProcessorAdvice.after(CamelInternalProcessor.java:609)
> > at
> > org.apache.camel.processor.CamelInternalProcessor$InternalCa
> llback.done(CamelInternalProcessor.java:239)
> > at
> > org.apache.camel.processor.CamelInternalProcessor$InternalCa
> llback.done(CamelInternalProcessor.java:250)
> > at
> > org.apache.camel.processor.RedeliveryErrorHandler$1.done(Red
> eliveryErrorHandler.java:480)
> > at
> > org.apache.camel.management.InstrumentationProcessor$1.done(
> InstrumentationProcessor.java:86)
> > at org.apache.camel.processor.Pipeline$1.done(Pipeline.java:148)
> > at
> > org.apache.camel.processor.ThreadsProcessor$ProcessCall.run(
> ThreadsProcessor.java:85)
> > at java.util.concurrent.Executors$RunnableAdapter.call(
> Executors.java:471)
> > at java.util.concurrent.FutureTask.run(FutureTask.java:262)
> > at
> > java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPool
> Executor.java:1145)
> > at
> > java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoo
> lExecutor.java:615)
> > at java.lang.Thread.run(Thread.java:745)
> >
> > As per our understanding camel picks up the file from the mentioned
> > directory i.e in my case-stageDummy and processed by process-myProcess
> > and moves it to .camel sub folder.We noticed that if we were having more
> > than or around 1000 files in the folder -stageDummy ,there were some
> > exceptions
> > thrown
> >
> > This exception is thrown randomly for n number of files,where n changes
> > everytime we execute our main.Note that we were not trying to access the
> > exchange
> > instead we kept our process simple.We could see that if there are 1000
> files
> > in our folder then the processor is getting called 1000+x times ,where x
> > changes everytime example it gets called 1024 times,and we could see that
> > camel is trying to process  these x number of files again
> >
> > We are attaching the Main class as well as the project zip
> >
> > AND HAPPY NEW YEAR.
> >
> > hoping to resolve it soon
> >
> >
> >
> >
>
>
>
> --
> Claus Ibsen
> -----------------
> http://davsclaus.com @davsclaus
> Camel in Action 2: https://www.manning.com/ibsen2
>

Reply via email to