Hi Claus,

Thanks for the reply. Can I still use the idempotent repository to prevent 
different nodes from processing the same file even though
the idempotent read lock is not supported? I don’t see any limitations on using 
an idempotent repository on the FTP endpoint.
On 23 Nov 2020, 09:44 +0100, Claus Ibsen <[email protected]>, wrote:
> The ftp component does not support idempotent read locks, so you
> cannot do what you do.
>
> On Sun, Nov 22, 2020 at 8:05 AM Robin Vanderhallen
> <[email protected]> wrote:
> >
> > Hi,
> >
> > I’m using file and FTP consumer endpoints (Camel 2.22.1) in a clustered 
> > setup, backed with Hazelcast and write-through persistence using the 
> > following endpoint options:
> > idempotent=true&idempotentRepository=#hazelcastIdempotentRepository&idempotentKey=#keyExpression&readLock=idempotent-changed
> >  for the file endpoint
> > idempotent=true&idempotentRepository=#hazelcastIdempotentRepository&idempotentKey=#keyExpression&readLock=changed
> >  for the FTP endpoint
> >
> > I discovered that there is a difference in the behaviour of both endpoints:
> >
> > file component: confirm method is invoked after processing the files
> > FTP component: confirm method is not invoked after processing the files
> >
> > When readLock=idempotent-changed is set on the file endpoint, 
> > FileIdempotentChangedRepositoryReadLockStrategy is being used and the 
> > releaseExclusiveReadLockOnCommit method is
> > invoked which triggers the confirm method on the idempotent repository.
> > According to the Camel docs this type of read lock is not supported by the 
> > FTP component. In my setup FtpChangedExclusiveReadLockStrategy is being 
> > used, which is never invoking the confirm method
> > on the idempotent repository (releaseExclusiveReadLockOnCommit is an empty 
> > method).
> >
> > I read the following in the Camel in Action book: To commit the Exchange, 
> > the confirm method is invoked on the idempotent repository
> >
> > Does this imply that the Exchange is never committed in case of the FTP 
> > component. Is this expected behaviour?
> >
> > Thanks,
> >
> > Robin
>
>
>
> --
> Claus Ibsen
> -----------------
> http://davsclaus.com @davsclaus
> Camel in Action 2: https://www.manning.com/ibsen2

Reply via email to