Hi all !
As the ftp can not be transacted, the poller should use a sendSync
robust-in-only mep to send the file to the next component. This would
return a fault if the
http invoker is down, and the ftp poller would not delete the file until
it has been sucessfully
processed. The downside of this is that the ftp poller should keep
track of redelivery.
This may be considered as a bug (please raise a jira).
IMHO, using message persistence, as George said, would not solve the
problem, as persistence is
about how to not loose messages in a crash. Currently, persistence is
only supported on the jca flow
but should be on ths jms flow too. The key point is that, when a message
is sent to the nmr, it will
be persisted and will be discarded when the component has succesfully
processed it. The main problem
is to identify a sucessfull handling: if like i said in the previous
paragraph, we use faults to send errors, the
processing of the message IS successfull, even if the result is not ...
But if the provider sends an exception,
it will not be propagated to the consumer :-( .
Cheers,
Guillaume Nodet
George Gastaldi wrote:
This is an excellent question and I would like that to be answered to.
Anyway, ServiceMix 1.1 has a Message Exchange persistence feature that
sounds nice. Haven´t figure how to do it yet, but seems interesting.
Guillaume please, can you explain us or supply a documentation for this
feature ?
Thx
"HURST, John B" <[EMAIL PROTECTED]> wrote
in message
news:[EMAIL PROTECTED]
Hi All,
Various items on the wiki describe ESB infrastructure and ServiceMix in
particular as "reliable". Does this imply it can guarantee delivery of
messages? What do I need to do to make that happen?
For example I wired an FTPPoller to an HttpInvoker. I run ServiceMix
when the HTTP destination service is down. My file is downloaded and
deleted from the FTP server, and lost. If I bring up the HTTP
destination later, the message is not delivered, it's given up. I'd
like to see it either delivered later somehow, or at least rerouted
somewhere else (reliably) for later retry.
Is this supposed to be part of ServiceMix, or is it supposed to be user
code? If ServiceMix can help with this, please refer me to where I can
learn about it. (Perhaps the problem is that HttpInvoker is not letting
SM know that it's failed -- but it does seem to be throwing an exception
...)
Alternatively, if this is a ridiculous question, let me know!
John Hurst
Wellington, New Zealand
______________________________________________________________________
This email has been scanned by the MessageLabs Email Security System.
For more information please visit http://www.messagelabs.com/email
______________________________________________________________________
