Hi
I have updated the FAQ with better examples how to stop a route (or
Camel) from a route
https://cwiki.apache.org/confluence/display/CAMEL/How+can+I+stop+a+route+from+a+route
The source code for the example will be committed, when ASF SVN again works.
There was in fact more problems with the old FAQ to stop a route using
the same thread.
Not only the problem identified by Babak, but you also have problem,
that when the route is stopped
it may close resources and whatnot currently in-use, that the current
message was needed to use
when it is done. For example a JMS consumer with TX would need to ack
the message etc.
A mail message need to set a SEEN flag on the mail etc.
So spinning off a new thread to stop a route is best practice, this is
also how you would normally stop routes, or the application itself.
On Sun, Feb 5, 2012 at 4:34 PM, Claus Ibsen <claus.ib...@gmail.com> wrote:
> Hi
>
> No we should not do that.
>
> People should really not stop a route using the same thread that is
> process currently messages from a route. But instead let another
> thread stop the route.
>
> There is no good solution for stopping a route with the same thread in
> mid flight of processing messages on a route. Its like sawing off the
> branch you are sitting on.
>
> I have added a note on the FAQ.
>
>
>
> On Sun, Feb 5, 2012 at 4:28 PM, Babak Vahdat
> <babak.vah...@swissonline.ch> wrote:
>> Hi Claus,
>>
>> with the usage of ScheduledBatchPollingConsumer and it's derivates inside
>> routes there's an issue while doing context.stopRoute("myCoolRoute") as
>> Camel tries to shutdown the route gracefully and as the mentioned class
>> does:
>>
>> if (answer == 0 && isPolling()) {
>> // force at least one pending exchange if we are polling as
>> there is a little gap
>> // in the processBatch method and until an exchange gets
>> enlisted as in-flight
>> // which happens later, so we need to signal back to the
>> shutdown strategy that
>> // there is a pending exchange. When we are no longer polling,
>> then we will return 0
>> LOG.trace("Currently polling so returning 1 as pending
>> exchanges");
>> answer = 1;
>> }
>>
>> it waits and waits until the (30 seconds) timeout reaches although no *real*
>> inflight exchanges are available at all! I realized this behavior as I was
>> trying to provide a working example for the user (you see already by this
>> thread).
>>
>> However doing main.stop() (as by the upper example) did *not* behave like
>> this!
>>
>> A possible fix (may be dirty) would be to check that if we already returned
>> X times a hard-coded value of 1 then we should better return 0 instead to
>> signal there's no real inflight exchange available.
>>
>> Babak
>>
>>
>> --
>> View this message in context:
>> http://camel.465427.n5.nabble.com/File-consumer-waiting-for-it-to-start-tp5455062p5458048.html
>> Sent from the Camel - Users mailing list archive at Nabble.com.
>
>
>
> --
> Claus Ibsen
> -----------------
> FuseSource
> Email: cib...@fusesource.com
> Web: http://fusesource.com
> Twitter: davsclaus, fusenews
> Blog: http://davsclaus.blogspot.com/
> Author of Camel in Action: http://www.manning.com/ibsen/
--
Claus Ibsen
-----------------
FuseSource
Email: cib...@fusesource.com
Web: http://fusesource.com
Twitter: davsclaus, fusenews
Blog: http://davsclaus.blogspot.com/
Author of Camel in Action: http://www.manning.com/ibsen/
