Hi

An aggregation strategy is not really intended to be slow, as you say also.

>From the JIRA its mostly because the consumer is not suspendable, and
therefore it cannot be suspended while shutting down and making
exisitng inflight messages complete gracefully.
So you have a window where the consumer may take in new messages while
the route / EIPs are shutting down, and therefore a message can be
rejected.



On Wed, Aug 15, 2018 at 2:19 PM, Dmitry Volodin <[email protected]> wrote:
> Hi Camellers!
>
> During the work on https://issues.apache.org/jira/browse/CAMEL-12668 issue,
> I've detected that if custom aggregation strategy is too slow or has a
> large delays (~5-10 secs and can be emulated with Thread.sleep), the
> default shutdown life-cycle doesn't wait for the aggregation and shutdown
> route early and cause RejectedExecutionException.
>
> I understand that in most cases the aggregation with 5-10 seconds is not
> permitted and deferred route shutdown could solve the problems but what do
> you think about potential investigation in this issue to detect why
> shutdown strategy not waiting for route and consumer?
>
> --
> Best regards,
> Dmitry Volodin



-- 
Claus Ibsen
-----------------
http://davsclaus.com @davsclaus
Camel in Action 2: https://www.manning.com/ibsen2

Reply via email to