You are viewing a plain text version of this content. The canonical link for it is here.
Posted to users@camel.apache.org by Sven Bauhan <sv...@ast.dfs.de> on 2013/04/26 15:49:08 UTC
AggregationStrategy and wrong sequence order
Hi,
I try to implement a fault handling in my AggregationStrategy for
missing segments or a wrong sequence order.
The standard for my protocol says, when the segments are not received in
correct order, the receiver shall send an error response and withdraw
all previously received segments of the message. The sender will send
the complete message again on error response.
In the method aggregate(Exchange oldExchange, Exchange newExchange) I
throw an exception. This is caught in the route and an error response is
sent back. But with the next segment, the oldExchange is unchanged.
What I would expect is that the aggregator discards all messages
previously received for this correlation key. So the next segment
received would have an oldExchange=null.
Is it possible to make the aggregator act this way? how can I implement
this?
Thanks, Sven
Re: AggregationStrategy and wrong sequence order
Posted by Sven Bauhan <sv...@ast.dfs.de>.
Ok, new issue: https://issues.apache.org/jira/browse/CAMEL-6325
On 04/27/13 08:06, Claus Ibsen wrote:
> Hi
>
> Yeah that may be possible. Feel free to log a JIRA ticket.
>
>
> On Fri, Apr 26, 2013 at 6:33 PM, Sven Bauhan <sv...@ast.dfs.de> wrote:
>> Ok, but then the timeout for the aggregation can occur, if the sender does
>> not resend the message fast enough.
>>
>> The only solution I found is to set the predicate to true, if an error
>> occured. Then the exception will be thrown in the onCompletion() method of
>> the AggregationStrategy. So the message is complete for the Aggregator.
>>
>> But then I have to implement the closeCorrelationKeyOnCompletion Cache on my
>> own.
>>
>> Better would be if there was a discardOnFailure() method for aggregate() in
>> the route similar to discardOnCompletionTimeout().
>>
>> Sven
>>
>>
>> On 04/26/13 16:46, Claus Ibsen wrote:
>>> Hi
>>>
>>> You can just store some state on the oldExchange before throwing that
>>> exception in the aggregate method.
>>> So the next time, you can find that state on oldExchange and then
>>> regard it as "null".
>>>
>>>
>>>
>>> On Fri, Apr 26, 2013 at 3:49 PM, Sven Bauhan <sv...@ast.dfs.de>
>>> wrote:
>>>> Hi,
>>>>
>>>> I try to implement a fault handling in my AggregationStrategy for missing
>>>> segments or a wrong sequence order.
>>>>
>>>> The standard for my protocol says, when the segments are not received in
>>>> correct order, the receiver shall send an error response and withdraw all
>>>> previously received segments of the message. The sender will send the
>>>> complete message again on error response.
>>>>
>>>> In the method aggregate(Exchange oldExchange, Exchange newExchange) I
>>>> throw
>>>> an exception. This is caught in the route and an error response is sent
>>>> back. But with the next segment, the oldExchange is unchanged.
>>>>
>>>> What I would expect is that the aggregator discards all messages
>>>> previously
>>>> received for this correlation key. So the next segment received would
>>>> have
>>>> an oldExchange=null.
>>>>
>>>> Is it possible to make the aggregator act this way? how can I implement
>>>> this?
>>>>
>>>> Thanks, Sven
>>>>
>>>
>
>
Re: AggregationStrategy and wrong sequence order
Posted by Claus Ibsen <cl...@gmail.com>.
Hi
Yeah that may be possible. Feel free to log a JIRA ticket.
On Fri, Apr 26, 2013 at 6:33 PM, Sven Bauhan <sv...@ast.dfs.de> wrote:
> Ok, but then the timeout for the aggregation can occur, if the sender does
> not resend the message fast enough.
>
> The only solution I found is to set the predicate to true, if an error
> occured. Then the exception will be thrown in the onCompletion() method of
> the AggregationStrategy. So the message is complete for the Aggregator.
>
> But then I have to implement the closeCorrelationKeyOnCompletion Cache on my
> own.
>
> Better would be if there was a discardOnFailure() method for aggregate() in
> the route similar to discardOnCompletionTimeout().
>
> Sven
>
>
> On 04/26/13 16:46, Claus Ibsen wrote:
>>
>> Hi
>>
>> You can just store some state on the oldExchange before throwing that
>> exception in the aggregate method.
>> So the next time, you can find that state on oldExchange and then
>> regard it as "null".
>>
>>
>>
>> On Fri, Apr 26, 2013 at 3:49 PM, Sven Bauhan <sv...@ast.dfs.de>
>> wrote:
>>>
>>> Hi,
>>>
>>> I try to implement a fault handling in my AggregationStrategy for missing
>>> segments or a wrong sequence order.
>>>
>>> The standard for my protocol says, when the segments are not received in
>>> correct order, the receiver shall send an error response and withdraw all
>>> previously received segments of the message. The sender will send the
>>> complete message again on error response.
>>>
>>> In the method aggregate(Exchange oldExchange, Exchange newExchange) I
>>> throw
>>> an exception. This is caught in the route and an error response is sent
>>> back. But with the next segment, the oldExchange is unchanged.
>>>
>>> What I would expect is that the aggregator discards all messages
>>> previously
>>> received for this correlation key. So the next segment received would
>>> have
>>> an oldExchange=null.
>>>
>>> Is it possible to make the aggregator act this way? how can I implement
>>> this?
>>>
>>> Thanks, Sven
>>>
>>
>>
>
--
Claus Ibsen
-----------------
Red Hat, Inc.
FuseSource is now part of Red Hat
Email: cibsen@redhat.com
Web: http://fusesource.com
Twitter: davsclaus
Blog: http://davsclaus.com
Author of Camel in Action: http://www.manning.com/ibsen
Re: AggregationStrategy and wrong sequence order
Posted by Sven Bauhan <sv...@ast.dfs.de>.
Ok, but then the timeout for the aggregation can occur, if the sender
does not resend the message fast enough.
The only solution I found is to set the predicate to true, if an error
occured. Then the exception will be thrown in the onCompletion() method
of the AggregationStrategy. So the message is complete for the Aggregator.
But then I have to implement the closeCorrelationKeyOnCompletion Cache
on my own.
Better would be if there was a discardOnFailure() method for aggregate()
in the route similar to discardOnCompletionTimeout().
Sven
On 04/26/13 16:46, Claus Ibsen wrote:
> Hi
>
> You can just store some state on the oldExchange before throwing that
> exception in the aggregate method.
> So the next time, you can find that state on oldExchange and then
> regard it as "null".
>
>
>
> On Fri, Apr 26, 2013 at 3:49 PM, Sven Bauhan <sv...@ast.dfs.de> wrote:
>> Hi,
>>
>> I try to implement a fault handling in my AggregationStrategy for missing
>> segments or a wrong sequence order.
>>
>> The standard for my protocol says, when the segments are not received in
>> correct order, the receiver shall send an error response and withdraw all
>> previously received segments of the message. The sender will send the
>> complete message again on error response.
>>
>> In the method aggregate(Exchange oldExchange, Exchange newExchange) I throw
>> an exception. This is caught in the route and an error response is sent
>> back. But with the next segment, the oldExchange is unchanged.
>>
>> What I would expect is that the aggregator discards all messages previously
>> received for this correlation key. So the next segment received would have
>> an oldExchange=null.
>>
>> Is it possible to make the aggregator act this way? how can I implement
>> this?
>>
>> Thanks, Sven
>>
>
>
Re: AggregationStrategy and wrong sequence order
Posted by Claus Ibsen <cl...@gmail.com>.
Hi
You can just store some state on the oldExchange before throwing that
exception in the aggregate method.
So the next time, you can find that state on oldExchange and then
regard it as "null".
On Fri, Apr 26, 2013 at 3:49 PM, Sven Bauhan <sv...@ast.dfs.de> wrote:
> Hi,
>
> I try to implement a fault handling in my AggregationStrategy for missing
> segments or a wrong sequence order.
>
> The standard for my protocol says, when the segments are not received in
> correct order, the receiver shall send an error response and withdraw all
> previously received segments of the message. The sender will send the
> complete message again on error response.
>
> In the method aggregate(Exchange oldExchange, Exchange newExchange) I throw
> an exception. This is caught in the route and an error response is sent
> back. But with the next segment, the oldExchange is unchanged.
>
> What I would expect is that the aggregator discards all messages previously
> received for this correlation key. So the next segment received would have
> an oldExchange=null.
>
> Is it possible to make the aggregator act this way? how can I implement
> this?
>
> Thanks, Sven
>
--
Claus Ibsen
-----------------
Red Hat, Inc.
FuseSource is now part of Red Hat
Email: cibsen@redhat.com
Web: http://fusesource.com
Twitter: davsclaus
Blog: http://davsclaus.com
Author of Camel in Action: http://www.manning.com/ibsen