[ 
https://issues.apache.org/jira/browse/CAMEL-9635?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15162981#comment-15162981
 ] 

Christoph Läubrich commented on CAMEL-9635:
-------------------------------------------

If this is intentional this really should be documented and at leas issue a 
warning if the settings mismatch. With current implementation/documentation it 
was a hard way to find out why settings regarding timout get silently ignored.

The UseCase is having different (independent) components all consuming from the 
same port with different demands regarding continuation setting (e.g. timeouts).

> Continuation settings are only evaluated on first creation for each 
> port/connector
> ----------------------------------------------------------------------------------
>
>                 Key: CAMEL-9635
>                 URL: https://issues.apache.org/jira/browse/CAMEL-9635
>             Project: Camel
>          Issue Type: Improvement
>          Components: camel-jetty
>    Affects Versions: 2.16.2
>            Reporter: Christoph Läubrich
>            Priority: Minor
>
> Assume the following endpoints configured (in this order!):
> {code}<from uri="jetty:http://0.0.0.0/myservice1/?useContinuation=false"/>
> <from uri="jetty:http://0.0.0.0/myservice2/?useContinuation=true"/>
> <from uri="jetty:http://0.0.0.0:9090/myservice3/"/>
> <from 
> uri="jetty:http://0.0.0.0:9090/myservice4/?continuationTimeout=10000"/>{code}
> This results in endpoints myservice1 and myservice2 NOT using continuation, 
> and myservice4 usind the default timeout of 30seconds.
> This is especially confusing if you use something like bluprint where 
> different bundles might use the jetty endpoint with different configurations. 
> In such a case it is highly dependant on the start order what configuration 
> wins that makes this issue more confusing and hard to debug.
> The cause of this is, that the configuration of the endpoint is only 
> evaluated once when the *Connector* is created and then never changed as long 
> as you do not shut down ALL endpoints of this connector.
> This can be fixed by reading the configuration on each request instead of 
> once at connector creation.
> I have seen this issue on Camel 2.10 but the code seems not to have changed 
> in the master until now, so I'll provide a patch for the master branch only.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

Reply via email to