[
https://issues.apache.org/jira/browse/CAMEL-4948?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13195000#comment-13195000
]
David Cifer commented on CAMEL-4948:
------------------------------------
I changed my mind. I was able to reproduce the issue.
I've created a new route, which generates an exception, the same way as above.
I made the previous queue to pass the elements at the end of the route.
I've run the test twice and again I begin to lose messages. This time activemq
restart didn't sorted the problem.
> Chaos arround durable jms transacted configuration
> --------------------------------------------------
>
> Key: CAMEL-4948
> URL: https://issues.apache.org/jira/browse/CAMEL-4948
> Project: Camel
> Issue Type: Bug
> Components: camel-activemq, camel-core, camel-jms
> Affects Versions: 2.8.1
> Environment: JDK=1.6.24 Oracle
> Reporter: David Cifer
> Priority: Minor
> Attachments: camelasync.zip
>
> Original Estimate: 168h
> Remaining Estimate: 168h
>
> Maybe this is two bug and maybe I miss understand something, but here is the
> issue.
> I've created a test. A route, which reads a JMS transacted durable, but every
> fifth message is thrown back by an exception.
> For example I send 10 message into the durrable, than I generate 2 exception.
> One for the fifth one and one for the nineth (because the fifth one going to
> be resend as transaction failed).
> Here is the route:
> from("jms:topic:test2?clientId=testClient&durableSubscriptionName=testSubs&transacted=true")
> .log("Processign from durable: ${body}").delay(100)
> .process(new ExceptionGenerator())
> .log("Retrieved from durable: ${body}").to("mock:durable");
> Code of the Exception generator:
> public void process(Exchange arg0) throws Exception {
> int i = counter.incrementAndGet();
> if (i % 5 == 0) {
> throw new IllegalArgumentException("Every fifth message is illegal: "
> + arg0.getIn().getBody());
> }
> }
> The issue, what I had is that the first message (the fifth) is rolled back
> just nice. But after that all message simply lost.
> This is I think the reason that since 2.8.x you have to use CACHE_NONE
> otherwise you mess up the transaction cache.
> Here is the second part of the problem:
> The whole test works just fine under 2.7.5
> But to make it start above that is a configuration nightmare. For example
> clientId and transacted options in the route definition cannot appear
> together.
> To make it work above 2.7.5 You have to add manually to the connection
> configuration:
> ActiveMQConnectionFactory connectionFactory = new
> ActiveMQConnectionFactory(ACTIVE_MQ_HOST);
> connectionFactory.setClientID("testClient"); // <-- This is necessary
> above 2.7.5 meaning 2.8.x. Under 2.8.x so example 2.7.5 will break
> configuration
> ctx.addComponent("activemq",
> ActiveMQComponent.activeMQComponent(ACTIVE_MQ_HOST));
> Summary:
> Above 2.7.5 clientId and transacted cannot appear in the same durable
> specification. Which makes configuration brake backward compatibilty. It has
> to be configured on connection factory level, which means I have to create a
> connectionfactory for each durable endpoint.
> Above 2.7.5 only the first transaction got rolled back. Upcomming failiures
> got lost.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators:
https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
