It could be more easy for us to understand your case by looking at the
routes you have.
On Sat Jul 21 06:40:34 2012, Christian Müller wrote:
Which version of Camel and ActiveMQ do you use?
Which MEP do you use?
Best,
Christian
Sent from a mobile device
Am 20.07.2012 16:26 schrieb "Marco Zapletal" <[email protected]>:
Hi folks,
We have an application where we have two Camel contexts (C1, C2) which
exchange messages via two queues (Q1, Q2). These queues are located on the
same ActiveMQ broker. Thereby, the message flow goes as follows:
C1 -> Q1 -> C2
C2 -> Q2 -> C1
C2 uses furthermore some "internal" queues on the AMQ broker, but I guess
they are not relevant to the problem.
The issue we are facing can be described as follows and happens only when
C1 or C2 go down or have to be restarted
- In case, no messages are produced of either C1/C2 while the other one
restarts, everything is fine - i.e., there is no problem with consuming
messages
- In case, messages are produced of either C1/C2 and are put in the
respective queue, during the absence of the other Camel application, we
gonna face problems with consuming messages from the queues.
We have especially tested this scenario by stopping C2. C1 produces
messages to Q1. Then we restart C2 again and (almost) nothing happened.
- By almost I mean, that the context of C2 starts up without errors. What
is also observed is that when we have 1 concurrentConsumer defined in the
AMQ consumer configuration in C2, 1 message is consumed (if 3
concurrentConsumers are defined, 3 messages are consumed). Afterwards,
consumption stops.
- When restarting C2 again, 1 message is consumed from Q1 (in case of 1
concurrentConsumer)
- C2 exposes also two CXF services as producers of routes. Both of the two
routes have one of those "internal" AMQ queues as their final destination.
When we want to access their respective WSDL URL, the request hangs.
- We have an admin Web application monitoring C2 via JMX. The admin
application hangs due to no response from C2's JMX services (although the
C2 context starts up properly according to the logs).
- Nothing special can be seen in the logs. We examined the logs on DEBUG
level (on Camel as well as on AMQ side) and nothing special could be seen.
- We went back to a rather base config. No transactions, no connection
pools, no caching of consumers/producers. We have experimented with the
prefetch (setting it to 1 or even 0) without success.
- In order to reach proper behavior again, Q1/Q2 (and maybe even the
"internal queues of C2) have to be purged. Then C2 has be to be restarted
again. After this procedure, message passing is back to normal.
Sorry for the long post, but I want to describe the problem as detailed as
possible. Since we have been working on this now for days any help would be
highly appreciated.
Thanks and best regards,
Marco
--
Willem
----------------------------------
FuseSource
Web: http://www.fusesource.com
Blog: http://willemjiang.blogspot.com (English)
http://jnn.javaeye.com (Chinese)
Twitter: willemjiang
Weibo: willemjiang
