Thanks Amila for bringing this up. The idea of this implementation was to
introduce a generic implementation for any AMQP based message broker to
detect a connection drop and re-connect. The problem was with the message
subscribers. Yes we need to have a message consumer for the ping topic to
avoid the message broker being overloaded.

I started a discussion on this topic here [1] with respective to ActiveMQ.
It seems like ActiveMQ provides a solution for this problem. The concern we
have here is that whether this is something the message broker should
handle. If so we could simply ignore this functionality in Stratos.

[1]
http://stackoverflow.com/questions/23503720/how-to-detect-a-connection-drop-in-an-activemq-subscriber

Thanks


On Sun, May 11, 2014 at 10:37 PM, Udara Liyanage <ud...@wso2.com> wrote:

> Hi,
>
> Yes it might cause problems in long run. Some cleanup process to subscribe
> to the ping topic and discard might be useful.
>
>
>
> Touched, not typed. Erroneous words are a feature, not a typo.
> On May 11, 2014 8:55 AM, "Amila Maha Arachchi" <ami...@wso2.com> wrote:
>
>> Hi Devs,
>>
>> I am working with a Stratos deployment these days. In our deployment, we
>> are using WSO2 Message Broker as the message broker. There I see the
>> following log getting printed very frequently (every second).
>>
>> TID: [0] [MB] [2014-05-05 08:31:52,352]  INFO
>> {org.wso2.andes.server.exchange.TopicExchange} -  Message routing key: ping
>> No routes. {org.wso2.andes.server.exchange.TopicExchange}
>>
>> TID: [0] [MB] [2014-05-05 08:31:52,352]  WARN
>> {org.wso2.andes.server.AMQChannel} -  MESSAGE DISCARDED: No routes for
>> message - Message[(HC:449176483 ID:366318838955900928 Ref:0)]:
>> 366318838955900928; ref count: 0 {org.wso2.andes.server.AMQChannel}
>>
>> TID: [0] [MB] [2014-05-05 08:31:52,621]  INFO
>> {org.wso2.andes.server.exchange.TopicExchange} -  Message routing key: ping
>> No routes. {org.wso2.andes.server.exchange.TopicExchange}
>>
>> TID: [0] [MB] [2014-05-05 08:31:52,621]  WARN
>> {org.wso2.andes.server.AMQChannel} -  MESSAGE DISCARDED: No routes for
>> message - Message[(HC:1522105342 ID:366318839026417664 Ref:0)]:
>> 366318839026417664; ref count: 0 {org.wso2.andes.server.AMQChannel}
>>
>> I came to know that this is due to a mechanism used to check whether the
>> Message Broker (MB) is alive. i.e. subscribers publish an event to this
>> "ping" topic to see whether the MB is alive, but no one is consuming it.
>> Thats why the above log is getting printed.
>>
>> Since no one is consuming this messages, is it possible that MB gets
>> loaded with this message and cause problems?
>>
>> Regards,
>> Amila.
>>
>> --
>> *Amila Maharachchi*
>> Senior Technical Lead
>> WSO2, Inc.; http://wso2.com
>>
>> Blog: http://maharachchi.blogspot.com
>> Mobile: +94719371446
>>
>>


-- 
Imesh Gunaratne

Technical Lead, WSO2
Committer & PPMC Member, Apache Stratos

Reply via email to