[ 
https://issues.apache.org/jira/browse/AMQNET-305?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12980693#action_12980693
 ] 

Kaminiecki commented on AMQNET-305:
-----------------------------------

Sorry.

So

3 brokers in one subnetwork and one in another.

connect clients with failover. make sure that none of them connect directly to 
teh 4th broker in another subnewtork.

make one of them to produce messages.

At this time break a connect to subnetwork with 3 brokers (turn off switch or 
router which know route to them) make sure You cant ping them. Only one should 
be pingable:) 

Solution with demonstrator application is available for download at 
http://rapidshare.com/#!download|863tl|442163550|amqFailoverNMS1.5.0.zip|842

> Failover reconnect doesn't work correct in NMS 1.5.0 and memory usage is 
> still growing.
> ---------------------------------------------------------------------------------------
>
>                 Key: AMQNET-305
>                 URL: https://issues.apache.org/jira/browse/AMQNET-305
>             Project: ActiveMQ .Net
>          Issue Type: Bug
>          Components: ActiveMQ, NMS
>    Affects Versions: 1.5.0
>         Environment: windows 7
>            Reporter: Kaminiecki
>            Assignee: Jim Gomes
>            Priority: Critical
>
> I use  four brokers 5.4.2 in the cluster.
> activemq client 1.4.1 and new NMS 1.5.0
> I just started testing a new NSM 1.5.0 and I had a problem with the network 
> through which I have not had access to three of the brokers during the test.
> However, I had one visible on the network but the client did not switched to. 
> Just so happens that one is in a different subnet than the three others.
> I had three clients connected to one topic with the uri  failover (4 brokes 
> url) All clients had the same uri. One of them was sending 10 messages per 
> second.
> When network connection dissapeard with three of the brokers. Message 
> producer began to consume more memory. It was growing and growing. 
> When I wanted to stop them, the producer stuck!

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.

Reply via email to