activemq-users mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Sharon Barker <sharon.bar...@intalgent.com>
Subject Re: Impact due to wireformat.maxInactivityDuration=0
Date Tue, 24 Jun 2008 13:45:07 GMT
Hi,

Our application had a similar issue bridging between brokers on  
different networks. We were able to reproduce the problem in a testing  
environment by fiddling with iptables to completely sever connectivity  
on one broker machine. In our case that made sense because in the wild  
the devices are physically separate and network issues (latency, lost  
packets, somebody's ill advised firewall rule, etc) happen all the  
time. In our tests we found that "shutdown" was either normal or  
sudden traffic stop (iptables change.)

One additional note which may (or may not) apply to your situation...  
we found that we had to set both brokers to have non-zero values for  
wireformat.maxInactivityDuration for the bridge to come back up at  
all. Once that was done, all was well in both 4.1.2 and 5.1.0 in our  
app.

Hope that helps a little.

Sharon



On Jun 16, 2008, at 6:16 AM, veechand wrote:

>
> Thanks Marco.
> By shutdown do you mean normal shutdown or client/broker getting  
> abruptly
> down. If getting abruptly down what could be reason for the same,  
> this will
> help in reproducing the same?
> We are using 4.1, yet to migrate to 5.1. Since this is critical time  
> in our
> project we are not migrating now and going with this fix.
> Moreover do you mean this issue is
> http://issues.apache.org/activemq/browse/AMQ-1146, if not can you  
> kindly
> point me to the same.
>
>
> Marco Buss wrote:
>>
>> Hi,
>>
>> we ave the same problem here. Setting  
>> wireformat.maxInactivityDuration=0
>> was not a good idea because the shutdown of client or broker (i cant
>> remeber) is not recognised. So try if your client get notice of  
>> broker
>> shutdown and vice versa.
>>
>> Which version of activeMQ did you use? With 5.1 the problem is fixed.
>>
>>
>> veechand wrote:
>>>
>>> In our application we are using ActiveMQ. Currently we facing
>>> "javax.jms.IllegalStateException: The Session is closed", this  
>>> exception
>>> occurs immediately after "Channel was inactive for too long". On
>>> searching on this we found an option for disabling the inactive  
>>> timer,
>>> namely wireformat.maxInactivityDuration=0. We have added this during
>>> intialization of TopicConnectionFactory. Since this issue is not
>>> reproduced consistently, we like to know on the following
>>>   1. Is there any impact due disabling timeout.
>>>   2. Will there be any increase in recourses like socket, file  
>>> handle,
>>> threads, memory
>>>   3. Any consistent way to reproduce the issue.
>>>
>>> Background:
>>>   Our application is a web based application. Uses ActiveMQ to pass
>>> events between JVMs. There are events which are passed so  
>>> infrequently
>>> and those that are passed very frequently.
>>>
>>
>>
>
> -- 
> View this message in context: http://www.nabble.com/Impact-due-to-wireformat.maxInactivityDuration%3D0-tp17859607p17861634.html
> Sent from the ActiveMQ - User mailing list archive at Nabble.com.
>


Mime
View raw message