activemq-users mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From JRR <...@cisco.com>
Subject Re: ActiveMQ-CPP Producer Flow Control Question
Date Mon, 09 Jul 2012 23:19:51 GMT
Hi Tim,

Before I do this. I think I found the culprit. And while I feel strongly
this is a defect, I could also see it argued as working as designed.

I revisited the configuration page
(http://activemq.apache.org/cms/configuring.html) after browsing the GDB
output and noticed the following configuration option.


http://activemq.apache.org/cms/configuring.html wrote
> 
>  http://activemq.2283324.n4.nabble.com/file/n4653814/amq_connection.jpg 
> 

So I changed my test to included '&connection.sendTimeout=500' (500 ms if I
recall correctly).

I repeated the test and this time it does not lock up. Instead it pauses
then throws an exception.

I would argue that this is a defect because the Producer Flow Control page
(http://activemq.apache.org/producer-flow-control.html) indicates the
following:

http://activemq.apache.org/producer-flow-control.html wrote
> 
> 
> http://activemq.2283324.n4.nabble.com/file/n4653814/client-side-exception.jpg 
> 

The text above makes me believe that we would simply add this configuration
on the broker and an exception would be raised on the client, regardless as
to the client's sendTimeout setting.  Otherwise the send could 'fail' with
either a timeout or a flow control issue. Where I'm only interested in a
flow control issue issue.

Moreover, I argue that the client shouldn't wait for the timeout to declare
the flow control exception. It should do so immediately. If I create a new
connection and invoked send, it will throw the flow control exception
immediately.

Although I suppose it could be argued that I should also be concerned with
timeouts too otherwise I could get an indefinite lockup if the broker never
responds. But I'd say that's beside the point. We shouldn't lock up the send
side, if we've followed the broker configuration rules to prevent permanent
throttling.

What are you thoughts? Should I still open this using Jira (I ask up front
since I see you are the lead developer for these changes in the CPP client).

Thank you for all your time,

-=John

--
View this message in context: http://activemq.2283324.n4.nabble.com/ActiveMQ-CPP-Producer-Flow-Control-Question-tp4653781p4653814.html
Sent from the ActiveMQ - User mailing list archive at Nabble.com.

Mime
View raw message