You are viewing a plain text version of this content. The canonical link for it is here.
Posted to users@activemq.apache.org by jgunz <sl...@twcny.rr.com> on 2008/05/07 15:19:18 UTC

ActiveMQ 4.1.1 server freezes

I have been using ActiveMQ 4.1.1 for awhile in a production product and
recently we have been seeing occasional lock-ups/freezes that go away when
we restart ActiveMQ. From our clients or from the ActiveMQ log we cannot see
that there is a problem, we just stop seeing message throughput .

I'm at a loss in regards to how I can debug this problem. Is there some way
I can see what the ActiveMQ server is doing and why it is hanging?

Incase it helps, here is our connect string:
failover:(tcp://localhost:61606)?initialReconnectDelay=1000&useExponentialBackOff=false

We register a transport listener and log interruptions but during these
freezes we do not get any log messages.
-- 
View this message in context: http://www.nabble.com/ActiveMQ-4.1.1-server-freezes-tp17104702s2354p17104702.html
Sent from the ActiveMQ - User mailing list archive at Nabble.com.


Re: ActiveMQ 4.1.1 server freezes

Posted by Hiram Chirino <hi...@hiramchirino.com>.
Could you simulate your error condition in a junit test and post it
for us to review?  I agree the broker should be more isolated.

On Mon, May 12, 2008 at 4:33 PM, jgunz <sl...@twcny.rr.com> wrote:
>
> We identified what was causing the problem. One of our consumers had an
> uncaught exception that was occasionally thrown in the MessageListener
> thread. By simply preventing this exception we no longer see the broker lock
> ups.
>
> Is this expected consumer behavior? It seems that the broker/server should
> be insulated against client problems.
> --
> View this message in context: http://www.nabble.com/ActiveMQ-4.1.1-server-freezes-tp17104702s2354p17189972.html
> Sent from the ActiveMQ - User mailing list archive at Nabble.com.
>
>



-- 
Regards,
Hiram

Blog: http://hiramchirino.com

Open Source SOA
http://open.iona.com

Re: ActiveMQ 4.1.1 server freezes

Posted by jgunz <sl...@twcny.rr.com>.
We identified what was causing the problem. One of our consumers had an
uncaught exception that was occasionally thrown in the MessageListener
thread. By simply preventing this exception we no longer see the broker lock
ups.

Is this expected consumer behavior? It seems that the broker/server should
be insulated against client problems.
-- 
View this message in context: http://www.nabble.com/ActiveMQ-4.1.1-server-freezes-tp17104702s2354p17189972.html
Sent from the ActiveMQ - User mailing list archive at Nabble.com.


Re: ActiveMQ 4.1.1 server freezes

Posted by jgunz <sl...@twcny.rr.com>.

Daryl Richter-3 wrote:
> 
> Do you open and close multiple sessions within the same consumer thread?
> 
> We have seen that this can very reliably cause the broker to lock up.
> 
> It is best to open a single session per thread and close all the way out 
> when done consuming.
> 

Is it wrong to open multiple sessions from the same thread? 

What we typically do is in one of our application threads, create sessions
and register a single consumer with a registered message listener on each.
Likewise we'd create a separate session for each producer we'd need.

-- 
View this message in context: http://www.nabble.com/ActiveMQ-4.1.1-server-freezes-tp17104702s2354p17134918.html
Sent from the ActiveMQ - User mailing list archive at Nabble.com.


Re: ActiveMQ 4.1.1 server freezes

Posted by Daryl Richter <ng...@comcast.net>.
jgunz wrote:
> I have been using ActiveMQ 4.1.1 for awhile in a production product and
> recently we have been seeing occasional lock-ups/freezes that go away when
> we restart ActiveMQ. From our clients or from the ActiveMQ log we cannot see
> that there is a problem, we just stop seeing message throughput .
> 
> I'm at a loss in regards to how I can debug this problem. Is there some way
> I can see what the ActiveMQ server is doing and why it is hanging?
> 
> Incase it helps, here is our connect string:
> failover:(tcp://localhost:61606)?initialReconnectDelay=1000&useExponentialBackOff=false
> 
> We register a transport listener and log interruptions but during these
> freezes we do not get any log messages.

Do you open and close multiple sessions within the same consumer thread?

We have seen that this can very reliably cause the broker to lock up.

It is best to open a single session per thread and close all the way out 
when done consuming.

--
Daryl