activemq-users mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Tlholoe, Peter" <PTlho...@fnb.co.za>
Subject RE: ActiveMQ hangs
Date Tue, 21 Aug 2012 13:36:27 GMT
Hi Torsten,
We are using ActiveMQ 5.4.2
Thank you for your help, I will try that and advise of the results. 

-----Original Message-----
From: Torsten Mielke [mailto:torsten@fusesource.com] 
Sent: 21 August 2012 03:25 PM
To: users@activemq.apache.org
Subject: Re: ActiveMQ hangs

What version of ActiveMQ are you using? 
Even with producer flow control disabled, the broker will still suspend producers when it
reaches any of the configured systemUsage limits. If you don't specify any systemUsage in
your brokers configuration, a default and hard coded systemUsage applies. So producers can
still be blocked. 
Its perhaps better to explicitly set the systemUsage to your requirements than rely on the
defaults.

Also, when you reach 70% of the configured queue destination memory limit, the cursor will
not store any more new msgs in memory but reload them from the persistence adapter when its
time to dispatch them. Persistent queue messages get written to kahadb right when being received
by the broker.

Regards,
Torsten Mielke



On Aug 21, 2012, at 2:30 PM, Tlholoe, Peter wrote:

> Hi,
> I am currently monitoring all the queues/ topics at the moment. In the interim I did
disable producer-flow-control, and increasing the memoryLimit=1024Mb, but I didn't activate
systemUsage, and I read that the memory limit configuration is dependent on the systemUsage
configuration. 
> My idea of this is that in the event of a slow consumer, all the message that are not
yet consumed can be offlined to a file system, so as not to exhaust broker resources and avoid
profucer flow control kicking in. 
> 
>        <destinationPolicy>
>            <policyMap>
>              <policyEntries>
>                <policyEntry topic=">" producerFlowControl="false" memoryLimit="1024mb">
>                  <pendingSubscriberPolicy>
>                    <vmCursor />
>                  </pendingSubscriberPolicy>
>                </policyEntry>
>                <policyEntry queue=">" producerFlowControl="false" memoryLimit="1024mb">
>                  <!-- Use VM cursor for better latency
>                       For more information, see:
> 
>                       http://activemq.apache.org/message-cursors.html
> 
>                  <pendingQueuePolicy>
>                    <vmQueueCursor/>
>                  </pendingQueuePolicy>
>                  -->
>                </policyEntry>
>              </policyEntries>
>            </policyMap>
>        </destinationPolicy>
> 
> 
> System Usage
>      <!--  <systemUsage>
>            <systemUsage>
>                <memoryUsage>
>                    <memoryUsage limit="20 mb"/>
>                </memoryUsage>
>                <storeUsage>
>                    <storeUsage limit="1 gb"/>
>                </storeUsage>
>                <tempUsage>
>                    <tempUsage limit="100 mb"/>
>                </tempUsage>
>            </systemUsage>
>        </systemUsage>		-->
> 
> 
> -----Original Message-----
> From: Torsten Mielke [mailto:torsten@fusesource.com]
> Sent: 21 August 2012 10:46 AM
> To: users@activemq.apache.org
> Subject: Re: ActiveMQ hangs
> 
> Hello,
> 
> In your Camel route you seem to route message from various destinations on the broker
to other destinations on the same broker.
> Is it possible that the destinations you route message to are getting filled up (not
all, but some of them). 
> If that happens, producer flow control might kick in, causing the Camel route producers
to be put on hold until more messages get consumed. 
> If the Camel route producer is stalled, the entire Camel route is stalled.
> http://activemq.apache.org/producer-flow-control.html
> 
> I suggest you monitor all queues and topics on the broker using either the web console
or JMX. 
> Also, a thread dump of the broker and the Camel route (in case it runs external to the
broker) could help you to find out in what state the broker and your Camel route are. 
> 
> 
> 
> Torsten Mielke
> torsten@fusesource.com
> tmielke@blogspot.com
> 
> 
> On Aug 21, 2012, at 9:32 AM, Tlholoe, Peter wrote:
> 
>> We currently have a problem in our production environment. We have one general purpose
topic where all messages are published to, and A Camel  subscribed to that general purpose
topic, apply rules to route the messages to different topics. That consumer constantly has
higher MessageCountAwaitingAcknowledge and higer DispachedQueueSize (around 2000). Our PROD
is a high message traffic message environment.
>> 
>> Any pointers on what might be causing this. Please see attached our activemq.xml
and camel.xml.
>> To read FirstRand Bank's Disclaimer for this email click on the following address
or copy into your Internet browser: 
>> https://www.fnb.co.za/disclaimer.html
>> 
>> If you are unable to access the Disclaimer, send a blank e-mail to firstrandbankdisclaimer@fnb.co.za
and we will send you a copy of the Disclaimer.
>> <activemq.xml><camel.xml>
> 
> 
> 
> 
> To read FirstRand Bank's Disclaimer for this email click on the following address or
copy into your Internet browser: 
> https://www.fnb.co.za/disclaimer.html
> 
> If you are unable to access the Disclaimer, send a blank e-mail to 
> firstrandbankdisclaimer@fnb.co.za and we will send you a copy of the Disclaimer.

Torsten Mielke
torsten@fusesource.com
tmielke@blogspot.com



To read FirstRand Bank's Disclaimer for this email click on the following address or copy
into your Internet browser: 
https://www.fnb.co.za/disclaimer.html 

If you are unable to access the Disclaimer, send a blank e-mail to
firstrandbankdisclaimer@fnb.co.za and we will send you a copy of the Disclaimer.

Mime
View raw message