You are viewing a plain text version of this content. The canonical link for it is here.
Posted to users@activemq.apache.org by "magic.moose" <ma...@avedya.com> on 2007/01/25 11:16:43 UTC
AMQ4.1.0: messages stucked in the queue when send fast
Hey,
I've been fighting the following problem for few hours now.
Lets limit it to a very trivial test. Send, say 10, messages in a for loop,
without waits between iterations via JmsTemplate of Spring. ObjectMessage.
Not all messages are consumed. Sometimes all, more often ca 8-9. Through
JMX, can see unconsumed messages stucked in the queue.
In fact such loss happens starting at 3 messages send in that loop.
No exeptions, warnings etc.
A word about config.
AMQ4.1.0, Jencks 1.1, Spring 1.2.7 and its jmsTemplate.
Consumer thread pool in Jenks set to 5.
Broker is embedded with JVM transport. (embedded in connection or separeted
bean with xml config - both behaves the same).
No persistance enabled. Well, at least I think so.
<broker persistent="false" useJmx="true">
<transportConnectors>
<transportConnector uri="vm://localhost" />
</transportConnectors>
<memoryManager>
<usageManager id="memory-manager" limit="20 MB"/>
</memoryManager>
</broker>
If that helps, the ObjectMessage contains streamed javax.mail.Message
(MimeMessage de facto)
If I wait, say 100ms or more, between iterations then things are fine.
It looks pretty scary to me. Maybe Jencks problem? Any idea?
M.M.
--
View this message in context: http://www.nabble.com/AMQ4.1.0%3A-messages-stucked-in-the-queue-when-send-fast-tf3097294.html#a8599367
Sent from the ActiveMQ - User mailing list archive at Nabble.com.
Re: AMQ4.1.0: messages stucked in the queue when send fast
Posted by "magic.moose" <ma...@avedya.com>.
Okay, patch applied. solved the problem (so far).
Just out of curiosity, I find this bug really crucial, I would even say a
blocker. If it really solves the problem I encoutered (not by accident but
directly) then I guess it MUST be very often. How do you think?
Thanks for helping me out.
M.M.
Christopher G. Stach II wrote:
>
> magic.moose wrote:
>>
>> How can I get a version with a patch applied?
>> svn only?
>>
>> M.M.
>>
>> Christopher G. Stach II wrote:
>>> AMQ-1078?
>
> I attached the diff to the issue. You just need to recompile that
> single class after you patch it.
>
> --
> Christopher G. Stach II
>
>
>
--
View this message in context: http://www.nabble.com/AMQ4.1.0%3A-messages-stuck-in-the-queue-when-sent-fast-tf3097294.html#a8647643
Sent from the ActiveMQ - User mailing list archive at Nabble.com.
Re: AMQ4.1.0: messages stucked in the queue when send fast
Posted by "Christopher G. Stach II" <cg...@ldsys.net>.
magic.moose wrote:
>
> How can I get a version with a patch applied?
> svn only?
>
> M.M.
>
> Christopher G. Stach II wrote:
>> AMQ-1078?
I attached the diff to the issue. You just need to recompile that
single class after you patch it.
--
Christopher G. Stach II
Re: AMQ4.1.0: messages stucked in the queue when send fast
Posted by "magic.moose" <ma...@avedya.com>.
How can I get a version with a patch applied?
svn only?
M.M.
Christopher G. Stach II wrote:
>
> magic.moose wrote:
>> Hey,
>>
>> I've been fighting the following problem for few hours now.
>> Lets limit it to a very trivial test. Send, say 10, messages in a for
>> loop,
>> without waits between iterations via JmsTemplate of Spring.
>> ObjectMessage.
>> Not all messages are consumed. Sometimes all, more often ca 8-9. Through
>> JMX, can see unconsumed messages stucked in the queue.
>> In fact such loss happens starting at 3 messages send in that loop.
>> No exeptions, warnings etc.
>>
>> A word about config.
>> AMQ4.1.0, Jencks 1.1, Spring 1.2.7 and its jmsTemplate.
>> Consumer thread pool in Jenks set to 5.
>> Broker is embedded with JVM transport. (embedded in connection or
>> separeted
>> bean with xml config - both behaves the same).
>> No persistance enabled. Well, at least I think so.
>>
>> <broker persistent="false" useJmx="true">
>> <transportConnectors>
>> <transportConnector uri="vm://localhost" />
>> </transportConnectors>
>> <memoryManager>
>> <usageManager id="memory-manager" limit="20 MB"/>
>> </memoryManager>
>> </broker>
>>
>> If that helps, the ObjectMessage contains streamed javax.mail.Message
>> (MimeMessage de facto)
>>
>> If I wait, say 100ms or more, between iterations then things are fine.
>>
>>
>> It looks pretty scary to me. Maybe Jencks problem? Any idea?
>>
>> M.M.
>>
>>
>
> AMQ-1078?
>
> --
> Christopher G. Stach II
>
>
>
--
View this message in context: http://www.nabble.com/AMQ4.1.0%3A-messages-stuck-in-the-queue-when-sent-fast-tf3097294.html#a8634427
Sent from the ActiveMQ - User mailing list archive at Nabble.com.
Re: AMQ4.1.0: messages stucked in the queue when send fast
Posted by "Christopher G. Stach II" <cg...@ldsys.net>.
magic.moose wrote:
> Hey,
>
> I've been fighting the following problem for few hours now.
> Lets limit it to a very trivial test. Send, say 10, messages in a for loop,
> without waits between iterations via JmsTemplate of Spring. ObjectMessage.
> Not all messages are consumed. Sometimes all, more often ca 8-9. Through
> JMX, can see unconsumed messages stucked in the queue.
> In fact such loss happens starting at 3 messages send in that loop.
> No exeptions, warnings etc.
>
> A word about config.
> AMQ4.1.0, Jencks 1.1, Spring 1.2.7 and its jmsTemplate.
> Consumer thread pool in Jenks set to 5.
> Broker is embedded with JVM transport. (embedded in connection or separeted
> bean with xml config - both behaves the same).
> No persistance enabled. Well, at least I think so.
>
> <broker persistent="false" useJmx="true">
> <transportConnectors>
> <transportConnector uri="vm://localhost" />
> </transportConnectors>
> <memoryManager>
> <usageManager id="memory-manager" limit="20 MB"/>
> </memoryManager>
> </broker>
>
> If that helps, the ObjectMessage contains streamed javax.mail.Message
> (MimeMessage de facto)
>
> If I wait, say 100ms or more, between iterations then things are fine.
>
>
> It looks pretty scary to me. Maybe Jencks problem? Any idea?
>
> M.M.
>
>
AMQ-1078?
--
Christopher G. Stach II