You are viewing a plain text version of this content. The canonical link for it is here.

Posted to users@servicemix.apache.org by Ivan Pryvalov <i....@smiss.com.ua> on 2009/11/10 12:41:32 UTC

Timeout Exception in MEX under heavy load

Hello to all,

Seems to be I faced some issue(probably bug) during heavy loading of SMX.

I created JIRA task for it.

https://issues.apache.org/activemq/browse/SM-1909


Any ideas how to check/detect cause of Exception?

I tried to research in SMX code, but it was not successfully yet.


Thanks,
Truly yours,
Ivan Pryvalov.

Re: Timeout Exception in MEX under heavy load

Posted by Ivan Pryvalov <i....@smiss.com.ua>.

Hi Maciej,

Thanks for you reply,

I have some comment:

Maciej Prochniak пишет:
> Hi,
> I think this is not really a bug, rather behaviour specified by JBI + 
> the way seda flow works.
> JBI 5.5.2.1.3: the exchange should be set to ERROR status - this is
> relevant note around line 500 of DeliveryChannelImpl
> 
> Under heavy load following scenario can occur:
> - you do sendSync with timeout

I use exactly sendSync() method. I checked sources of SMX, default value used for timeout is 0. Method .wait(0) is called, i.e. infinity. Am I right here? 


Truly yours,
Ivan Pryvalov.

> - the exchange stucks is seda queue
> - it times out - it is marked as timed out (aborted) - and JBI spec says
> it cannot be dealt with anymore
> - the message is taken from seda queue by destination endpoint - and
> when it tries to do sth with it (e.g. do channel.doSend()...) - error 
> is thrown.
> 
> The point is that destination endpoint should be aware that sender 
> does no longer wait for result
> 
> br,
> maciek
> 
> On Tue, 2009-11-10 at 13:41 +0200, Ivan Pryvalov wrote:
>> Hello to all,
>>
>> Seems to be I faced some issue(probably bug) during heavy loading of SMX.
>>
>> I created JIRA task for it.
>>
>> https://issues.apache.org/activemq/browse/SM-1909
>>
>>
>> Any ideas how to check/detect cause of Exception?
>>
>> I tried to research in SMX code, but it was not successfully yet.
>>
>>
>> Thanks,
>> Truly yours,
>> Ivan Pryvalov.
> 
>

Re: Timeout Exception in MEX under heavy load

Posted by Maciej Prochniak <mp...@touk.pl>.

Hi,
I think this is not really a bug, rather behaviour specified by JBI + 
the way seda flow works.
JBI 5.5.2.1.3: the exchange should be set to ERROR status - this is
relevant note around line 500 of DeliveryChannelImpl

Under heavy load following scenario can occur:
- you do sendSync with timeout
- the exchange stucks is seda queue
- it times out - it is marked as timed out (aborted) - and JBI spec says
it cannot be dealt with anymore
- the message is taken from seda queue by destination endpoint - and
when it tries to do sth with it (e.g. do channel.doSend()...) - error 
is thrown.

The point is that destination endpoint should be aware that sender 
does no longer wait for result

br,
maciek

On Tue, 2009-11-10 at 13:41 +0200, Ivan Pryvalov wrote:
> Hello to all,
> 
> Seems to be I faced some issue(probably bug) during heavy loading of SMX.
> 
> I created JIRA task for it.
> 
> https://issues.apache.org/activemq/browse/SM-1909
> 
> 
> Any ideas how to check/detect cause of Exception?
> 
> I tried to research in SMX code, but it was not successfully yet.
> 
> 
> Thanks,
> Truly yours,
> Ivan Pryvalov.

Re: Timeout Exception in MEX under heavy load

Posted by Jean-Baptiste Onofré <jb...@nanthrax.net>.

Hi Ivan,

I've just seen your Jira. We will give you a feedback as soon as possible.

Regards
JB

Ivan Pryvalov wrote:
> Hello to all,
> 
> Seems to be I faced some issue(probably bug) during heavy loading of SMX.
> 
> I created JIRA task for it.
> 
> https://issues.apache.org/activemq/browse/SM-1909
> 
> 
> Any ideas how to check/detect cause of Exception?
> 
> I tried to research in SMX code, but it was not successfully yet.
> 
> 
> Thanks,
> Truly yours,
> Ivan Pryvalov.