[sr-dev] Dialog module: "CRITICAL bogus event 6 in state 2

[Klaus Feichtinger]

Hello,

today I've tested the committed fix in an actual development version
(tar.gz) from yesterday.

In this test I have compared the released Kamailio 3.0.2 version with
the development version. The bug was reproduceable within a minute in
the released version. But in the development version this fault didn't
occur. The configuration is the same as at the beginning when I found
this problem!

How was it tested?
With SIPp (UAC + UAS) in both cases. The development version was tested
with a standard call setup over sip-route and a call rate of 20 calls /
second for a period of 60 minutes. No bogus event! In the released
version the bogus event occured after 30 seconds with a call rate of 10
calls / seconds.

Note - this is not a classification about a call rate when this fault
may start! It was only a free decision without testing the intervals in
between. At the beginning I started with a call rate of 2 / second. In
that case no error occured in both versions.

I hope this test is enough for a first comment.

regards,

Klaus

Am 05.07.2010 12:29, schrieb Daniel-Constantin Mierla:
> Hello,
>
> last week I committed an update for this. If any of you can run some
> tests and report back, would be great. Plan is to backport to 3.0 as
> soon as some testing was done to be sure it does not break something
> else.
>
> Thanks,
> Daniel
>
>
> On 6/14/10 5:35 PM, Ovidiu Sas wrote:
>> There is a new TM callback, but it is an alias to the old
>> TMCB_RESPONSE_OUT.
>>
>> /* response ready - might need to be executed before sending reply
>> out */
>> #define TMCB_RESPONSE_READY             TMCB_RESPONSE_OUT
>>
>>
>> Regards,
>> Ovidiu Sas
>>
>> On Mon, Jun 14, 2010 at 11:31 AM, Daniel-Constantin Mierla
>> <miconda at gmail.com>  wrote:
>>   
>>>
>>> On 6/14/10 4:06 PM, Ovidiu Sas wrote:
>>>     
>>>> [...]
>>>> In kamailio 3.0, this is a regression due to the new ser tm module.
>>>> The sip-router tm module must be updated in order to avoid this
>>>> regression.
>>>>
>>>>        
>>> IIRC, there was a corresponding TM callback in 3.0 that was used
>>> instead --
>>> a callback executed before sending the reply. Unfortunately I cannot
>>> check
>>> right now the sources, I will do it later.
>>>
>>> Cheers,
>>> Daniel
>>>
>>> -- 
>>> Daniel-Constantin Mierla
>>> Asterisk SIP Router Masterclass
>>> Malaga, Spain - July 5-9, 2010
>>> http://www.asipto.com/
>>>
>>>
>>>      
>