[OpenSIPS-Users] assigning dialog profiles

Bogdan-Andrei Iancu bogdan at voice-system.ro
Mon Mar 9 19:09:17 CET 2009


Hi Jeff,

Jeff Pyle wrote:
> Hi Bogdan,
>
> Ok.  Is it safe to assume that this is "caused" by low network latency? 
not necessary...it might be because of long 200OK processing, like 
tracing in DB.
>  Do
> you have any feel for how long the dialog module needs to process the 200?
>   
it is not dialog module, but about other module doing work on the 200 OK.
> I suppose that's not fair to ask because it would depend on machine speed.
>   
no, it is not the machine, but the CFG. Anyhow this is more or less a 
bug in how the dialog state machine is updated.
> Does this problem cause any problems with the actual operation of the dialog
> module at this point?
>   
not really (except the err logs) - your dialogs will be in "confirmed 
but not acknowledged" state but it is equivalent to "confirmed and 
acknowledged" state
> With the restart... It's as if I just restarted it.  No log data at debug=3.
> I changed it to 6.  There was just too much data for my mental parser to
> deal with.  I went to 4.  By this point I couldn't make it restart anymore.
> I have no idea what had changed.  Perhaps it will reoccur and I can get more
> data.  We're going to be starting testing soon with the mi_datagram module;
> perhaps if that causes the same issues I'll be able to get more data.
>   
OK.

Regards,
Bogdan
>
> - Jeff
>
>
>
> On 3/9/09 12:51 PM, "Bogdan-Andrei Iancu" <bogdan at voice-system.ro> wrote:
>
>   
>> Hi Jeff,
>>
>> The error message translates in receiving the ACK before finishing
>> processing the 200 OK....It is race I know about and I'm planing the fix it.
>>
>> In regards to "opensipsctl fifo dlg_list", by restart, you mean crash?
>> does the log says something about?
>>
>> Thanks and regards,
>> Bogdan
>>
>> Jeff Pyle wrote:
>>     
>>> Hi Bogdan,
>>>
>>> I updated from svn and the dialog profile assignments seem to be behaving
>>> now.
>>>
>>> However, I get this error:
>>>
>>> CRITICAL:dialog:log_next_state_dlg: bogus event 6 in state 2 for dlg
>>> 0xb614953c [3865:384697978] with clid
>>> '5f55f99253a9471f3672b30a49367e92 at ww.xx.yy.zz' and tags 'as4156018e' ''
>>>
>>> It shows up when the call goes to 200 OK on the second PSTN carrier, after
>>> failing on the first one, if that's relevant.
>>>
>>> The dialog still shows up in the database in the proper profile with the
>>> proper value.
>>>
>>> I don't know if this is related, but until just a moment ago anytime I ran
>>> "opensipsctl fifo dlg_list", Opensips would restart.  I moved debug to 6,
>>> then to 4, and then back to 3.  And it didn't happen anymore.  Odd.
>>>
>>>
>>> - Jeff
>>>
>>>
>>>
>>> On 3/9/09 10:29 AM, "Bogdan-Andrei Iancu" <bogdan at voice-system.ro> wrote:
>>>
>>>   
>>>       
>>>> Hi Jeff,
>>>>
>>>> Thank you for the report - there was a bug in the new code (when
>>>> create_dialog() was added), but now it is fixed on SVN.
>>>>
>>>> Please update and test again with your initial configuration (see the
>>>> email I sent to Brett).
>>>>
>>>> Regards,
>>>> Bogdan
>>>>
>>>> Jeff Pyle wrote:
>>>>     
>>>>         
>>>>> Hello,
>>>>>
>>>>> I'm configuring Opensips' dialog module to keep count of the number of
>>>>> calls
>>>>> I have on each outbound PSTN carrier.  Here's my thinking:
>>>>>
>>>>> In request route:
>>>>> - create_dialog() on new INVITE
>>>>> - select carrier
>>>>> - set_dlg_profile() to assign to proper profile with this carrier's value
>>>>>
>>>>> In failure route:
>>>>> - unset_dlg_profile() to remove from profile, since it failed
>>>>> - send to original request route to select next carrier, assign profile,
>>>>> etc
>>>>>
>>>>> When I run this, the first set_dlg_profile() works properly, but the
>>>>> unset_dlg_profile() in the failure route logs:
>>>>>
>>>>>  ERROR:dialog:unset_dlg_profile: dialog was not yet created - script error
>>>>>  ERROR:dialog:w_unset_dlg_profile: failed to unset profile
>>>>>
>>>>> All subsequent set/unset_dlg_profile give the same error.  Does the dialog
>>>>> somehow get destroyed when the failure_route is hit?  Is it necessary to
>>>>> create_dialog() each time the failure_route is hit by sending it back
>>>>> around
>>>>> to a request route?
>>>>>
>>>>>
>>>>> Thanks,
>>>>> Jeff
>>>>>
>>>>>
>>>>> _______________________________________________
>>>>> Users mailing list
>>>>> Users at lists.opensips.org
>>>>> http://lists.opensips.org/cgi-bin/mailman/listinfo/users
>>>>>
>>>>>   
>>>>>       
>>>>>           
>>>   
>>>       
>
>
>   




More information about the Users mailing list