[SR-Users] pua_usrloc, PUBLISH not send when contact is deleted or expires

[Daniel-Constantin Mierla]

Hi Giacomo,

On 1/19/12 11:49 AM, Giacomo Vacca wrote:
>
> Ah thanks Daniel,
>
> So I had a wrong expectation on the behaviour of pua_usrloc.
>
> It does make sense that the PUBLISH already carries its expiry and so 
> the expiration moment is implicit and should/can be handled elsewhere.
>
> I think what confused me were two things: the pua_usrloc documentation 
> and the fact that pua_usrloc does register for EXPIRE callbacks with 
> usrloc.
>
> I'm referring to:
>
> "The pua_usrloc module is the connector between the usrloc and pua 
> modules. It creates the environment to send PUBLISH requests for user 
> location records, on specific events (e.g., when new record is added 
> in usrloc, a PUBLISH with status open (online) is issued; when 
> expires, it sends closed (offline))."
>

perhaps the documentation is misleading a bit with its phrasing. It send 
offline (closed) when it is an un-register, before the expiration. I 
haven't looked in the code, but based on log messages and correlated 
with the expired carried by PUBLISH, the behavior is correct over all.

There is another aspect, usrloc cleans up expired records on timer, so 
usrloc callback can be executed later than actual expiration. A publish 
at that time could be late. Of course, presence does cleanup on timer as 
well, but it may be more accurate to handle expiration there. It is like 
when an UA with presence support disappears from networks, without 
sending any PUBLISH updates.

Cheers,
DAniel



> *From:*Daniel-Constantin Mierla [mailto:miconda at gmail.com]
> *Sent:* 19 January 2012 10:37
> *To:* SIP Router - Kamailio (OpenSER) and SIP Express Router (SER) - 
> Users Mailing List
> *Cc:* Giacomo Vacca
> *Subject:* Re: [SR-Users] pua_usrloc, PUBLISH not send when contact is 
> deleted or expires
>
> Hello,
>
> On 1/19/12 11:22 AM, Giacomo Vacca wrote:
>
> Hi Daniel,
>
> Thanks for your feedback. Yes, using mysql and enabling presence made 
> it work for un-registrations too.
>
> Just for the purpose of this scenario I'm sending the PUBLISH back to 
> the same entity, and it's being handled as a presentity.
>
> I still have a missing piece though (which is in fact the reason for 
> this configuration): having PUBLISH requests triggered by the 
> expiration of a contact.
>
> When a contact expires, pua_usrloc is now logging:
>
> Jan 19 10:12:03 debian6 /usr/sbin/kamailio[4359]: INFO: pua_usrloc 
> [ul_publish.c:211]: should not send ul publish
>
> (please see below for traces and details).
>
> Since I'm marking a registration with pua_set_publish(), why is that 
> flag not seen as true when usrloc fires an EXPIRE callback?
>
> I've tried marking every REGISTER request with pua_set_publish(), and 
> not just the first one, before calling save("location") but the result 
> was the same.
>
>
> the PUBLISH for registration expiration does not make sense, as 
> PUBLISH has also an expire interval, so the record in the presence 
> server should expire at the same time. A publish with update state to 
> close will eventually hit the presence server when there is no record 
> for it. Presence server will send notifications when the record in its 
> table expires.
>
> Cheers,
> Daniel
>
>


-- 
Daniel-Constantin Mierla -- http://www.asipto.com
http://linkedin.com/in/miconda -- http://twitter.com/miconda

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.sip-router.org/pipermail/sr-users/attachments/20120120/780c0b19/attachment.htm>