I would assume the phone sends multiple REGISTER requests with same CallID, one or more of which has an expire=0, as a NAT traversal technique trying to discover its public IP at first. May be it doesn't do it very well.

Since you are using Kamailio for auth/usrloc, sending its REGISTER with expire=0 would indicate that the Kamailio contact has expired (probably after such a request was received from the phone). Try checking it with 'kamctl ul show'

On Tue, Nov 6, 2018 at 4:57 PM Henning Westerholt <hw@kamailio.org> wrote:
Am Dienstag, 6. November 2018, 22:50:54 CET schrieb Kevin Olbrich:
> Am Di., 6. Nov. 2018 um 22:40 Uhr schrieb Sergiu Pojoga <pojogas@gmail.com>:
> > It's not clear what kamailio/asterisk integration method you are using.
> > Looking at the 2 provided messages - the 2nd one is not a relay of the 1st
> > one.
>
> I might have matched the wrong transaction. I use HEP/HOMER to observe
> communication and Kamailio starts a new flow (=Call-Id) to asterisk (this
> message is no coming from the phone).
>
> > handle authentication/usrloc in Kamailio?
> > or
> > using PATH extension?
>
> I do auth + usrloc in Kamailio, no PATH.
>
> Maybe the Kamailio debug would lead me to the problem but verbose level 3
> has too much info.

Hi Kevin,

you could control the time when the specific not-working phone send a
REGISTER. Then you can enable the debugging for a few seconds during this
time, and then deactivate it again. This should work even on a production
server. Debugging it on a test server is of course an even better way.

Best regards,

Henning

--
Henning Westerholt - https://skalatan.de/blog/
Kamailio services - https://skalatan.de/services
Kamailio security assessment - https://skalatan.de/de/assessment

_______________________________________________
Kamailio (SER) - Users Mailing List
sr-users@lists.kamailio.org
https://lists.kamailio.org/cgi-bin/mailman/listinfo/sr-users