[sr-dev] handling of locally generated 478 errors

Henning Westerholt hw at skalatan.de
Wed Nov 25 18:52:59 CET 2020


Hello,

I want to ask for your opinion on the best approach regarding the handling of locally generated 478 errors.

To give an example, like the ones generated from TM during t_relay() on an unresolvable destination.

Nov 25 17:40:13 kamailio[19345]: ERROR: {28607414 INVITE bba500ac-a9df-1239-6693-00505682c04d} tm [ut.h:286]: uri2dst2(): failed to resolve "invalid.skalatan.de" :unresolvable A or AAAA request (-7)
Nov 25 17:40:13 kamailio[19345]: ERROR: {28607414 INVITE bba500ac-a9df-1239-6693-00505682c04d} tm [t_fwd.c:1738]: t_forward_nonack(): failure to add branches
Nov 25 17:40:13 kamailio[19345]: CRITICAL: {28607414 INVITE bba500ac-a9df-1239-6693-00505682c04d} rtpengine [../../core/parser/../ip_addr.h:658]: ip_addr2sbuf(): unknown address family 0

These errors will not show up in onreply or failure_route.  A long time ago this was discussed on the list [1], as some functionality were phased out out that support these scenarios.

Kamailio will try to generate a 478 with TM, this will obviously fail as well, and then generate a 478 with SL.

Question 1)

Is this intentional that the internally generated 478 is not showing up in the failure_route, like for for 408? This has been tested several times, but it is a complicated configuration.

Question 2)

Are there any other (better) ideas how to handle that besides using a "event_route[sl:local-response]" to catch this, e.g. to tear down otherwise stale rtpengine sessions etc..? As a side note, event_route[tm:local-response] seems not to work as well because of the tm failure.

Thanks,

Henning

[1] https://lists.kamailio.org/pipermail/sr-users/2011-June/069020.html

--
Henning Westerholt - https://skalatan.de/blog/
Kamailio services - https://gilawa.com<https://gilawa.com/>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.kamailio.org/pipermail/sr-dev/attachments/20201125/00a1af4c/attachment-0001.htm>


More information about the sr-dev mailing list