On Sep 24, 2009 at 13:06, Alex Hermann alex@speakup.nl wrote:
On Thursday 24 September 2009, Andrei Pelinescu-Onciul wrote:
By default sip-router tm uses exactly what was sent on the last branch when doing dns failover. See http://sip-router.org/docbook/sip-router/branch/master/modules/tm/tm.html #reparse_on_dns_failover
The documented limitations make this seem like a hack. Why not make sure it works all the time?
We gladly accept patches :-) The case when it wouldn't work is a corner case (how many of you use mhomed=1 in the cfg?) and even in that case is unlikely that it would cause problem (I could of course come with a scenario where it will cause problems, but the likelihood of encountering it is quite low IMHO). Making it work in all the cases would also add a bigger performance hit.
So this should solve the header added in a branch route and not preserved on dns failover and you don't have to add anything to the script (provided you use sip-router or kamailio 3.0).
Well, it's a bit early to use that in production, isn't it?
Yes, using it now directly in production would be risky. This code in tm comes from ser 2.1 where it was tested (and is used in production). k 3.0 and sip-router as a whole, did see quite little testing until now, but hopefully they will see much more in the next month.
One thing to note though, is that branch route is not called when doing dns failover, if reparse_on_dns_failover is set to 1.
If set to 0, will it call branch route for every DNS-based branch? Even if it is the result of an internally generate 408?
Yes.
Andrei