[OpenSIPS-Users] DNS SRV failover not working for B2BUA (1.10)

[Jeff Pyle]

I have same the SRV failover problem with uac_registrant-generated REGISTER
requests.

After little bit of digging it looks like there is no proxy structure
associated with internally generated requests, and no DNS-based failover
without it.

Issue #140 <https://github.com/OpenSIPS/opensips/issues/140> logged.


- Jeff


On Tue, Nov 26, 2013 at 9:23 AM, Jeff Pyle <jpyle at fidelityvoice.com> wrote:

> Hello,
>
> I have an SRV name as follows:
>
> $ dig +short -tSRV _sip._udp.proxyname.domain.com
> 1 10 5060 host1.domain.com.
> 2 10 5060 host2.domain.com.
>
> If I t_relay() a request with proxyname.domain.com as the request domain,
> and host1 does not answer within fr_timer seconds, the request is
> re-relayed to host2.  This is good.
>
> If I b2b_init_request("top hiding") instead of t_relay(), the 408 is
> forwarded back towards the UAC after host1 doesn't answer.  The request is
> never re-relayed to host2.
>
> There is a difference in the debug output after the 408 is generated for
> host1.  With the B2BUA it looks like is_3263_failure() never happens.
>
> t_relay:
> Nov 26 09:03:38 [16296] DBG:tm:timer_routine: timer
> routine:0,tl=0x7fbe6c7c0060 next=(nil), timeout=61
> Nov 26 09:03:38 [16296] DBG:tm:final_response_handler: Cancel sent out,
> sending 408 (0x7fbe6c7bfe10)
> Nov 26 09:03:38 [16296] DBG:tm:t_should_relay_response: T_code=100,
> new_code=408
> Nov 26 09:03:38 [16296] DBG:tm:t_pick_branch: picked branch 0, code 408
> (prio=800)
> Nov 26 09:03:38 [16296] DBG:tm:is_3263_failure: dns-failover test:
> branch=0, last_recv=408, flags=1
> Nov 26 09:03:38 [16296] DBG:tm:t_should_relay_response: trying DNS-based
> failover
> Nov 26 09:03:38 [16296] DBG:tm:do_dns_failover: new destination available
> Nov 26 09:03:38 [16296] DBG:core:check_ip_address: params 127.0.0.1,
> 127.0.0.1, 0
> Nov 26 09:03:38 [16296] DBG:tm:set_timer: relative timeout is 500000
>
> b2b_init_request:
> Nov 26 09:12:21 [16401] DBG:tm:timer_routine: timer
> routine:0,tl=0x7f09ddcbf7e8 next=(nil), timeout=11
> Nov 26 09:12:21 [16401] DBG:tm:final_response_handler: Cancel sent out,
> sending 408 (0x7f09ddcbf598)
> Nov 26 09:12:21 [16401] DBG:tm:t_should_relay_response: T_code=0,
> new_code=408
> Nov 26 09:12:21 [16401] DBG:tm:t_pick_branch: picked branch 0, code 408
> (prio=800)
> Nov 26 09:12:21 [16401] DBG:tm:local_reply: branch=0, save=0, winner=0
> Nov 26 09:12:21 [16401] DBG:tm:local_reply: local transaction completed
> Nov 26 09:12:21 [16401] DBG:tm:run_trans_callbacks: trans=0x7f09ddcbf598,
> callback type 256, id 0 entered
> Nov 26 09:12:21 [16401] DBG:b2b_entities:b2b_tm_cback: tm [0x7f09ddcbf598]
> notification cb for [408] reply
> Nov 26 09:12:21 [16401] DBG:b2b_entities:b2b_parse_key: hash_index = [472]
>  - local_index= [2611231]
> Nov 26 09:12:21 [16401] DBG:b2b_entities:b2b_tm_cback: Received reply
> [408] for dialog [0x7f09ddcbf2b8], method [INVITE]
>
> Is there something extra that must occur in the script to get DNS failover
> behavior with the B2BUA?  Or, is this a bug?
>
>
> Regards,
> Jeff
>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.opensips.org/pipermail/users/attachments/20131126/e2d04642/attachment-0001.htm>