Hi Daniel,

Thanks for the suggestion.  I'm not modifying the R-URI in any way but tested with revert_uri() regardless.  The behavior is the same.  I'll open a ticket.  Thanks.


Cindy

On Wed, Aug 19, 2020 at 10:18 AM Daniel-Constantin Mierla <miconda@gmail.com> wrote:

Looking a bit at the node, I see that the r-uri branch is ignored if not changed at all. So if you updated r-uri in the config, then it will be kept. But the same behaviour seems to be in 5.3. Maybe in the new config you do operations over $ru or use some functions changing it.

You can try to do revert_uri() before t_load_contacts(0).

Cheers,
Daniel

On 19.08.20 16:12, Daniel-Constantin Mierla wrote:

Hello,

On 18.08.20 08:05, Cindy Leung wrote:
Hello all,

Just started using v5.4.0 on our system and I noticed a change in behavior when doing the append_branch, t_load_contacts, and t_next_contacts combo.  Previously using v5.3.4 and it appears to be fine.

Here's the call scenario: Kamailio receives a call to sip:1001@carrierB.  Kamailio sees carrierB and appends 2 contacts: gateway1.carrierB.com;q=0.3 and gateway2.carrierB.com;q=0.2.  After t_load_branches(0),

first, I am not finding the t_load_branches() function, is it supposed to be t_load_contacts()?

The, supposing it was the later, afaik, the load contacts was putting the branches in internal xavps, not pushing to the r-uri. The append_branch() was not changing the first branch (the r-uri) but adding extra. So if you have the invite coming in to u1@s.com and do append_branch(u2@s.com) and append_branch(u3@s.com), then it will be 3 branches that will go out in case of parallel forking.

With t_load_contacts() and t_next_contacts(), these 3 branches should be prepared to do serial forking.

Now, I haven't really used these functions myself to comment more, it is based on what append_branch() is doing: adding extra branches and keeping the first branch (r-uri) untouched.

But if you found a different behaviour than expected, open a bug on issue tracker and we can refer the developer of commit 1399714fbba63732f94eb8034dabb1e565ca832a (which added proportional load) to review it.

Cheers,
Daniel

I expect to see $rd to be changed to gateway1.carrierB.com, but it's not.

This is the piece of config I'm trying to debug:
        xlog ("=== branch 0: $(branch(uri)[0]), $(branch(q)[0])\n");
    xlog ("=== branch 1: $(branch(uri)[1]), $(branch(q)[1])\n");
    t_load_contacts(0);
    while (t_next_contacts()) {
        xlog ("=== rd = $rd\n");
    }

This is the log I get.  It appears to use the backup contact first.
ERROR: IBG_LOG: sipp-ci1-20200818053427-1-23@172.18.1.21: === branch 0: sip:1001@gateway2.carrierB.com;transport=udp, 200
ERROR: IBG_LOG: sipp-ci1-20200818053427-1-23@172.18.1.21: === branch 1: sip:1001@gateway1.carrierB.com;transport=tcp, 300
DEBUG: IBG_LOG: sipp-ci1-20200818053427-1-23@172.18.1.21: tm [t_serial.c:522]: t_load_contacts(): load_contact mode selected: 0
DEBUG: IBG_LOG: sipp-ci1-20200818053427-1-23@172.18.1.21: tm [t_serial.c:340]: ki_t_load_contacts_mode(): nr_branches is 2
DEBUG: IBG_LOG: sipp-ci1-20200818053427-1-23@172.18.1.21: <core> [core/xavp.c:539]: xavp_destroy_list(): destroying xavp list 0x7f983cb66608
DEBUG: IBG_LOG: sipp-ci1-20200818053427-1-23@172.18.1.21: tm [t_serial.c:890]: ki_t_next_contacts(): Appending branch uri-'sip:1001@gateway1.carrierB.com;transport=tcp' dst-'' path-'' inst-'' ruid-'' location_ua-''
DEBUG: IBG_LOG: sipp-ci1-20200818053427-1-23@172.18.1.21: <core> [core/xavp.c:539]: xavp_destroy_list(): destroying xavp list 0x7f983cb66350
ERROR: IBG_LOG: sipp-ci1-20200818053427-1-23@172.18.1.21: === rd = carrierB
DEBUG: IBG_LOG: sipp-ci1-20200818053427-1-23@172.18.1.21: <core> [core/xavp.c:539]: xavp_destroy_list(): destroying xavp list 0x7f983cb66078
ERROR: IBG_LOG: sipp-ci1-20200818053427-1-23@172.18.1.21: === rd = gateway2.carrierB.com
DEBUG: IBG_LOG: sipp-ci1-20200818053427-1-23@172.18.1.21: tm [t_serial.c:627]: ki_t_next_contacts(): no contacts in contacts_avp - we are done!

t_load_contacts(1) works a little better.  But we don't need the probability feature.
ERROR: IBG_LOG: sipp-ci1-20200818054736-1-21@172.18.1.21: === branch 0: sip:1001@gateway2.carrierB.com;transport=udp, 200
ERROR: IBG_LOG: sipp-ci1-20200818054736-1-21@172.18.1.21: === branch 1: sip:1001@gateway1.carrierB.com;transport=tcp, 300
DEBUG: IBG_LOG: sipp-ci1-20200818054736-1-21@172.18.1.21: tm [t_serial.c:522]: t_load_contacts(): load_contact mode selected: 1
DEBUG: IBG_LOG: sipp-ci1-20200818054736-1-21@172.18.1.21: tm [t_serial.c:340]: ki_t_load_contacts_mode(): nr_branches is 2
DEBUG: IBG_LOG: sipp-ci1-20200818054736-1-21@172.18.1.21: tm [t_serial.c:280]: t_load_contacts_proportional(): proportionally selected contact with uri: sip:1001@gateway1.carrierB.com;transport=tcp (q: 300, random: 287, q_index: 500, q_total: 500)
DEBUG: IBG_LOG: sipp-ci1-20200818054736-1-21@172.18.1.21: tm [t_serial.c:280]: t_load_contacts_proportional(): proportionally selected contact with uri: sip:1001@gateway2.carrierB.com;transport=udp (q: 200, random: 157, q_index: 200, q_total: 200)
DEBUG: IBG_LOG: sipp-ci1-20200818054736-1-21@172.18.1.21: tm [t_serial.c:303]: t_load_contacts_proportional(): proportionally added backup contact with uri: sip:1001@carrierB SIP/2.0#015#012Via: SIP/2.0/UDP 172.18.1.21:5060;branch=z9hG4bK-21-1-0 blah blah blah (q: -1)
DEBUG: IBG_LOG: sipp-ci1-20200818054736-1-21@172.18.1.21: <core> [core/xavp.c:539]: xavp_destroy_list(): destroying xavp list 0x7f6e457ec078
ERROR: IBG_LOG: sipp-ci1-20200818054736-1-21@172.18.1.21: === rd = gateway1.carrierB.com
DEBUG: IBG_LOG: sipp-ci1-20200818054736-1-21@172.18.1.21: <core> [core/xavp.c:539]: xavp_destroy_list(): destroying xavp list 0x7f6e457ec350
ERROR: IBG_LOG: sipp-ci1-20200818054736-1-21@172.18.1.21: === rd = gateway2.carrierB.com
DEBUG: IBG_LOG: sipp-ci1-20200818054736-1-21@172.18.1.21: <core> [core/xavp.c:539]: xavp_destroy_list(): destroying xavp list 0x7f6e457ec608
ERROR: IBG_LOG: sipp-ci1-20200818054736-1-21@172.18.1.21: === rd = carrierB
DEBUG: IBG_LOG: sipp-ci1-20200818054736-1-21@172.18.1.21: tm [t_serial.c:627]: ki_t_next_contacts(): no contacts in contacts_avp - we are done!

As a comparison, this is what we've been getting in 5.3.4
ERROR: IBG_LOG: sipp-ci1-20200818055736-1-21@172.18.1.21: === branch 0: sip:1001@gateway2.carrierB.com;transport=udp, 200
ERROR: IBG_LOG: sipp-ci1-20200818055736-1-21@172.18.1.21: === branch 1: sip:1001@gateway1.carrierB.com;transport=tcp, 300
DEBUG: IBG_LOG: sipp-ci1-20200818055736-1-21@172.18.1.21: tm [t_serial.c:191]: ki_t_load_contacts(): nr_branches is 2
DEBUG: IBG_LOG: sipp-ci1-20200818055736-1-21@172.18.1.21: <core> [core/xavp.c:529]: xavp_destroy_list(): destroying xavp list 0x7fe13146a680
ERROR: IBG_LOG: sipp-ci1-20200818055736-1-21@172.18.1.21: === rd = gateway1.carrierB.com
DEBUG: IBG_LOG: sipp-ci1-20200818055736-1-21@172.18.1.21: <core> [core/xavp.c:529]: xavp_destroy_list(): destroying xavp list 0x7fe13146a3a8
ERROR: IBG_LOG: sipp-ci1-20200818055736-1-21@172.18.1.21: === rd = gateway2.carrierB.com
DEBUG: IBG_LOG: sipp-ci1-20200818055736-1-21@172.18.1.21: <core> [core/xavp.c:529]: xavp_destroy_list(): destroying xavp list 0x7fe13146a0d0
ERROR: IBG_LOG: sipp-ci1-20200818055736-1-21@172.18.1.21: === rd = carrierB
DEBUG: IBG_LOG: sipp-ci1-20200818055736-1-21@172.18.1.21: tm [t_serial.c:460]: ki_t_next_contacts(): no contacts in contacts_avp - we are done!

Is there anything else that's been changed in branch building that we should pay attention to?  Thanks.



Cindy


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

-- 
Daniel-Constantin Mierla -- www.asipto.com
www.twitter.com/miconda -- www.linkedin.com/in/miconda
Funding: https://www.paypal.me/dcmierla
-- 
Daniel-Constantin Mierla -- www.asipto.com
www.twitter.com/miconda -- www.linkedin.com/in/miconda
Funding: https://www.paypal.me/dcmierla