[SR-Users] Via in locally generated PUBLISH (using different presence server address)

[Asgaroth]

Update:

I've just noticed that the error's I'm seeing roughly appear every 20 
seconds after an attempted locall generated publish is sent when 
rewriting the via header in the event-route.

This seems to corrolate with my db update interval which is set to 20 
seconds.

I'm wondering if the via header re-write was not written to the pua 
table correctly. I cannot run the msg_apply_changes() function in 
event-route so I'm stuck with this error message.

I wonder, would it be possible to have a module parameter to specify the 
sending socket if/when the outbound_proxy parameter is set? I presume, 
then, the Via header would be generated correctly.

Any thoughts on how I could work around this issue?

Thanks



On 09/08/2016 09:31, Asgaroth wrote:
> Hi Daniel,
>
> I've just sent the captures on to you privately.
>
> Thanks
>
> On 09/08/2016 08:34, Daniel-Constantin Mierla wrote:
>> Hello,
>>
>> can you send the pcap with sip messages for such situation? It will help
>> to understand if Via is used corerctly for replies.
>>
>> Cheers,
>> Daniel
>>
>>
>> On 08/08/16 18:51, Asgaroth wrote:
>>> Hi All,
>>>
>>> I have a configuration where we run a multi-homed proxy/loadbalancer
>>> and a seperate internal presence server. We then use
>>> pua/pua-dialoginfo on the proxies to publish dialog state to the
>>> internal presence server.
>>>
>>> The proxies listen on udp:external_ip:5062 and udp:internal_ip1:5060.
>>> The presence server listens on udp:internal_ip2:5060 on a seperate
>>> server.
>>>
>>> The problem I am coming accross is that when kamailio generates a
>>> PUBLISH message for some dialogs it creates the via header with
>>> address external:ip:5062, however the pua module is configured to send
>>> to the presence server address on internal_ip2:5060. I have a route
>>> which forces the sending socket ($fs) to be udp:internal_ip1:5060.
>>>
>>> What happens then is that the presence server tries to send the 200
>>> reply to the PUBLISH message back to internal_ip1:5062 (when it should
>>> be internal_ip1:5060), I'm thinking this is because the Via header has
>>> host as external_ip:5062.
>>>
>>> Is it possible, from event-route[tm:local-request] to rewrite the via
>>> header so that it reflects internal_ip1:5060 instead of 
>>> external_ip:5062?
>>>
>>> I did try the following snippet from event-route[tm:local-request],
>>> but I got module parse errors on the Via header:
>>>
>>> if (is_method("PUBLISH")) {
>>>        subst_hf("Via", "/ 1\.2\.3\.[0-9]+:5062;/
>>> $sel(cfg_get.socket.internal_ip);/", "f");
>>> }
>>>
>>> [7370]: ERROR: <core> [parser/parse_via.c:2607]: parse_via():
>>> ERROR:parse_via on <1> state 5 (default)
>>> [7370]: ERROR: <core> [parser/parse_via.c:2708]: parse_via(): ERROR:
>>> parse_via on: <SIP/2.0/UDP
>>> external_ip:5062;branch=z9hG4bK96eb.0f014ae4000000000000000000000000.0#015#012 
>>>
>>> internal_ip1;
>>> [7370]: ERROR: <core> [parser/parse_via.c:2712]: parse_via(): ERROR:
>>> parse_via parse error, parsed so far:<SIP/2.0/UDP
>>> external_ip:5062;branch=z9hG4bK96eb.0f014ae4000000000000000000000000.0#015#012 
>>>
>>> [7370]: ERROR: <core> [parser/msg_parser.c:125]: get_hdr_field():
>>> ERROR: get_hdr_field: bad via
>>> [7370]: INFO: <core> [parser/msg_parser.c:338]: parse_headers():
>>> ERROR: bad header field [Via: SIP/2.0/UDP 1]
>>> [7370]: ERROR: <core> [parser/msg_parser.c:690]: parse_msg(): ERROR:
>>> parse_msg: message=<PUBLISH sip:user at domain;transport=UDP
>>> SIP/2.0#015#012Via: SIP/2.0/UDP
>>> external_ip:5062;branch=z9hG4bK96eb.0f014ae4000000000000000000000000.0#015#012 
>>>
>>> internal_ip1;To: <sip:user at domain;transport=UDP>#015#012From:
>>> <sip:user at domain;transport=UDP>;tag=3e8fdaca277026bf3e6d75c5e5313096-f162#015#012CSeq: 
>>>
>>> 10 PUBLISH#015#012Call-ID:
>>> 5634294b786134b3-7370 at external_ip#015#012Content-Length:
>>> 692#015#012User-Agent: kamailio (bfievklb04)#015#012Max-Forwards:
>>> 70#015#012Event: dialog#015#012Expires: 10901#015#012Content-Type:
>>> application/dialog-info+xml#015#012#015#012<?xml
>>> version="1.0"?>#012<dialog-info
>>> xmlns="urn:ietf:params:xml:ns:dialog-info" version="0" state="full"
>>> entity="sip:user at domain;transport=UDP">#012  <dialog
>>> id="mvqXXaAr24DHdfsffYKZeA.." call-id="mvqXXaAr24DHdfsffYKZeA.."
>>> direction="initiator">#012 <state>Trying</state>#012    <remote>#012
>>> <identity>sip:1234567 at domain;transport=UDP</identity>#012 <target
>>> uri="sip:1234567 at domain;transport=UDP"/>#012 </remote>#012
>>> <local>#012 <identity>sip:user at domain;transport=UDP</identity>#012
>>> <target uri="sip:user at domain;transport=UDP"/>#012 </local>#012
>>> </dialog>#012</dialog-info>#012>
>>> [7370]: ERROR: <core> [msg_translator.c:3057]:
>>> build_sip_msg_from_buf(): parsing failed
>>> [7370]: ERROR: tm [uac.c:163]: uac_refresh_hdr_shortcuts(): failed to
>>> parse msg buffer
>>> [7370]: ERROR: tm [uac.c:487]: t_uac_prepare(): failed to refresh
>>> header shortcuts
>>> [7370]: ERROR: pua [send_publish.c:700]: send_publish(): in t_request
>>> tm module function
>>> [7370]: ERROR: pua_dialoginfo [dialog_publish.c:358]:
>>> dialog_publish(): while sending publish
>>>
>>> To me, it looks like the replace section of the subst_hf sub
>>> expression is being appended to the end of the via header instead of
>>> replacing the address section found.
>>>
>>> I have sanitized the above output to not show internal details, but if
>>> you need it, I can send it on to you privately.
>>>
>>> Am I going about this issue in the correct way, and, if so, where am I
>>> going wrong with the subst_hf snippet above?
>>>
>>> Kamailio version is 4.3.6.
>>>
>>> Any pointers/tips/trciks are greatly appreciated.
>>>
>>> Thanks
>>>
>>> _______________________________________________
>>> SIP Express Router (SER) and Kamailio (OpenSER) - sr-users mailing list
>>> sr-users at lists.sip-router.org
>>> http://lists.sip-router.org/cgi-bin/mailman/listinfo/sr-users
>