13 Jan
2021
13 Jan
'21
9:24 a.m.
SIP transactions are decoupled from the transport layer, by specs, the
connections have to be reused for the same target ip/port.
Cheers,
Daniel
On 12.01.21 16:51, Charles Phillips wrote:
It is my understanding that for outbound connections,
subsequent
transactions to the same destination IP:port reuse an existing TLS
socket (if one exists) by design. By the logs, it appears that this
matching takes place early in the processing so there is no regard for
a new outbound transaction that has different SNI. Is this correct?
Is there a way to force a new outbound TLS connection for a new
transaction based on some other identifier?
- Charles Phillips
--
Daniel-Constantin Mierla -- www.asipto.com
www.twitter.com/miconda -- www.linkedin.com/in/miconda
Funding: https://www.paypal.me/dcmierla
On Jan 11, 2021, at 9:00 AM, Charles Phillips
<charles(a)rustybike.com
<mailto:charles@rustybike.com>> wrote:
That is what I figured was happening. I have tried sending it back
to a standard routing block, but perhaps I am doing it incorrectly.
When I try to send it back to a regular routing block I get the
following error:
CRITICAL: tm [tm.c:1754]: _w_t_relay_to(): unsupported route type: 64
Config:
event_route[tm:local-request] {
sip_trace();
if(is_method("OPTIONS") && $ru =~ "pstnhub.microsoft.com
<http://pstnhub.microsoft.com/>") {
$var(domain) = $fd;
append_hf("Contact: <sip:$var(domain):5061;transport=tls
<sip:$var(domain):5061;transport=tls>>\r\n");
xlog("L_INFO", “TEAMS Contact: $var(domain)\r\n");
route(TEAMS_SEND);
}
xlog("L_INFO", "Sent out tm request: $mb\n");
}
route[TEAMS_SEND] {
$var(domain) = $fd;
$xavp(tls=>server_name) = $var(domain);
$xavp(tls[0]=>server_id) = $var(domain);
$du = "sip:$var(domain):5061;transport=tls
<sip:$var(domain):5061;transport=tls>";
t_relay();
}
For testing, I also tried generating the packets in a normal route
using t_uac_send and controlling it with rtimer. As ugly a hack that
this approach is, it did manage to create the packets and set the
xavp as required (although, it certainly wouldn’t help dispatcher
know if a gateway is offline…). Additional trouble is that if a
second domain attempts to send OPTIONS packets in the while loop (see
below) it goes out the same TLS connection, so it is rejected.
Config:
route["PING-TEAMS"] {
sql_query("db", "select domain from
domain;", "domain_list");
$var(i) = 0;
while ($dbr(domain_list=>[$var(i),0]) != $null) {
$var(domain) = $dbr(domain_list=>[$var(i),0]);
xlog(“OPTIONS from domain name $var(domain)");
$xavp(tls=>server_name) = $var(domain);
$xavp(tls[0]=>server_id) = $var(domain);
$du = "sip:$var(domain):5061;transport=tls
<sip:$var(domain):5061;transport=tls>";
t_uac_send ("OPTIONS", "sip:$var(domain):5061;transport=tls
<sip:$var(domain):5061;transport=tls>", "sip:sip3.pstnhub.microsoft.com;transport=tls
<sip:sip3.pstnhub.microsoft.com;transport=tls>", "", "From:
sip:$var(domain) <sip:$var(domain)>\r\nTo:
sip:sip3.pstnhub.microsoft.com;transport=tls
<sip:sip3.pstnhub.microsoft.com;transport=tls>\r\nContact:
<sip:$var(domain):5061;transport=tls
<sip:$var(domain):5061;transport=tls>>\r\n", "");
sleep(2);
t_uac_send ("OPTIONS", "sip:$var(domain):5061;transport=tls
<sip:$var(domain):5061;transport=tls>", "sip:sip2.pstnhub.microsoft.com;transport=tls
<sip:sip2.pstnhub.microsoft.com;transport=tls>", "", "From:
sip:$var(domain) <sip:$var(domain)>\r\nTo:
sip:sip2.pstnhub.microsoft.com;transport=tls
<sip:sip2.pstnhub.microsoft.com;transport=tls>\r\nContact:
<sip:$var(domain):5061;transport=tls
<sip:$var(domain):5061;transport=tls>>\r\n", "");
sleep(2);
t_uac_send ("OPTIONS", "sip:$var(domain):5061;transport=tls
<sip:$var(domain):5061;transport=tls>", "sip:sip.pstnhub.microsoft.com;transport=tls
<sip:sip.pstnhub.microsoft.com;transport=tls>", "", "From:
sip:$var(domain) <sip:$var(domain)>\r\nTo:
sip:sip.pstnhub.microsoft.com;transport=tls
<sip:sip.pstnhub.microsoft.com;transport=tls>\r\nContact:
<sip:$var(domain):5061;transport=tls
<sip:$var(domain):5061;transport=tls>>\r\n", "");
sleep(5);
$var(i) = $var(i) + 1;
}
}
200 from MS on domain 0:
2021/01/11 08:48:31.291048 52.114.7.24:5061 -> *myip*:52606
SIP/2.0 200 OK
FROM: <sip:100.sbc.*mydomain*.net
<sip:100.sbc.*mydomain*.net>>;tag=3393f0703fb0ccaca74109ff37de39f5-5a503a0a
TO: <sip:sip3.pstnhub.microsoft.com <sip:sip3.pstnhub.microsoft.com>>
CSEQ: 10 OPTIONS
CALL-ID: 0f37a09f409a4e41-24410(a)127.0.0.1
<mailto:0f37a09f409a4e41-24410@127.0.0.1>
VIA: SIP/2.0/TLS
*myip*:5061;branch=z9hG4bK9306.42d92227000000000000000000000000.0
CONTENT-LENGTH: 0
ALLOW: INVITE,ACK,OPTIONS,CANCEL,BYE,NOTIFY
SERVER: Microsoft.PSTNHub.SIPProxy v.2020.12.28.1 i.ASEA.0
403 from MS on domain 1:
2021/01/11 08:49:39.755005 52.114.7.24:5061 -> *myip*:52606
SIP/2.0 403 Forbidden
FROM: <sip:101.sbc.*mydomain*.net
<sip:101.sbc.*mydomain*.net>>;tag=3393f0703fb0ccaca74109ff37de39f5-69555a28
TO: <sip:sip3.pstnhub.microsoft.com <sip:sip3.pstnhub.microsoft.com>>
CSEQ: 10 OPTIONS
CALL-ID: 0f37a09f409a4e44-24410(a)127.0.0.1
<mailto:0f37a09f409a4e44-24410@127.0.0.1>
VIA: SIP/2.0/TLS
*myip*:5061;branch=z9hG4bK4306.973e1562000000000000000000000000.0
REASON:
Q.850;cause=63;text="00cce062-66c6-45a4-8b5c-ccd48b71a9f6;Provided
Trunk FQDN '101.sbc.*mydomain*.net' is not allowed. Connection allows
ollowing fqdns: 100.sbc.*mydomain*.net, 100.sbc.*mydomain*.net."
CONTENT-LENGTH: 0
ALLOW: INVITE,ACK,OPTIONS,CANCEL,BYE,NOTIFY
SERVER: Microsoft.PSTNHub.SIPProxy v.2020.12.28.1 i.ASEA.0
- Charles Phillips
On Jan 11, 2021, at 4:13 AM, Daniel-Constantin
Mierla
<miconda(a)gmail.com <mailto:miconda@gmail.com>> wrote:
Hello,
the xavp_cfg set in the event_route is not propagated (kept) to the
moment when the message is sent out to tls. The
event_route[tm:local-request] is executed when the local request is
constructed, terminated before sending out, so whatever avp/xavp is
set in the event route are deleted when the block execution is
terminated.
It requires another solution here, I am thinking about what can be
done and will be added soon in the master branch.
Meanwhile, a workaround is to look traffic back to kamailio so the
routing happens over request_route block, where you can set the xavp.
Cheers,
Daniel
On 08.01.21 22:23, Charles Phillips wrote:
Thanks Daniel, I needed some certainty to get
unstuck. It appears
that the problem is actually related to the TLS config. I am using
multiple TLS configs, so it looks like the problem may be that the
server_name server_id are not being set, so the reply is returning
to the default TLS config. Not to mention that when I set the
testing domains certs under the default config, it works...
This is observed in the logs:
tls_get_connect_server_name(): xavp with outbound server name not found
tls_get_connect_server_id(): xavp with outbound server id not found
tls_complete_init(): Using initial TLS domain TLSc<default>
Is there a way to set this with xavp_cfg in the event_route? I
have read that section and tried may combinations, but since the
OPTIONs packets from the dispatcher module do not seem to traverse
the standard routes, I am not sure how to handle this.
Any advice would be greatly appreciated!
- Charles Phillips
> On Jan 8, 2021, at 12:56 PM, Daniel-Constantin Mierla
> <miconda(a)gmail.com <mailto:miconda@gmail.com>> wrote:
>
> Hello,
>
> there is an option that you can set to reuse the port for tcp/tls
> connections, but even so it is a best effort and it is not going
> to ensured -- all these are practically flags set to the sockets
> and the kernel (tcp stack) decides after all.
>
> Anyhow, the rport is mainly useful for connectionless
> communication, like UDP. For tcp/tls, the SIP specs demand to
> reuse the existing connections. As I did several Kamailio-MSTeams
> interconnectivity deployments, I can tell that the source port was
> never a problem. The TLS connection is kept open and MSTeams sends
> back traffic on it.
>
> Cheers,
> Daniel
>
> On 08.01.21 14:32, Charles Phillips wrote:
>> Thanks for the quick response Joel. Yes, I have read the article
>> and I have tested and confirmed that I am correctly appending the
>> contact header (I probably should have left that in the snippet
>> for clarity). Below is an example of Kamailio setting up the
>> connection. It is going out port 46245 this time, but it is random.
>>
>> 07:59:23.572319 IP *my.kamailio.server*.46245 >
>> *ms.teams.server*.sip-tls: Flags [P.], seq 1:518, ack 1, win 502,
>> length 517
>> 07:59:23.802458 IP *ms.teams.server*.sip-tls >
>> *my.kamailio.server*.46245: Flags [P.], seq 1:3767, ack 518, win
>> 2051, length 3766
>>
>> The TLS connection shows as successful in the logs.
>>
>>
>> - Charles
>>
>>
>> Date: Thu, 7 Jan 2021 19:12:10 -0800
>> From: Joel Serrano <joel(a)textplus.com <mailto:joel@textplus.com>>
>> To: "Kamailio (SER) - Users Mailing List"
>> <sr-users(a)lists.kamailio.org <mailto:sr-users@lists.kamailio.org>>
>> Subject: Re: [SR-Users] Source Port on TLS OPTIONS from Dispatcher
>> Message-ID:
>> <CAMtXxQnLtEyD=40cwKembxiyj3D778eK=+5JD7sL4CvYbYXF1g(a)mail.gmail.com
>>
<mailto:CAMtXxQnLtEyD=40cwKembxiyj3D778eK=+5JD7sL4CvYbYXF1g@mail.gmail.com>>
>> Content-Type: text/plain; charset="utf-8"
>>
>> Hi Charles,
>>
>> I don't think your issue is rport, make sure you are setting the
>> Contact
>> header correctly.
>>
>> Have you checked this blog post:
>> https://skalatan.de/en/blog/kamailio-sbc-teams
>> <https://skalatan.de/en/blog/kamailio-sbc-teams> ?
>>
>> There is a specific section that talks about how to tell Kamailio
>> to send
>> the OPTIONS like MS Teams wants them.
>>
>> Good luck,
>> Joel.
>>
>>
>>> On Jan 7, 2021, at 7:31 PM, Charles Phillips
>>> <charles(a)rustybike.com <mailto:charles@rustybike.com>> wrote:
>>>
>>> Hello all. As they say in radio, “long time listener, first
>>> time caller”
>>>
>>> Anyway, I am having trouble getting past the following road
>>> block and any help would be greatly appreciated.
>>>
>>> Kamailio version is 5.4.3
>>>
>>> When attempting to use dispatcher to send OPTIONS packets to
>>> several TLS destinations, the packets are leaving the Kamailio
>>> server on random ports. This is a problem because the servers I
>>> am sending the OPTIONS to (MS Teams) are enforcing rport so the
>>> responses are returned to a port that Kamailio is not listening
>>> on. I have tried to force the socket in the event route
>>> (relevant parts of snippet below) but it does not appear to
>>> help. I should also mention that I am not behind NAT and the
>>> TLS socket is specified in the dispatcher attrs.
>>>
>>> event_route[tm:local-request] {
>>> sip_trace();
>>>
>>>
>>> $fs = “tls:**ip-address**:5061”;
>>>
>>>
>>> }
>>>
>>> I have used Kamailio as a TLS server for many projects, but this
>>> is my first time as a client. I am sure I am missing something.
>>>
>>> - Charles
>>>
>>>
>>>
>>>
>>
>>
>> _______________________________________________
>> Kamailio (SER) - Users Mailing List
>> sr-users(a)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