sr-users January 2015

sr-users@lists.kamailio.org

94 participants
135 discussions

Load balancer - Path and Route header issue
by Mickael Marrache 18 Jan '15

18 Jan '15

Hi, I'm trying to use one Kamailio instance as load balancer for multiple Kamailio instances. The load balancer only performs NAT handling and dispatching of messages. The other instances handle registrations, routing, RTP proxy, authentication, etc. When a REGISTER request is received by the load balancer, a Path header is added to the request which is then forwarded to one of the registrars. The registration request is successfully handled and the Path header is saved in database. All subscribers register through the load balancer. When an INVITE is received from one subscriber by the load balancer, a Record-Route header is added by the load balancer which then forwards the INVITE to one of the other instances. The selected instance authenticates the subscriber, looks for the callee using the REGISTRAR module, adds a Record-Route header, adds a Route header according to the Path header associated with the callee, and forwards the INVITE to the load balancer. The load balancer performs loose routing (removing the Route header), adds another Record-Route header, and forwards the INVITE to the callee using the right transport address (indicated by the received parameter present in the Route header). The flow is right for the INVITE and its replies. However, when the caller sends the ACK, the route set is: load balancer, proxy, load balancer - without any information about how to reach the callee. Therefore, when the ACK reaches the "second" load balancer, the latter forwards it according to the RURI which is the Contact URI present in the 200 OK (which is not routable because of the NAT). I expected Kamailio to reuse the information that was present in the Path header to route the ACK correctly. What am I doing wrong? Thanks, Mickael

1 0

Kamailio with rtpproxy and asterisk in internal lan
by Willian Mazzardo - SYSSVOIP 17 Jan '15

17 Jan '15

Hi all. I want to create a Kamailio server who have dispatcher module, to loadbalance some asterisk servers who have my billing system. But, if I call from internet to my Public IP on Kamailio server, and then call goes to internal IP asterisk, who execute Echo App, I cant hear myself. If I call to my Public IP on Kamailio server, and call goes to EXTERNAL IP on Asterisk, everything works nice. I want this : Sip Client > Public IP Kamailio > Internal IP Kamailio > Internal IP Asterisk Server > Echo() I have rtpproxy running on same machine that Kamailio, and Im running with this command: rtpproxy -F -s udp:127.0.0.1:7722 -l PU.BL.IC.IP/172.16.1.1 -d DBUG:LOG_LOCAL0 My asterisk have 172.16.1.2 as internal IP. https://www.cloudshark.org/captures/15724d2d7e51 This is my tcpdump file captured. Thanks for any help. Willian Mazzardo Depto TI - SYSSVOIP www.syssvoip.com.br 55 3537 2030

1 0

Using the iptel Switch for both home & office environments...
by H. KAIMARAH 15 Jan '15

15 Jan '15

Hello there, Kindly advise me on how i can download and install the iptel soft switch for both Windows and Linux platforms. I intend to install a hosted Sip VoiP service using your open source switch for both home and office environments. I have jst been checking out the documentation from the site but seams some documentation pages are failing to load (timing out). Kindly advise or share with me some of the download links for the application so i can make a successful installation. Sincere regards, -- ***/Happy New Year/*** *HENRY K. Kaimarah* /Cell:/ +256-754462010 /Now in Kampala/ - UGANDA.

2 2

15 Jan '15

Hello all, We are moving our registration and location services to dedicated instances geographically separate from our proxies and then using redirects to locate the registered user. The idea is to protect the proxies from a potential thundering herd of registrations if there was some network outage I’m using dispatcher to find a registrar that has a registration for the user, if a user is not locally registered a 503 is returned and the next server is contacted. Registrations are replicated to other servers when the client is not behind NAT. If the client is behind NAT, the registrar where the client initially registered performs keep-alives and NAT traversal and also acts as an outbound proxy instead of sending a redirect. I’m attempting to simplify the configuration by using a dispatcher group to classify incoming registrations and to select replication targets but the registrar that initially received the request is in the group as well. Is there an elegant way to chose all destinations except the local instance? I’m currently doing this with the hack below but any suggestions would be greatly appreciated. Also, if anyone has any comments or recommendations on the topology, please share! Thanks, Spencer if (isflagset(FLT_FROMREG)) return; subst('/^From:(.*)sip:.*@[a-zA-Z0-9.:]+(.*)$/From:\1sip:$au@DOMAIN\2/ig'); subst('/^To:(.*)sip:.*@[a-zA-Z0-9.:]+(.*)$/To:\1sip:$au@DOMAIN\2/ig’); $rd = "DOMAIN"; $var(local_uri) = "sip:" + $Ri + ":" + $Rp; ds_select_dst("10", "4"); if ($du != $var(local_uri)) { t_replicate($du); } while (ds_next_dst()) { if ($du != $var(local_uri)) { t_replicate($du); } }

1 0

Changing SDP with subst after rtpproxy_answer corrupts SDP
by Matthias van der Vlies 15 Jan '15

15 Jan '15

Dear all, I have an on-reply route that needs to change the SDP version for the reply coming in. The use case is that I have a mobile originated call and there is some Ericsson switch that doesn't like it when the SDP version is updated (in this case by asterisk) although nothing has changed to the actual SDP (183 session progress and then OK.) Funny thing is that Asterisk will actually drop a call if it receives a re-INVITE with same version... That's why they invented ignoresdpversion, but now it's the other way around.... :) Mobile phone -> Ericsson MSC -> ACME packet -> (18X.4X.XXX.XX) Kamailio (10.41.0.21) -> Asterisk The issue is that the asterisk sends a reply 200 OK, with an updated version because it already sent SDP for 183 session progress. This can be patched in asterisk, but in my scenario I can unfortunately not do that. Thus trying to fix this on Kamailio. I am able to 'fix' this currently by performing a subst on the sdp owner variable: onreply_route[WITHSDP] { if (has_body("application/sdp")) { if(ds_is_from_list()) { rtpproxy_answer("wrei"); *if(subst("/^o=someowner ([0-9]+) ([0-9]+) IN IP4 (.*)$/o=someowner \1 \1 IN IP4 \3/")) {* xlog("L_INFO", "Fixed Asterisk incorrect version number in SDP"); } # tried the answer here as well, but that corrupts it even more } else { rtpproxy_answer("wrie"); } } exit; } However this corrupts the SDP: v=0 o=tismi 652858233 652858233 IN IP4 10.41.0.21*18X.4X.XXX.XX* s=Some server c=IN IP4 18X.4X.XXX.XX t=0 0 m=audio 57644 RTP/AVP 8 0 18 3 101 a=rtpmap:8 PCMA/8000 a=rtpmap:0 PCMU/8000 a=rtpmap:18 G729/8000 a=fmtp:18 annexb=no a=rtpmap:3 GSM/8000 a=rtpmap:101 telephone-event/8000 a=fmtp:101 0-16 a=silenceSupp:off - - - - a=ptime:20 a=sendrecv a=nortpproxy:yes When I do not substitute the SDP looks perfectly fine and the external address shows as the IN IP4. But of course the version is incremented: v=0 o=tismi 1606876535 *1606876536* IN IP4 *18X.4X.XXX.XX* s=Some server c=IN IP4 18X.4X.XXX.XX t=0 0 m=audio 55410 RTP/AVP 8 0 18 101 a=rtpmap:8 PCMA/8000 a=rtpmap:0 PCMU/8000 a=rtpmap:18 G729/8000 a=fmtp:18 annexb=no a=rtpmap:101 telephone-event/8000 a=fmtp:101 0-16 a=silenceSupp:off - - - - a=ptime:20 a=sendrecv a=sdpmangled:yes The ericsson is now accepting this (although it's corrupt, I know.... probably the ACME doing something funky with it), but it causes issues with another unknown piece of equipment that fails on parsing the session owner. I hope there is something wrong with my subst, but I'm afraid I can not do this from the on_reply route because SDP is only updated once it finishes? I know it's dangerous to alter the session version like this, so I made sure the Asterisk will never send a re-INVITE. Now I need a way to not corrupt the o= Kind regards, Matthias van der Vlies

3 3

websocket convert and forward to asterisk
by jaflong jaflong 15 Jan '15

15 Jan '15

Hi Everyone, Does anyone have a example of the config where I can get the following to work I want Kamailio to process websocket converting wss to tcp and srtp to rtp and forward to asterisk as tcp and rtp incoming call on websocket (wss rtp/savpf) --> kamailio/rtpproxy (protocol - convert wss to tcp) and (sdp - convert rtp/savpf to rtp/avp) and (audio - convert srtp to rtp) --> asterisk (recieve protocol: tcp, sdp: rtp/avp and audio:rtp) This application would only have incoming call from websocket to forward on to asterisk.

1 0

Fosdem 2015
by Alexandr Dubovikov 15 Jan '15

15 Jan '15

Hello everybody, what is about Fosdem this year ? Who is going there ? Wbr, Alexandr

2 1

Kamailio with a third party operator
by Jyaim Jyaim 15 Jan '15

15 Jan '15

Hello all, I cannot find documentation on how to configure Kamailio to interwork with another operator's IMS Core. I think the s-cscf should be in charge of detecting a user is not known locally, in order to route requests toward other operator. Do you know where I should look to get infos on this topic ? Thank you!

1 0

Redirect Server Including Path/Recieved Information
by Asgaroth 15 Jan '15

15 Jan '15

Hi All, I am attempting to setup a standalone redirect server which will lookup contact info and redirect to appropriate outbound proxy. The problem I am having is that the registrar is storing the path and recieved information, however, when I perform a lookup and reply with 302, the recieved and path information is not included. Are there module parameters in the registrar module that will include these contact parameters (recieved) or create the appropriate headers (path/route) in the 302 response, or, is this something I need to do manually. Currently I have the following lookup code: route { t_check_trans(); if ( method == "INVITE" ) { xlog("route[MAIN] : $rm : ruri=$ru"); xlog("route[MAIN] : $rm : lookup=sip:$rU@registered.domain"); if ( lookup("location", "sip:$rU@registered.domain") ) { send_reply("302", "Moved Temporarily"); exit; } } } I had a look at the path_mode parameter but this looks like it only takes affect for REGISTER methods. The INVITE that comes in supports the PATH header but I dont see this being passed back. In fact, I'm not entirely sure if the path header is a supported header in the 302 response message. I had a quick google and I cant seem to easily find what headers are suported in the 302 message, and where I need to put the path/recieved information. I presume I can add the recieved info as a parameter to the contact header, but how would i specify the outbound proxy to use, would that be in a route or a path header in the 302 response? I can see that the $du pseudo variable is set with the appropriate outbound path and received information, now I just need to include this information in the 302 response. Any suggestions/comments to assist in how I get this info into the 302 message would be greatly appreciated. Thanks in advance.

4 12

Crash in usrloc, get_all_db_ucontacts
by Kristian F. Høgh 14 Jan '15

14 Jan '15

Hi, On an ancient kamailio 3.1, we occasionally experience crash in the function get_all_db_ucontacts in usrloc. A backtrace reveals the following: #0 0x00002b505f2fb4ff in get_all_db_ucontacts (buf=0xadd270, len=<value optimized out>, flags=64, part_idx=1, part_max=30) at dlist.c:176 176 if (VAL_NULL(ROW_VALUES(row)+4) || p1==0 || p1[0]==0){ The lines around line 176 (which look similar to trunk, except p and p1 are str today) 159 for(i = 0; i < RES_ROW_N(res); i++) { 160 row = RES_ROWS(res) + i; 161 162 /* received */ 163 p = (char*)VAL_STRING(ROW_VALUES(row)); 164 if ( VAL_NULL(ROW_VALUES(row)) || p==0 || p[0]==0 ) { 165 /* contact */ 166 p = (char*)VAL_STRING(ROW_VALUES(row)+1); 167 if (VAL_NULL(ROW_VALUES(row)+1) || p==0 || p[0]==0) { 168 LM_ERR("empty contact -> skipping\n"); 169 continue; 170 } 171 } 172 p_len = strlen(p); 173 174 /* path */ 175 p1 = (char*)VAL_STRING(ROW_VALUES(row)+4); 176 if (VAL_NULL(ROW_VALUES(row)+4) || p1==0 || p1[0]==0){ 177 p1 = NULL; 178 p1_len = 0; 179 } else { 180 p1_len = strlen(p1); 181 } The client from frame #2 (bt full below), send a REGISTER, which was replied "200 OK" at the same second as the crash. Is it a timing issue between REGISTER and nathelper ping? I have a trace and the core. Regards, Kristian Høgh Uni-tel bt full #0 0x00002b505f2fb4ff in get_all_db_ucontacts (buf=0xadd270, len=<value optimized out>, flags=64, part_idx=1, part_max=30) at dlist.c:176 sock = 0x0 dbflags = 11088 host = {s = 0x375ef8e "PROXY_IP:5060", len = 13} i = <value optimized out> row = 0xa13488 dom = 0x2b5062723278 p1 = 0x2 <Address 0x2 out of bounds> now_s = "'2015-01-13 12:58:51'\177\000\000)" port = 11088 p_len = 73 cp = <value optimized out> needed = 16 res = 0xa2ed38 p = 0xcfcdffaf <Address 0xcfcdffaf out of bounds> now_len = 21 proto = 1642847400 p1_len = 0 shortage = 0 query_buf = "select received, contact, socket, cflags, path from location where expires > '2015-01-13 12:58:51' and cflags & 64 = 64 and id % 30 = 1", '\000' <repeats 376 times> query_str = {s = 0x2b505f515520 "select received, contact, socket, cflags, path from location where expires > '2015-01-13 12:58:51' and cflags & 64 = 64 and id % 30 = 1", len = 135} #1 get_all_ucontacts (buf=0xadd270, len=<value optimized out>, flags=64, part_idx=1, part_max=30) at dlist.c:398 No locals. #2 0x00002b5061eb6b1a in nh_timer (ticks=<value optimized out>, timer_idx=<value optimized out>) at nathelper.c:1623 rval = <value optimized out> buf = 0xadd270 cp = <value optimized out> c = {s = 0x0, len = 0} opt = {s = 0x0, len = 0} path = {s = 0x0, len = 0} curi = {user = {s = 0x0, len = 0}, passwd = {s = 0x0, len = 0}, host = {s = 0xadd836 "CLIENT_IP:62252\250u\220", len = 14}, port = {s = 0xadd845 "62252\250u\220", len = 5}, params = {s = 0x0, len = 0}, sip_params = {s = 0x0, len = 0}, headers = { s = 0x7fff110134b0 "", len = 566862356}, port_no = 50336, proto = 70, type = ERROR_URI_T, flags = 65536, transport = {s = 0x8cbfa0 "", len = 5441948}, ttl = {s = 0x7fff1101352f "", len = 4448297}, user_param = {s = 0x375ef8e "PROXY_IP:5060", len = 13}, maddr = {s = 0x9204b0 "sip3", len = 10677560}, method = {s = 0x0, len = 0}, lr = {s = 0xadd84a "\250u\220", len = 0}, r2 = {s = 0xa2ebc0 "\250\272\252", len = 285291296}, transport_val = {s = 0x40 <Address 0x40 out of bounds>, len = 335544320}, ttl_val = { s = 0x3221c302d0 "H\307\300\017", len = 0}, user_param_val = {s = 0x0, len = 0}, maddr_val = {s = 0x0, len = 0}, method_val = {s = 0x0, len = 0}, lr_val = {s = 0x0, len = 0}, r2_val = {s = 0x0, len = 0}} he = <value optimized out> send_sock = <value optimized out> flags = <value optimized out> proto = <value optimized out> dst = {send_sock = 0x0, to = {s = {sa_family = 0, sa_data = '\000' <repeats 13 times>}, sin = {sin_family = 0, sin_port = 0, sin_addr = {s_addr = 0}, sin_zero = "\000\000\000\000\000\000\000"}, sin6 = {sin6_family = 0, sin6_port = 0, sin6_flowinfo = 0, sin6_addr = {in6_u = {u6_addr8 = '\000' <repeats 15 times>, u6_addr16 = {0, 0, 0, 0, 0, 0, 0, 0}, u6_addr32 = {0, 0, 0, 0}}}, sin6_scope_id = 0}}, id = 0, proto = 0 '\000', send_flags = {f = 0 '\000', blst_imask = 0 '\000'}} iteration = 1 __FUNCTION__ = "nh_timer" #3 0x0000000000521c54 in fork_dummy_timer (child_id=<value optimized out>, desc=<value optimized out>, make_sock=<value optimized out>, f=0x2b5061eb6a70 <nh_timer>, param=0x0, interval=1) at timer_proc.c:71 pid = <value optimized out> #4 0x00002b5061eb42ee in child_init (rank=<value optimized out>) at nathelper.c:673 i = 58055190 #5 0x00000000004db1d0 in init_mod_child (m=0x917dc0, rank=0) at sr_module.c:829 No locals. #6 0x00000000004db1a9 in init_mod_child (m=0x918420, rank=0) at sr_module.c:826 No locals. #7 0x00000000004db1a9 in init_mod_child (m=0x918aa0, rank=0) at sr_module.c:826 No locals.

2 10

← Newer
1
...
7
8
9
10
11
12
13
14
Older →

Jump to page:

2025

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

2006

2005

2004

2003

2002

sr-users January 2015