Hi Daniel-Constantin!
On 17.11.11 10:56, Daniel-Constantin Mierla wrote:
> Hello,
>
> On 11/16/11 9:15 AM, Dmitry Petrakoff wrote:
>> Hi Daniel-Constantin!
>>
>> Many thanks for your reply
>> Sorry for the wrong output from DEBUG.
>> I took it when such directives were set in config:
>>
>> #disable_tcp=yes
>> udp_mtu=1300
>> udp_mtu_try_proto=TCP
>>
>> Actually, that host 10.198.3.100 works just fine via UDP.
>>
>> But when those directives were enabled, next INVITE has gone there via
>> TCP and the error was the same as with Lync.
> yes, the question was if the host is able to take tcp since the
> conversion from tcp to udp is done by the sip server automatically
> when the size of the message exceeds udp mtu.
>
When both these udp_mtu and udp_mtu_try_proto are set, when server
receives INVITE from UAC via UDP with size more then set in udp_mtu,
then he tries to forward this INVITE to other party using TCP. But this
attempt generates the same error when sending as you may see in my
previous posts.
>From the other side, if server getting INVITE via TCP and forwarding it
via UDP, everything works good.
I.e. server listens on TCP and can to respond for incoming TCP requests.
But can't create a socket ( under FreeBSD 8 at least ) to initiate TCP
session.
Under tcpdump I see fisrt SYN from kamailio to destination; destination
reply with SYNACK but kamailio sends RESET instead of ACK:
09:44:41.869687 IP 10.200.245.250.10274 > 10.200.253.186.5068: Flags
[S], seq 3232460831, win 65535, options [mss 1460,nop,wscale 3,sackOK,TS
val 519365170 ecr 0], length 0
09:44:41.870304 IP 10.200.253.186.5068 > 10.200.245.250.10274: Flags
[S.], seq 1977596128, ack 3232460832, win 8192, options [mss
1460,nop,wscale 8,sackOK,TS val 7346469 ecr 519365170], length 0
09:44:41.870364 IP 10.200.245.250.10274 > 10.200.253.186.5068: Flags
[R], seq 3232460832, win 0, length 0
> If you have another kamailio listening on tcp, are you able to send
> traffic to it over tcp?
>
Yes we've set up quickly Kamalio-3.1.5 under Ubuntu ( Linux ubu-kam
2.6.32-33-generic-pae #72-Ubuntu SMP Fri Jul 29 22:06:29 UTC 2011 i686
GNU/Linux Ubuntu 10.04.3 LTS ) with just dispatcher's pinging
functionality for testing and it works.
Here is DEBUG output:
9(1415) DEBUG: <core> [tcp_main.c:1824]: tcp_send: no open tcp
connection found, opening new one
9(1415) DEBUG: <core> [ip_addr.c:247]: tcpconn_new: new tcp connection:
10.200.253.186
9(1415) DEBUG: <core> [tcp_main.c:1081]: tcpconn_new: on port 5068, type 2
9(1415) DEBUG: <core> [tcp_main.c:1382]: tcpconn_add: hashes: 3509:0:0, 2
9(1415) DEBUG: <core> [tcp_main.c:2700]: pending write on new
connection 0xb5ac3f5c (-1/305 bytes written)
17(1423) DEBUG: <core> [tcp_main.c:3583]: handle_ser_child: read
response= b5ac3f5c, 5, fd 29 from 9 (1415)
17(1423) DEBUG: <core> [io_wait.h:390]: DBG: io_watch_add(0x82430c0, 29,
2, 0xb5ac3f5c), fd_no=22
17(1423) DEBUG: <core> [io_wait.h:796]: DBG: io_watch_chg (0x82430c0,
29, 0x1, 0xffffffff) fd_no=23 called
17(1423) DEBUG: <core> [io_wait.h:617]: DBG: io_watch_del (0x82430c0,
29, -1, 0x0) fd_no=23 called
17(1423) DEBUG: <core> [tcp_main.c:4204]: tcp: DBG: sending to child,
events 2001
17(1423) DEBUG: <core> [tcp_main.c:3889]: send2child: to tcp child 1
14(1420), 0xb5ac3f5c
And here is tcpdump:
14:44:10.534273 IP 10.200.251.6.53280 > 10.200.253.186.5068: Flags [S],
seq 2844516733, win 5840, options [mss 1460,sackOK,TS val 600506 ecr
0,nop,wscale 6], length 0
14:44:10.534736 IP 10.200.253.186.5068 > 10.200.251.6.53280: Flags [S.],
seq 100170775, ack 2844516734, win 8192, options [mss 1460,nop,wscale
8,sackOK,TS val 356125 ecr 600506], length 0
14:44:10.534793 IP 10.200.251.6.53280 > 10.200.253.186.5068: Flags [P.],
seq 1:308, ack 1, win 92, options [nop,nop,TS val 600506 ecr 356125],
length 307
14:44:10.536831 IP 10.200.253.186.5068 > 10.200.251.6.53280: Flags [F.],
seq 1, ack 308, win 260, options [nop,nop,TS val 356125 ecr 600506],
length 0
14:44:10.537004 IP 10.200.251.6.53280 > 10.200.253.186.5068: Flags [F.],
seq 308, ack 2, win 92, options [nop,nop,TS val 600507 ecr 356125], length 0
14:44:10.537427 IP 10.200.253.186.5068 > 10.200.251.6.53280: Flags [.],
ack 309, win 260, options [nop,nop,TS val 356126 ecr 600507], length 0
I.e. TCP sending under Linux works just fine.
As you can see, the error happens right after string where
[tcp_main.c:1382]. Under FreeBSD we've got an error, but under Linux - not.
Unfortunately, I was not able to check this under OpenBSD-5.0 because
can't compile kamailio at all.
But that was a different story.
Thank you very much!
I'm ready to provide any additional info you will need.
WBR
Dimon
> Cheers,
> Daniel
>
>>
>> More over. Today I run dispatcher module as "stupid" pinger. And again,
>> faced with the same error.
>>
>> Now it is output from kamctl ps:
>>
>> Process:: ID=0 PID=40677 Type=attendant
>> Process:: ID=1 PID=40678 Type=udp receiver child=0 sock=1.2.3.4:5060
>> Process:: ID=2 PID=40679 Type=udp receiver child=1 sock=1.2.3.4:5060
>> Process:: ID=3 PID=40680 Type=udp receiver child=2 sock=1.2.3.4:5060
>> Process:: ID=4 PID=40681 Type=udp receiver child=3 sock=1.2.3.4:5060
>> Process:: ID=5 PID=40682 Type=udp receiver child=0 sock=127.0.0.1:5060
>> Process:: ID=6 PID=40683 Type=udp receiver child=1 sock=127.0.0.1:5060
>> Process:: ID=7 PID=40684 Type=udp receiver child=2 sock=127.0.0.1:5060
>> Process:: ID=8 PID=40685 Type=udp receiver child=3 sock=127.0.0.1:5060
>> Process:: ID=9 PID=40686 Type=udp receiver child=0
>> sock=10.200.245.250:5060
>> Process:: ID=10 PID=40687 Type=udp receiver child=1
>> sock=10.200.245.250:5060
>> Process:: ID=11 PID=40688 Type=udp receiver child=2
>> sock=10.200.245.250:5060
>> Process:: ID=12 PID=40689 Type=udp receiver child=3
>> sock=10.200.245.250:5060
>> Process:: ID=13 PID=40690 Type=slow timer
>> Process:: ID=14 PID=40691 Type=timer
>> Process:: ID=15 PID=40692 Type=MI FIFO
>> Process:: ID=16 PID=40693 Type=ctl handler
>> Process:: ID=17 PID=40694 Type=TIMER NH
>> Process:: ID=18 PID=40695 Type=tcp receiver child=0
>> Process:: ID=19 PID=40696 Type=tcp receiver child=1
>> Process:: ID=20 PID=40697 Type=tcp receiver child=2
>> Process:: ID=21 PID=40698 Type=tcp receiver child=3
>> Process:: ID=22 PID=40699 Type=tcp main process
>>
>> As you can see, TCP is present and working for receiving. And we were
>> able to receive INVITE from Lync via TCP.
>>
>>
>> Now, part of newly added dispatcher functionality:
>>
>> loadmodule "dispatcher.so"
>> modparam("dispatcher", "db_url", DBURL);
>> modparam("dispatcher", "ds_ping_from", "sip:openser@10.200.245.250");
>> modparam("dispatcher", "ds_ping_interval", 5);
>> modparam("dispatcher", "ds_probing_mode", 1);
>>
>> and table:
>>
>> mysql> select * from dispatcher;
>> +----+-------+---------------------------------------+-------+----------+-------+-------------+
>>
>> | id | setid | destination | flags | priority
>> | attrs | description |
>> +----+-------+---------------------------------------+-------+----------+-------+-------------+
>>
>> | 1 | 1 | sip:10.200.253.186:5068;transport=tcp | 0 | 0
>> | | Lync-Test |
>> +----+-------+---------------------------------------+-------+----------+-------+-------------+
>>
>>
>>
>>
>> corresponding tcp options are:
>>
>> #disable_tcp=yes
>> tcp_children=4
>>
>> and another, probably important, option which I forgot to inform about:
>>
>> mhomed=1
>>
>> And what I'm getting.
>>
>> From syslog
>>
>> Nov 16 11:07:04 fbsd-kam /usr/local/sbin/kamailio[60981]: ERROR:<core>
>> [tcp_mai
>> n.c:2748]: connect 10.200.253.186:5068 failed Socket is not connected
>> Nov 16 11:07:04 fbsd-kam /usr/local/sbin/kamailio[60981]: ERROR:<core>
>> [tcp_mai
>> n.c:2754]: 10.200.253.186:5068: connect& send for 0x28db87f8 failed:
>> Socket is
>> not connected (57)
>> Nov 16 11:07:04 fbsd-kam /usr/local/sbin/kamailio[60981]: ERROR: tm
>> [../../forwa
>> rd.h:170]: msg_send: ERROR: tcp_send failed
>> Nov 16 11:07:04 fbsd-kam /usr/local/sbin/kamailio[60981]: ERROR: tm
>> [uac.c:496]:
>> t_uac: Attempt to send to precreated request failed
>>
>> When run with DEBUG
>>
>> 13(67161) DEBUG: dispatcher [dispatch.c:2305]: probing set #1, URI
>> sip:10.200.253.186:5068;transport=tcp
>> 13(67161) DEBUG: tm [uac.c:242]: DEBUG:tm:t_uac:
>> next_hop=<sip:10.200.253.186:5068;transport=tcp>
>> 13(67161) DEBUG: tm [uac.c:181]: DEBUG: dlg2hash: 62915
>> 13(67161) DEBUG:<core> [tcp_main.c:1824]: tcp_send: no open tcp
>> connection found, opening new one
>> 13(67161) DEBUG:<core> [ip_addr.c:247]: tcpconn_new: new tcp
>> connection: 10.200.253.186
>> 13(67161) DEBUG:<core> [tcp_main.c:1081]: tcpconn_new: on port 5068,
>> type 2
>> 13(67161) DEBUG:<core> [tcp_main.c:1382]: tcpconn_add: hashes:
>> 3509:0:0, 4
>> 13(67161) ERROR:<core> [tcp_main.c:2748]: connect 10.200.253.186:5068
>> failed Socket is not connected
>> 13(67161) ERROR:<core> [tcp_main.c:2754]: 10.200.253.186:5068: connect
>> & send for 0x28dbc308 failed: Socket is not connected (57)
>> 13(67161) DEBUG:<core> [tcp_main.c:2955]: tcpconn_chld_put: destroying
>> connection 0x28dbc308 (4, -1) flags 0060
>> 13(67161) ERROR: tm [../../forward.h:170]: msg_send: ERROR: tcp_send
>> failed
>>
>>
>> Other options from config:
>>
>> fork=yes
>> children=4
>> user="kamailio"
>> group="kamailio"
>> server_header="DPRS's SIP proxy"
>> server_signature=no
>>
>> So as you can see, here is the same error.
>>
>> Many thanks in advance!
>>
>> WBR
>> Dimon
>>
>> On 16.11.11 11:46, Daniel-Constantin Mierla wrote:
>>> Hello,
>>>
>>> you haven't disabled tcp support, right? Can you do 'kamctl ps' and
>>> see if you have tcp workers?
>>>
>>> I see the error is related to 10.198.3.100:5060 :
>>>
>>> 10(36629) ERROR:<core> [tcp_main.c:2748]: connect 10.198.3.100:5060
>>> failed Socket is not connected
>>>
>>> What is listening on this ip and port? In drouting table you have
>>> 10.198.3.100:5068;transport=tcp -- a different port.
>>>
>>> Then, do you need to enable and play with mtu params? This is for
>>> auto-switching from UDP to TCP when size of SIP message exceeds
>>> udp_mtu value in bytes, done by SIP server, but if the other party is
>>> not listening on tcp, then it won't work (reason is disabled by
>>> default) -- note that this switching will be done for all sip messages.
>>>
>>> Cheers,
>>> Daniel
>>>
>>>
>>> On 11/15/11 11:43 AM, Dmitry Petrakoff wrote:
>>>> Hi List!
>>>>
>>>> I'm running Kamailio for about 1 year without any serious problems.
>>>> When it works with UDP only, it runs perfectly well for us. Without
>>>> any
>>>> major issues.
>>>>
>>>> But now we have a customer who wants to connect Lync to our network.
>>>> Unfortunately, Lync does not support UDP for SIP signalling, so we
>>>> tried
>>>> to teach kamailio to send SIP messages via TCP.
>>>> And getting strange error:
>>>>
>>>> 10(36629) DEBUG: tm [t_reply.c:659]: DEBUG: reply sent out.
>>>> buf=0x832c7ac: SIP/2.0 100 trying -..., shmem=0x28d97c7c: SIP/2.0 100
>>>> trying -
>>>> 10(36629) DEBUG: tm [t_reply.c:669]: DEBUG: _reply_light: finished
>>>> 10(36629) DEBUG:<core> [forward.c:213]: DEBUG: get_out_socket:
>>>> socket
>>>> determined: 0x833ddc0
>>>> 10(36629) DEBUG:<core> [msg_translator.c:457]: clen_builder:
>>>> content-length: 553 (553)
>>>> 10(36629) DEBUG:<core> [msg_translator.c:204]:
>>>> check_via_address(10.200.245.200, 10.200.245.200, 0)
>>>> 10(36629) DEBUG:<core> [tcp_main.c:1824]: tcp_send: no open tcp
>>>> connection found, opening new one
>>>> 10(36629) DEBUG:<core> [ip_addr.c:247]: tcpconn_new: new tcp
>>>> connection: 10.198.3.100
>>>> 10(36629) DEBUG:<core> [tcp_main.c:1081]: tcpconn_new: on port 5060,
>>>> type 2
>>>> 10(36629) DEBUG:<core> [tcp_main.c:1382]: tcpconn_add: hashes:
>>>> 0:0:0, 1
>>>> 10(36629) ERROR:<core> [tcp_main.c:2748]: connect 10.198.3.100:5060
>>>> failed Socket is not connected
>>>> 10(36629) ERROR:<core> [tcp_main.c:2754]: 10.198.3.100:5060:
>>>> connect&
>>>> send for 0x28d988ac failed: Socket is not connected (57)
>>>> 10(36629) DEBUG:<core> [tcp_main.c:2955]: tcpconn_chld_put:
>>>> destroying
>>>> connection 0x28d988ac (1, -1) flags 0060
>>>> 10(36629) ERROR: tm [../../forward.h:170]: msg_send: ERROR: tcp_send
>>>> failed
>>>>
>>>> We tested two different variants: one when defined Lync in dr_gateways
>>>> table as 10.198.3.100:5068;transport=tcp
>>>>
>>>> And then, tried global options
>>>> udp_mtu=1300
>>>> udp_mtu_try_proto=TCP
>>>>
>>>> As soon as these options were enabled, all other gateways became
>>>> unreachable with the same error as above.
>>>>
>>>> Could you please to point me out, where I'm wrong?
>>>>
>>>> fbsd-kam# /usr/local/sbin/kamailio -V
>>>> version: kamailio 3.1.5 (i386/freebsd) ec672a
>>>> flags: STATS: Off, USE_IPV6, USE_TCP, USE_TLS, TLS_HOOKS,
>>>> USE_RAW_SOCKS,
>>>> DISABLE_NAGLE, USE_MCAST, DNS_IP_HACK, SHM_MEM, SHM_MMAP, PKG_MALLOC,
>>>> DBG_QM_MALLOC, FAST_LOCK-ADAPTIVE_WAIT, USE_DNS_CACHE,
>>>> USE_DNS_FAILOVER,
>>>> USE_NAPTR, USE_DST_BLACKLIST, HAVE_RESOLV_RES
>>>> ADAPTIVE_WAIT_LOOPS=1024, MAX_RECV_BUFFER_SIZE 262144, MAX_LISTEN 16,
>>>> MAX_URI_SIZE 1024, BUF_SIZE 65535, PKG_SIZE 4MB
>>>> poll method support: poll, select, kqueue.
>>>> id: ec672a
>>>> compiled on 11:02:58 Nov 15 2011 with gcc 4.2.1
>>>>
>>>> FreeBSD fbsd-kam.dprs-consulting.com 8.2-RELEASE-p2 FreeBSD
>>>> 8.2-RELEASE-p2 #0: Sun Jul 3 17:49:13 GMT 2011
>>>> r...@fbsd-kam.dprs-consulting.com:/usr/obj/usr/src/sys/GENERIC i386
>>>>
>>>> Routing script could be sent on request.
>>>>
>>>> Thank you very much!
>>>>
>>>> WBR
>>>> Dimon
>>>>
>>>>
>>>> _______________________________________________
>>>> SIP Express Router (SER) and Kamailio (OpenSER) - sr-users mailing
>>>> list
>>>> sr-users@lists.sip-router.org
>>>> http://lists.sip-router.org/cgi-bin/mailman/listinfo/sr-users
>> _______________________________________________
>> SIP Express Router (SER) and Kamailio (OpenSER) - sr-users mailing list
>> sr-users@lists.sip-router.org
>> http://lists.sip-router.org/cgi-bin/mailman/listinfo/sr-users
>
_______________________________________________
SIP Express Router (SER) and Kamailio (OpenSER) - sr-users mailing list
sr-users@lists.sip-router.org
http://lists.sip-router.org/cgi-bin/mailman/listinfo/sr-users
