[SR-Users] Strange core.tcp_list / tls.list output

[Daniel-Constantin Mierla]

Hello,

quickly checking the code it seems that the dst_ip is holding the local
address always, even for outbound connection.

Not being the author of the code, my guess is that the lookup of
connection is done always on src_ip matching the remote peer address.
Also, the structure used there is the one specific for holding received
info for sip packets.

The output is somehow misleading, maybe the fields should be renamed to
local_ip/port and remote_ip/port in the output of the rpc commands.

Cheers,
Daniel

On 03/06/15 20:27, Mikko Lehto wrote:
> Hi, I noticed thath command core.tcp_list in kamcmd gives output that
> does not make sense to me:
> ---
> kamcmd> core.tcp_list
> {
>         id: 1
>         type: TLS
>         state: CONN_OK
>         timeout: 3599
>         ref_count: 1
>         src_ip: 192.168.47.132
>         src_port: 5061
>         dst_ip: 10.10.235.9
>         dst_port: 54769
> }
> {
>         id: 2
>         type: TLS
>         state: CONN_OK
>         timeout: 3599
>         ref_count: 1
>         src_ip: 192.168.47.132
>         src_port: 32854
>         dst_ip: 10.10.235.9
>         dst_port: 5041
> }
> ---
>
> In my scenario I am connecting via TLS to host 192.168.47.132:5061.
> And host 192.168.47.132 is connecting to my proxy host 10.10.235.9:5041.
> pcap verifies this. Also tls.list gives similar output.
>
> Values regarding incoming connection (id=2) makes total sense.
> It seems that values regarding outgoing connection (id=1) have
> dst_* values swapped with src_* values.
>
>
> To debug, I placed this logging to tcp_main.c:
> ---
> $ git diff tcp_main.c
> diff --git a/tcp_main.c b/tcp_main.c
> index 5830c8e..3717cf6 100644
> --- a/tcp_main.c
> +++ b/tcp_main.c
> @@ -1207,6 +1207,7 @@ again:
>         }
>         from=&my_name; /* update from with the real "from" address */
>         su2ip_addr(&ip, &my_name);
> +       LM_WARN("after getsockname()  my_name=%s server=%s ip=%s\n",
> 		su2a(&my_name, sizeof(my_name)), su2a(server, sizeof(*server)),
> 		ip_addr2a(&ip));
>  find_socket:
>  #ifdef USE_TLS
>         if (unlikely(type==PROTO_TLS))
> @@ -1223,6 +1224,7 @@ find_socket:
>                 else *res_si=sendipv6_tcp;
>         }
>         *res_local_addr=*from;
> +       LM_WARN("before return        my_name=%s server=%s ip=%s\n",
> 		su2a(&my_name, sizeof(my_name)), su2a(server, sizeof(*server)),
> 		ip_addr2a(&ip));
>         return s;
>  error:
>         if (s!=-1) tcp_safe_close(s);
> ---
>
> This gives following output:
> WARNING: <core> [tcp_main.c:1210]: tcp_do_connect(): after getsockname() my_name=192.168.47.132:5061 server=192.168.47.132:5061 ip=10.10.235.9
> WARNING: <core> [tcp_main.c:1227]: tcp_do_connect(): before return my_name=192.168.47.132:5061 server=192.168.47.132:5061 ip=10.10.235.9
>
>
> Are tcp endpoint values really wrong way around or did I undertand
> something wrong about their meaning?
>
> I am running Kamailio 4.2.5 (x86_64/linux) 61d84c-dirty from git.
>

-- 
Daniel-Constantin Mierla
http://twitter.com/#!/miconda - http://www.linkedin.com/in/miconda
Book: SIP Routing With Kamailio - http://www.asipto.com