Unable to Register SIP client with IPv6

[Andy-2639]

What version of ejabberd are you using?
How did you install ejabberd (source, package, distribution)?

https://www.process-one.net/downloads/downloads-action.php?file=/ejabberd/17.03/ejabberd-17.03-windows-installer.exe

What operating system (version) are you using?

Windows 10 Pro x64 Anniversary Update (Version 1607 Build 14393.969)

What did not work as expected? Are there error messages in the log? What
was the unexpected behavior? What was the expected result?

My SIP client (Phoner Lite 2.47, http://www.phonerlite.de/ ) can register to my ejabberd SIP server via IPv4 (TCP and UDP) without problems.
When I try to use IPv6, the register fails and PhonerLite says a timeout occurred. This happens with TCP and UDP.

SIP client and ejabberd are running on different maschines in the same LAN. No firewall is enabled on either maschine.

I attached the logs of the connections (ejabberd log, PhonerLite debug output and wireshark packet captures) zipped. Wireshark ran on the same maschine on which ejabberd runs.
logs_pcaps.zip

SIP configuration in my ejabberd.yml:

listen: 
  - 
    ip: "0.0.0.0"
    port: 65060
    transport: udp
    module: ejabberd_sip
  - 
    ip: "::"
    port: 65060
    transport: udp
    module: ejabberd_sip
  - 
    ip: "0.0.0.0"
    port: 65060
    module: ejabberd_sip
  - 
    ip: "::"
    port: 65060
    module: ejabberd_sip
modules: 
  mod_sip: []

[zinid]

Don't you have any errors in the log? It's not always possible to define the same port globally to both IP protocols. Do you see the ports are opened on all IP interfaces?

[Andy-2639]

Don't you have any errors in the log?

No, there are no errors in the log.
The ejabberd logs are in logs_pcaps.zip "ejabberd [tcp/udp] v[4/6].log". For every protocol combination, I deleted the old logs and started a fresh ejabberd instance.

It's not always possible to define the same port globally to both IP protocols. Do you see the ports are opened on all IP interfaces?

I tested it again with port 5060 for IPv4 TCP/UDP and port 65060 for IPv6 TCP/UDP. This gives the same behavior. I also testes the bahavior of Phoner Lite 2.47: when the SIP server is not reacheable I get an connection error message instead of an timeout.
Also wireshark indicates that the ports are open: no TCP RST packets. In the ejabberd logs, I can see the REGISTER messages from Phoner Lite, too (ejabberd tcp v6.log, line 327; ejabbers udp v6.log, 326 - 509).
XMPP works fine with IPv4 and IPv6 on the same port number.

[zinid]

I tested it again with port 5060 for IPv4 TCP/UDP and port 65060 for IPv6 TCP/UDP.

Do you mean to say you disabled ipv4 listener and still don't have it working with ipv6 listener?

[Andy-2639]

Neighter

listen: 
  - 
    ip: "0.0.0.0"
    port: 5060
    transport: udp
    module: ejabberd_sip
  - 
    ip: "::"
    port: 65060
    transport: udp
    module: ejabberd_sip
  - 
    ip: "0.0.0.0"
    port: 5060
    module: ejabberd_sip
  - 
    ip: "::"
    port: 65060
    module: ejabberd_sip

nor

listen: 
  - 
    ip: "::"
    port: 5060
    transport: udp
    module: ejabberd_sip
  - 
    ip: "::"
    port: 5060
    module: ejabberd_sip

works on my system.

[Andy-2639]

I tested SIP with IPv6 again with ejabberd 17.08. It still fails.
The logs of phonerlite and ejabberd and the complete ejabberd configuration is in the zip:
logs_config.zip

SIP with IPv4 works fine. I enabled listening only on one IP version at a time.

It seems that ejabberd is receiving the requests vom the SIP client but just does not send a reply.

The test ran in a VM with a newly installed Windows 10 Pro x64 and a newly installed ejabberd 17.08. The windows firewall was deactivated and phonelite and ejabberd ran on the same VM.

[zinid]

I'm not sure I can help because I don't have windows installed and IPv6 works fine for me on Linux.

[aylesm]

Any progress with this? I'm running centos and have enabled IPv6. I can see from the logs the server receives the SIP REGISTER message. But never responds.

[zinid]

I still have it working:

$ telnet ::1 5060
Trying ::1...
Connected to ::1.
Escape character is '^]'.
REGISTER sip:zinid.ru SIP/2.0
Via: SIP/2.0/TCP 198.211.126.52:43403;branch=z9hG4bK.23e5441e;rport;alias
From: sip:xram@zinid.ru;tag=4cc1fd1f
To: sip:xram@zinid.ru
Call-ID: 1287781663@198.211.126.52
CSeq: 1 REGISTER
Content-Length: 0
Max-Forwards: 70
User-Agent: sipsak 0.9.7pre
Expires: 15
Contact: sip:xram@198.211.126.52:43403;transport=TCP

SIP/2.0 401 Unauthorized
Via: SIP/2.0/TCP 198.211.126.52:43403;branch=z9hG4bK.23e5441e;alias;received=[::1];rport=54724
From: <sip:xram@zinid.ru>;tag=4cc1fd1f
Call-ID: 1287781663@198.211.126.52
CSeq: 1 REGISTER
Max-Forwards: 70
To: <sip:xram@zinid.ru>;tag=2332915767-1429473788
Server: ejabberd 17.09.121
WWW-Authenticate: Digest realm="zinid.ru", qop="auth", nonce="8aa552d7fd797fec03f613ef91f856859d72ec97"
Content-Length: 0

[zinid]

I have tried on two Linux machines, btw.

[aylesm]

Have you tested over TLS?

[zinid]

Yes, it works as well.

[zinid]

Have you enabled debug (loglevel: 5)? What do you see in the log?

[aylesm]

I see the incoming request in the logs. But I don't see the response. So it feels like this is an outbound issue in my setup. Not sure if I've waited long enough to see a timeout or other relevant log msg.

[zinid]

This is strange because you should see data_out log message within 32 seconds max.