replace-session-connection

[H Yavari]

Hi all 

Updating to the latest version seems broken the IP translation from ipv6 to ipv4 in "c=". We were using the replace-session-connection flag, but it's been removed. 

What is the solution? How can we fix it ?

Thanks,

Hossein

[Richard Fuchs]

What exactly is "it" you wish fixed?

Post a concrete example of what you think is broken.

Also be specific as to what the "latest version" is for you.

Cheers

[H Yavari]

Thanks for the prompt response.

When users have this SDP:

...
s=-
c=IN IP6 2001:56f:c011:9fc2:cd4:dc2e:a8be:841c
t=0 0
a=sendrecv
m=audio 49120 RTP/AVP 109 104 110 102 108 105 100
b=AS:49
b=RS:0
b=RR:0
....

And we have a public interface IPv4 so we do [ipv6,ipv4]
and we are using the ["origin", "session-connection"] We updated to the latest version (mr13.3.1.4), and now RTPEngine keeps using IPv6 and doesn't translate ipv6 to IPv4, which we expect.

Please let me know if I have missed any information that you need.

Thanks,

Hossein

[Richard Fuchs]

On 02/09/2025 13.42, H Yavari wrote:
> Please let me know if I have missed any information that you need.

Yes, things like your configuration and the complete set of offer/answer
flags would be relevant, especially those relevant to interfaces and
address family selection.

A "concrete example" ideally would be something you take directly from
the log file with full debugging enabled.

Cheers

[Andrew Pogrebennyk]

Hi all,

To chime in, in release 12.5, the session connection already triggers a log message indicating that an obsolete flag is being used - Hossein, you might want to look out for that.

The default behavior to my understanding is: in cases like the one that might be illustrated here, when the media-level connection is absent, the session-level connection would be used instead, no flag needed.

I didn’t mean to hijack the thread, but I’m also curious about the scenario where both session-level and media-level c= lines are present. If only the media-level line is replaced, wouldn’t that potentially reveal more about the network topology from one side than intended?

Looking forward to your thoughts.

[Richard Fuchs]

On 09/09/2025 06.00, Andrew Pogrebennyk wrote:
> I didn’t mean to hijack the thread, but I’m also curious about the
> scenario where both session-level and media-level c= lines are
> present. If only the media-level line is replaced, wouldn’t that
> potentially reveal more about the network topology from one side than
> intended?

The behaviour in 12.5+ is to always use media-level connection lines in
output SDPs and never session-level lines, regardless of where the
line(s) were present in the received SDP. That's ultimately why the flag
became obsolete. So far I have not heard any of any problems this might
cause. 🤞

Cheers

[Andrew Pogrebennyk]

Thanks. Should it remove the session-level c= line when there are media-level connection lines too?

[Richard Fuchs]

On 09/09/2025 07.56, Andrew Pogrebennyk wrote:
> Thanks. Should it remove the session-level c= line when there are
> media-level connection lines too?

Yes, there never are any session-level c= lines in output SDPs.

Cheers

[Andrew Pogrebennyk]

OK I switched to mr13.2 and can confirm the behavior. Just mr12.5 did not strip the line. Thanks!