IPv6 issues with candidates

53 views
Skip to first unread message

Rani Yaroshinski

unread,
Dec 31, 2021, 11:34:49 AM12/31/21
to meetecho-janus
We noticed some weird issues with the ICE protocol and  Janus-Gateway, when there are candidates gathered with IPv6 addresses also.
First we get the infamous "sockaddr 17" error, and that is why I know there is such a problem. The weird thing, is that this is causing an ICE failed error, with the annoying, and not giving any information message "stream 1 in component 1 failed".
I have use the lib_nice_debug flag there, and nothing can be seen cause actually Janus is able to overcome this issue, but this is still causing a rather unexplainable issue there.

There are weirder things here :
[1] It seems to only happen with the ICE gathering protocol (no-trickle) mechanism, but when allowing trickling is seems to be avoided - which probably means there is some sort of bug there, cause there is no reason that trickle will solve the issue, since the IPv6 addresses are not supported in Janus-Gateway, and nothing can be done with them, anyway, and actually the trickling is not needed.
[2] It is also solved when the network stack of the machine sending those is setup in such a way that IPv6 is not used.

The issue here is that not always I can control all the peers with IPv6 candidates, and there must be  a more robust solution to this.

Any suggestion, or at least explanation to this issue ?
Reply all
Reply to author
Forward
0 new messages