Not receiving a TCP (srflx) candidate in EC2 while using Docker about jitsi-videobridge HOT 7 CLOSED

We don't currently support automatically detecting the public address for TCP. You can work around this by setting it up manually. The relevant properties are:
org.jitsi.videobridge.NAT_HARVESTER_PUBLIC_ADDRESS
org.jitsi.videobridge.NAT_HARVESTER_LOCAL_ADDRESS

from jitsi-videobridge.

We have tried setting both NAT_HARVESTER_PUBLIC_ADDRESS to the public IP address of EC2 and NAT_HARVESTER_LOCAL_ADDRESS to the private IP addresses without success. Might there be anything else we can try?

from jitsi-videobridge.

Sounds like this could be a bug. Is there anything different in the logs with these properties set?

from jitsi-videobridge.

You could try to disable the AWS harvester (you don't need it if you set NAT_HARVESTER manually). I don't understand why the problem occurs, so I don't know if this will help. I will try and look into it in detail when I get a chance, but probably not until at least next week.

from jitsi-videobridge.

We have actually solved the issue. We had thought we knew all of the ports that the Videobridge uses but it seems like we missed exposing something in Docker. After running it in HOST network mode in Docker, it looks like we are receiving all the UDP and TCP candidates. Thank you!

from jitsi-videobridge.

Glad to hear that you got it to work. Keeping this open, because we should probably implement automatically creating the mapping for TCP when we detect we're on EC2.

from jitsi-videobridge.

Should be fixed with ice4j r505 in 9a5bf25

from jitsi-videobridge.