Videoconference TCP fallback not working

Ports used see diagram https://github.com/jitsi/jitsi-meet/blob/master/doc/manual-install.md

root@heu20:/etc# netstat -tulpn | grep 4443
tcp6       0      0 10.47.48.20:4443        :::*                    LISTEN      508/java        
root@heu20:/etc# netstat -tulpn | grep 443
tcp        0      0 0.0.0.0:4438            0.0.0.0:*               LISTEN      488/nginx -g daemon
tcp        0      0 127.0.0.1:6443          0.0.0.0:*               LISTEN      455/sandstorm   
tcp6       0      0 10.47.48.20:4443        :::*                    LISTEN      508/java        
tcp6       0      0 :::443                  :::*                    LISTEN      532/sniproxy

Starting Nmap 7.01 ( https://nmap.org ) at 2017-12-07 20:28 CET
Nmap scan report for meet.fairchat.net (178.63.167.183)
Host is up (0.037s latency).
rDNS record for 178.63.167.183: static.183.167.63.178.clients.your-server.de
PORT     STATE SERVICE VERSION
4443/tcp open  pharos?

Service detection performed. Please report any incorrect results at https://nmap.org/submit/ .
Nmap done: 1 IP address (1 host up) scanned in 129.26 seconds
ras@kar:~$ nmap -sV -p 4443 meet.ucom.gv.at

Starting Nmap 7.01 ( https://nmap.org ) at 2017-12-07 20:52 CET
Nmap scan report for meet.ucom.gv.at (62.99.130.124)
Host is up (0.026s latency).
PORT     STATE SERVICE VERSION
4443/tcp open  pharos?

Service detection performed. Please report any incorrect results at https://nmap.org/submit/ .
Nmap done: 1 IP address (1 host up) scanned in 129.22 seconds

Ports scheinen zu passen.

In /etc/jitsi/meet/meet.fairchat.net-config.js

useStunTurn: true,

gesetzt wie in ucom, obwohl es bei Jitsi heisst dass kein TURN server gebraucht wird, war false. Die Relay Funktion übernimmt jvb.

added explicitly TCP_HARVESTER_PORT in /etc/jitsi/videobridge/sip-communicator.properties

org.jitsi.videobridge.AUTHORIZED_SOURCE_REGEXP=focus@auth.meet.fairchat.net/.*
org.ice4j.ice.harvest.NAT_HARVESTER_LOCAL_ADDRESS=10.47.48.20
org.ice4j.ice.harvest.NAT_HARVESTER_PUBLIC_ADDRESS=178.63.167.183
org.jitsi.videobridge.TCP_HARVESTER_PORT=4443

    /**
     * The default port that the <tt>TcpHarvester</tt> will
     * bind to.
     */
    private static final int TCP_DEFAULT_PORT = 443;

    /**
     * The port on which the <tt>TcpHarvester</tt> will bind to
     * if no port is specifically configured, and binding to
     * <tt>DEFAULT_TCP_PORT</tt> fails (for example, if the process doesn't have
     * the required privileges to bind to a port below 1024).
     */
private static final int TCP_FALLBACK_PORT = 4443;

Still problems with conferences with 2+ patricipants: one often drops.

Harvester addresses in /etc/jitsi/videobridge/sip-communicator.properties really improve connectivity in same LAN (tested again) (not there in meet.ucom.gv.at).

Set p2p mode to false.

Maybe an issue of the stable Nov2017 version: audio drops, when moderator leaves channel: https://github.com/jitsi/jitsi-meet/issues/2299

closed

/etc/jitsi/videobridge/sip-communicator.properties is configured

org.jitsi.videobridge.TCP_HARVESTER_PORT=4443

Connecting App Hatler Brunnen - no media

JVB 2018-12-15 20:53:23.907 INFO: [5781008] org.ice4j.ice.ConnectivityCheckClient.log() Pair failed: 54.37.204.4:10000/udp/host -> 192.168.179.111:35764/udp/host (stream.RTP)
JVB 2018-12-15 20:53:23.928 INFO: [5781008] org.ice4j.ice.ConnectivityCheckClient.log() Pair failed: 54.37.204.4:4443/tcp/host -> 10.73.207.155:9/tcp/host (stream.RTP)

JVB 2018-12-15 20:53:26.711 WARNING: [5779433] org.jitsi.videobridge.EndpointMessageTransport.log() SCTP connection with fb6d8d25 not ready yet.
JVB 2018-12-15 20:53:26.712 WARNING: [5779433] org.jitsi.videobridge.EndpointMessageTransport.log() No available transport channel, can't send a message

Hatler Brunnen to yesss network - no media

JVB 2018-12-15 21:09:44.971 INFO: [5785422] org.ice4j.ice.ConnectivityCheckClient.log() Pair failed: 54.37.204.4:10000/udp/host -> 192.168.179.119:43050/udp/host (stream.RTP)
JVB 2018-12-15 21:09:44.991 INFO: [5785422] org.ice4j.ice.ConnectivityCheckClient.log() Pair failed: 54.37.204.4:4443/tcp/host -> 192.168.179.119:9/tcp/host (stream.RTP)

"make sure the process is able to listen on 443 (from a permissions perspective)"

reopened

use authbind 443 https://community.jitsi.org/t/jvb-how-to-listen-to-port-443/15017/3

But 443 is being used by nginx?? Do we need a second IP? https://community.jitsi.org/t/jvb-how-to-listen-to-port-443/15017/4?u=rasos

Yes, we need to multiplex 443 or uye a turn server https://community.jitsi.org/t/videobridge-behind-nginx-proxy-443-4443-because-4443-port-is-closed/14913

meet.jit.si has a turn server on an extra machine with extra LE certificate on port 443 https://community.jitsi.org/t/one-way-media-only-nat-issues-after-upgrade/15328/6?u=rasos

Hi @roland.alton - just adding a small note on here that I'm interested in this issue too since it appears very similar to something I've reported at https://github.com/jitsi/jitsi-meet/issues/3738 ; if there's anything we can do to collaborate and find the root cause I'd be glad to help.

Hi @jayaddison looks like we have slightly different usage scenarios. In fact we came up with the conclusion to split signalling and the videobridge as outlined here #19 (closed)

Thanks @roland.alton - I'm being a little stubborn about the single-jitsi-machine approach I'm trying to take, but thank you for the suggestion and link; I'll take that route if I can't get the conferencing working on an individual machine.

We finally added a second IP address to the VM and let the media stream connect directly to the videobridge.

closed

Videoconference TCP fallback not working

Designs

Child items ...

Activity