Path: blob/master/webpage/versioned_docs/version-v2/troubleshooting.md
1464 views
------Troubleshooting
Neko UI loads, but you don't see the screen, and it gives you connection timeout or disconnected error?
Test your client
Some browsers may block WebRTC access by default. You can check if it is enabled by going to about:webrtc or chrome://webrtc-internals in your browser.
Check if your extensions are not blocking WebRTC access. Following extensions are known to block or does not work properly with WebRTC:
Privacy Badger
Private Internet Access
PIA VPN (even if disabled)
Test whether your client supports and can connect to WebRTC.
Networking
If you are absolutely sure, that your client is working correctly, then most likely your networking is not set up correctly.
Check if your ports are correctly exposed using docker
Check that your ephemeral port range NEKO_EPR is correctly exposed as /udp port range.
In following example, specified range 52000-52100 must be also exposed using docker.
Validate UDP ports reachability
Ensure, that your ports are reachable through your external IP.
To validate UDP connection the simplest way, run this on your server:
And this on your local client:
Then try to type on one end, you should see characters on the other side.

If it does not work for you, then most likely your port forwarding is not working correctly. Or your ISP is blocking traffic.
If you get Command 'nc' not found. error, you can install netcat package using:
Check if your external IP was determined correctly
One of the first logs, when the server starts, writes down your external IP that will be sent to your clients to connect to.
Note: Some newer versions of docker-compose use docker compose instead of docker-compose.
You should see this:
If your IP is not correct, you can specify own IP resolver using NEKO_IPFETCH. It needs to return IP address that will be used.
Or you can specify your IP address manually using NEKO_NAT1TO1: (It's read as NAT 1 to 1, so it's capital letter 'O', not zero '0', in NAT1TO1)
If you want to use n.eko only locally, you must put here your local IP address, otherwise public address will be used.
Neko works externally, but not locally
You are probably missing NAT Loopback (NAT Hairpinning) setting on your router.
Example for pfsense with truecharts docker container:
First, port forward the relevant ports
8080and52000-52100/udpfor the container.Then turn on
Pure NATpfsense (under system > advanced > firewall and nat).Make sure to check the two boxes so it works.
Make sure
NEKO_NAT1TO1is blank andNEKO_IPFETCHaddress is working correctly (if unset default value is chosen).Test externally to confirm it works.
Internally you have to access it using
<your-public-ip>:port
If your router does not support NAT Loopback (NAT Hairpinning), you can use turn servers to overcome this issue. See more details here on how to setup local coturn instance.
Neko works locally, but not externally
Make sure, that you are exposing your ports correctly.
If you put local ip as NEKO_NAT1TO1, external clients try to connect to that ip. But it is unreachable for them, because it is your local IP. You must use your public IP address with port forwarding.
Debug mode
To see verbose information from n.eko server, you can enable debug mode using NEKO_DEBUG.
Ensure, that you have enabled debug mode in javascript console too, in order to see verbose information from client.
Frequently Encountered Errors
Common server errors
Check your NEKO_NAT1TO1 or ensure, that NEKO_IPFETCH returns correct IP.
Check if your DNS is set up correctly, and if your IPv6 connectivity is working properly, or is disabled.
Check if your UDP ports are exposed correctly and reachable.
Common client errors
Check if your TCP port is exposed correctly and your reverse proxy is correctly proxying websocket connections. And if your browser has not disabled websocket connections.
Most likely you forgot to add -cap-add=SYS_ADMIN when using chromium-based browsers.
Unrelated server errors
This error originates from browser, that it could not connect to dbus. This does not affect us and can be ignored.
Broadcast pipeline not working with some ingest servers
See related issue.
Some ingest servers require live=1 parameter in the URL (e.g. nginx-rtmp-module). Some do not and do not accept aphostrophes (e.g. owncast). You can try to change the pipeline to: