K3s - Traefik 2.2 UDP packetloss #7000
Assignees
Labels
area/udp
kind/bug/possible
a possible bug that needs analysis before it is confirmed or fixed.
status/5-frozen-due-to-age
Projects
Comments
rtribotte
added
kind/bug/possible
|
Hi @dragon2611 , could you share your configuration, logs and provide a reproducible example ? |
|
Edit: Mumble doesn't show much in the way of Diagnostic info, infect I didn't notice something was wrong until it was reported by my friend there was no audio. (it being a bit difficult to test that without a 2nd client connected) TS3 is showing the packet loss in the client info on a user connected through traefik, In the outbound direction if it helps any (I.e from the server back towards the client), Inbound packet loss seems fine. The TS config, sadly it's prior to putting it in git so I just commented out the nodeport and uncommented the ingress routes My Traefik 2.2 config TS3 server log Traefik logs from the pod I believe I was hitting `These logs from one of the traefik pods might be useful, I can grab all of the logs but probably need to upload them somewhere Note I removed and redeployed TS and Traefik, I suspect the middleware error is due to that and I need to re-apply it, Although that middleware isn't used in either the TS or mumble configs. |
|
I do have the same problem. I was very exited, when Traefik finally anounced UDP Support, but I wasnt able to use it. My Teamspeak Server (running on Docker) had a very high packet loss and I was forced to revert the traffik back to the bridge network. This is my very basic configuration of traefik: traefik.yml entryPoints:
teampseak-voice:
address: ":9987/udp"dynamic_conf.yml udp:
routers:
teamspeak-voice:
entryPoints:
- "teamspeak-voice"
service: teamspeak-voice@file
services:
teamspeak-voice:
loadBalancer:
servers:
- address: "teamspeak:9987"I have the official teamspeak Container named The Client successfully connects to the server, but for some moments, the trafiic completly cuts off and others are stuttering. After I have reverted to the direct bridge network, all of the mentioned errors went away. |
|
Hey all, I tried to reproduce your issue on a fresh digitalocen droplet with the given docker-compose version: "3.7"
services:
teamspeak:
image: mbentley/teamspeak
environment:
TS3SERVER_LICENSE: accept
labels:
- "traefik.udp.routers.ts3.entrypoints=ts3"
- "traefik.enable=true"
traefik:
image: traefik:v2.3
restart: always
command:
- "--log.level=DEBUG"
- "--api.insecure=true"
- "--providers.docker=true"
- "--providers.docker.exposedbydefault=false"
- "--entrypoints.traefik.address=:8080"
- "--entrypoints.ts3.address=:9987/udp"
ports:
- "80:80"
- "8080:8080"
- "9987:9987/udp"
volumes:
- "/var/run/docker.sock:/var/run/docker.sock:ro"However, I couldn't. tested with a group of 4 people and all stayed stable throughout our test. I guess, we will need more information / or another reproducable example. |
|
Hi! I'm Træfiker 🤖 the bot in charge of tidying up the issues. I have to close this one because of its lack of activity 😞 Feel free to re-open it or join our Community Forum. |
|
@SantoDE it's normal because you use 2.3 of traefik and author use 2.2.1. 2.2.2 version have two fixes about udp : It probably fixe the problem encourtered by author, he just need to update traefik (tested yesterday) |
|
I'll check which version I'm running as I've updated since opening this and re-enable the UDP listener |
Sign up for free
to subscribe to this conversation on GitHub.
Already have an account?
Sign in.
What did you do?
Configured Traefik to listen on UDP for TeamSpeak 3 (UDP 9987, TCP:30033 TCP:10011) and Mumble (TCP/UDP 64738)
Traefik is configured with HOST networking on the nodes as there's no load balancers and I want Traefik to see the origin IP for some middlewares I use for other services.
What did you expect to see?
Traefik Passes through UDP connections to pod
What did you see instead?
UDP connections are successful, but TS is reporting packet loss between 5% and 20% , Mumble you connect but there is no voice traffic.
Doesn't seem to make a difference if you are hitting Traefik on the same node or a different one to the one running the pod.
Output of
traefik version: (What version of Traefik are you using?)2.2.1
What is your environment & configuration (arguments, toml, provider, platform, ...)?
K3s v1.18.4+k3s1 running on 3 nodes (master + 2 workers, Calcio + Wireguard networking)
Reconfiguring to use Nodeport instead of via Traefik solves the packetloss issue.
The text was updated successfully, but these errors were encountered: