This message was deleted.

Seems to be a regression: v1.32.2+k3s1 doesn't crash.

Running k3s with docker container

You mean you run k3d or something else?

No. k3s docker image.

How do you start the image? Just curious

services:

k3s:

labels:

com.centurylinklabs.watchtower.enable: "false"

image: rancher/k3s:v1.32.2-k3s1

container_name: k3s-server

privileged: true

restart: unless-stopped

networks:

- npm

volumes:

- /var/run/docker.sock:/var/run/docker.sock

- /var/lib/rancher/k3s:/var/lib/rancher/k3s

- /var/lib/kubelet:/var/lib/kubelet

- /etc/rancher:/etc/rancher

- /home/ubuntu/docker:/home/ubuntu/docker

- /mnt/vol1/data:/mnt/vol1/data

ports:

- "6443:6443"

- "80:80"

- "443:443"

- "30006:30006"

- "30003:30003"

command: ["server", "--disable", "traefik"]

I started getting this constantly a few days ago. I was running on an Ubuntu 20.04 VM, installed with the script, and it's been running fine for a few years, but then a few days ago it hit a restart loop with this "failed to find interface with specified node ip" error. I chalked it up to being an old deployment and decided to start fresh with a clean install of Ubuntu 24.04 and K3s, but it's now in the same restart loop.

Turned out disabling ipv6 with sysctl caused this bug:

cat /etc/sysctl.d/70-disable-ipv6.conf

net.ipv6.conf.default.disable_ipv6=1

net.ipv6.conf.lo.disable_ipv6=1

net.ipv6.conf.all.disable_ipv6=1

@bland-easter-37603 caused or fixes? IPv6 isn't disabled in my case

@dry-musician-87363 reverting that file in my case fixed the bug.

hm, alright, thanks, doesn't seem to apply to my situation. I've narrowed it down to the upgrade from v1.32.5+k3s1 to v1.32.6+k3s1, in which the dependency on kube-router was updated from v2.2.1 to v2.5.0 (the "failed to find interface with specified node ip" error message comes from there)

Seems about right. v1.32.5+k3s1 was the last working version as well before I got the above breakthrough. Good luck bug hunting!

Thanks. Given the vanillaness of my setup I thought it'd be easier to replicate. It's really just an Ubuntu 24.04 instance on GCE with K3s and pretty much nothing else installed, and reproduces within a day, but if I go create an instance now I can restart all I want and it doesn't happen. Wait a day and it does and can't recover.