Kubernetes flannel cni0 degraded carrier

General Discussions
1

I have kubernetes on Ubuntu 22.04. I’ve had networking issues showing up in the syslog with weave or with flannel. You can see various networking links going up and down every few minutes. In the case of flannel, cni0 is in status degraded carrier. Does anyone know what is going on ?

Calling systemctl status systemd-networkd.service return the text below.

Aug 11 17:55:02 ubuntu systemd-networkd[20424]: vethd8f8d04b: Lost carrier
Aug 11 18:00:00 ubuntu systemd-networkd[20424]: vethd34da19a: Link UP
Aug 11 18:00:00 ubuntu systemd-networkd[20424]: veth6eb806f2: Link UP
Aug 11 18:00:00 ubuntu systemd-networkd[20424]: vethd34da19a: Gained carrier
Aug 11 18:00:00 ubuntu systemd-networkd[20424]: veth6eb806f2: Gained carrier
Aug 11 18:00:02 ubuntu systemd-networkd[20424]: veth6eb806f2: Link DOWN
Aug 11 18:00:02 ubuntu systemd-networkd[20424]: veth6eb806f2: Lost carrier
Aug 11 18:00:02 ubuntu systemd-networkd[20424]: vethd34da19a: Gained IPv6LL
Aug 11 18:00:02 ubuntu systemd-networkd[20424]: vethd34da19a: Link DOWN
Aug 11 18:00:02 ubuntu systemd-networkd[20424]: vethd34da19a: Lost carrier

Syslog entries show similar results:

Aug 13 11:50:00 ubuntu kernel: [155362.178296] IPv6: ADDRCONF(NETDEV_CHANGE): veth89866373: link becomes ready
Aug 13 11:50:00 ubuntu kernel: [155362.178360] cni0: port 44(veth89866373) entered blocking state
Aug 13 11:50:00 ubuntu kernel: [155362.178364] cni0: port 44(veth89866373) entered forwarding state
..
Aug 13 11:50:02 ubuntu systemd-networkd[20424]: veth89866373: Link DOWN
Aug 13 11:50:02 ubuntu systemd-networkd[20424]: veth89866373: Lost carrier
Aug 13 11:50:02 ubuntu kernel: [155364.341038] cni0: port 44(veth89866373) entered disabled state
Aug 13 11:50:02 ubuntu kernel: [155364.342241] device veth89866373 left promiscuous mode
Aug 13 11:50:02 ubuntu kernel: [155364.342245] cni0: port 44(veth89866373) entered disabled state
Aug 13 11:50:02 ubuntu networkctl[3451580]: Interface "veth89866373" not found.

Cluster information:

Kubernetes version: 1.25.4
Cloud being used: Azure VM
Host OS: Ubuntu 22.04
CNI and version: Flannel 0.21.5

2

We’ve discovered out flannel configuration was using the default setting of off for ipv6, but ubuntu was using ipv6. Disabled ipv6 and now managed to get something that looks better. All the veth interfaces are in state carrier. Just the cni0 is in carrier-degraded