0

I deployed a kubernetes cluster of v1.13.0 a month ago, using kubeadm, with one master-node and three worker-nodes in it. And everything was ok.
but when i was going to register a new worker to this cluster. kube-proxy deamonset container start up with the error log below: 

W0103 09:41:57.631256       1 proxier.go:477] Failed to load kernel module ip_vs with modprobe. You can ignore this message when kube-proxy is running inside container without mounting /lib/modules  
W0103 09:41:57.631256       1 proxier.go:477] Failed to load kernel module ip_vs with modprobe. You can ignore this message when kube-proxy is running inside container without mounting /lib/modules
W0103 09:41:57.631983       1 proxier.go:477] Failed to load kernel module ip_vs_rr with modprobe. You can ignore this message when kube-proxy is running inside container without mounting /lib/modules
W0103 09:41:57.632666       1 proxier.go:477] Failed to load kernel module ip_vs_wrr with modprobe. You can ignore this message when kube-proxy is running inside container without mounting /lib/modules
W0103 09:41:57.633330       1 proxier.go:477] Failed to load kernel module ip_vs_sh with modprobe. You can ignore this message when kube-proxy is running inside container without mounting /lib/modules
W0103 09:41:57.638341       1 server_others.go:295] Flag proxy-mode="" unknown, assuming iptables proxy
W0103 09:41:57.802455       1 node.go:108] Failed to retrieve node IP: host IP unknown; known addresses: []
I0103 09:41:57.802480       1 server_others.go:148] Using iptables Proxier.
W0103 09:41:57.802597       1 proxier.go:314] invalid nodeIP, initializing kube-proxy with 127.0.0.1 as nodeIP

As the kube-proxy is inside container, so the main problem is : 

W0103 09:41:57.802455       1 node.go:108] Failed to retrieve node IP: host IP unknown; known addresses: []

e

(version compatibility has been proved not the main cause of this issue)

hoping for someone who expert on this can help me.

George Liang
  • 131
  • 1
  • 10

2 Answers2

1

I finally locate the essential cause of this issue by checking the kubelet log, which reports a warning that the route doesn't have a default route. After adding a default route manually, the problem is fixed and everything goes well.

George Liang
  • 131
  • 1
  • 10
0

Quoting from the official documentation for contributors regarding Kubernetes release versioning and backward compatibility/component skew:

...Nodes may lag masters components by up to two minor versions but should be at a version no newer than the master...

so I guess this is simply not supported cluster setup.

Nepomucen
  • 4,449
  • 3
  • 9
  • 24
  • Thank you for answering. I've downgrade kubelet and kubeadm on this worker. But the kube-proxy still failed to retrieve node ip.... – George Liang Jan 04 '19 at 02:45