Replies: 2 comments 3 replies
-
In general it's not a good idea to switch out major bits of your cluster networking configuration (CNI plugin, CNI backend, subnets, etc) once the cluster is already up and running. You may be able to resolve this by rebooting all the nodes in order to clear the old interfaces and routes, but if you continue to run into problems after rebooting all the nodes with the new flag set, you're probably best off rebuilding the cluster. |
Beta Was this translation helpful? Give feedback.
-
New Installation. Why is k3s not managing the routes?
|
Beta Was this translation helpful? Give feedback.
-
Environmental Info:
K3s Version:
k3s version v1.21.0+k3s1 (2705431)
go version go1.16.2
Node(s) CPU architecture, OS, and Version:
Centos 7 x64
Cluster Configuration:
3 servers, 0 agents
Describe the bug:
After boostraping a 3 nodes cluster and deploying some workload, I want to change flannel backend from default VXLAN to host-gw.
I followed instructions from #538 and restarted k3s service on all nodes after changing systemd but I have the feeling that this is not working.
How can I check that host-gw is working fine?
Steps To Reproduce:
--flannel-backend=host-gw
Expected behavior:
Flannel should switch to host-gw
Actual behavior:
Unknown
Beta Was this translation helpful? Give feedback.
All reactions