hetznercloud
Integrated Robot servers - Cloud Network
TL;DR
Hello,
maybe someone can help out, we have a vSwitch configured, and connected with our cloud servers. The dedicated machine is configured, and can reach all our control panels on the cloud side.
We use k3s and have a cluster deployed on several cloud machines.
On the deciated machine, we have connected it to our cluster sucessfully.
curl -sfL https://get.k3s.io | K3S_URL="https://10.1.0.4:6443" K3S_TOKEN="xyz" INSTALL_K3S_VERSION="v1.32.0+k3s1" sh -s - agent --flannel-iface=enp5s0.4001 --node-external-ip=1.2.3.4 --kubelet-arg="cloud-provider=external" --node-name=mynode --node-ip=10.1.1.30
We can schedule Pods there, but all Pods can not reach the CoreDns Server or other IPs of the cloud network. Does some has similar problems and found a solution for that situation?
Thanks!
Expected behavior
It Works :-)
Observed behavior
It does not :-(
Minimal working example
No response
Log output
Additional information
No response
Hey,
so from what you have provided, you might have the setup for Robot and private networks incorrectly configured. Could you please take a look at guides/robot/private-networks.md, explanation/private-networks.md and explanation/robot.md? By default, if you supply a valid private Networks configuration the routes-controller of HCCM will run and try to create the necessary routes for the cluster. This won't work when you use Robot nodes, as there are some limitations to the product with respect to adding custom routes.
Furthermore, I suggest you use the --disable-cloud-controller flag of k3s to disable the k3s cloud controller.
If none of this solves your issue, could you please provide me more details on your HCCM config and possibly some logs messages and Kubernetes events of HCCM and other affected resources?
Best Regards Lukas
