WARNING: THIS SITE IS A MIRROR OF GITHUB.COM / IT CANNOT LOGIN OR REGISTER ACCOUNTS / THE CONTENTS ARE PROVIDED AS-IS / THIS SITE ASSUMES NO RESPONSIBILITY FOR ANY DISPLAYED CONTENT OR LINKS / IF YOU FOUND SOMETHING MAY NOT GOOD FOR EVERYONE, CONTACT ADMIN AT ilovescratch@foxmail.com
Skip to content

Conversation

@radekg
Copy link

@radekg radekg commented Nov 5, 2025

This PR addresses #676. Install the software after workers so that when Cilium is installed, there are workers to deploy the CNI pods to.

@sonarqubecloud
Copy link

sonarqubecloud bot commented Nov 5, 2025

@vitobotta
Copy link
Owner

Hi @radekg , sorry for the delay. I cannot merge the pull request as is because it breaks the creation of the cluster for default and more common configurations. By moving the installation of the software to after the installation of k3s on the workers, the create command gets stuck after setting up the workers because they cannot reach a ready state without the CCM being installed first. And then the Cluster Autoscaler installation has to wait for at least one worker to be ready. So the process hangs. Can you please try to figure out a way to make it work with your Cilium config without affecting more common use cases? As another user pointed out in the original thread, they don't have problems with Cilium on several clusters, so it has to be an issue with your particular configuration that requires some changes.

Copy link
Owner

@vitobotta vitobotta left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

See other comment :)

@radekg
Copy link
Author

radekg commented Dec 8, 2025

I'm not sure what breaks the nodes in this case because the exact fixed the problem for Cilium installation.
Having said that, I am not in a position to be burn cycles on figuring out the edge cases I do not use. Feel free to close the issue. I can live with my patched version.

@vitobotta
Copy link
Owner

Let's leave it open for now so I can see if I can find the time to reproduce the issue and look into finding a fix that doesnt break more common configurations.

@radekg
Copy link
Author

radekg commented Dec 8, 2025

Thank you. I understand your concerns regarding other use cases getting broken for whatever reason. The issue with the CNI is that it has to go in first before anything else, and before the first worker can join when workloads land on workers instead of master nodes.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants