Skip to content

[🔴] ACTION REQUIRED: Change of default simplestreams server #180

@neoaggelos

Description

@neoaggelos

Summary

Dear users,

The default simplestreams server used by cluster-api-provider-incus is migrating to a new place. This means existing CAPN deployments must be updated to v0.8.3 or newer, to use the new simplestreams server.

What Do You Need To Do?

IF YOU ARE:

  • using cluster-api-provider-incus version v0.8.2 or earlier.
  • relying on the default kubeadm images.

YOU MUST DO ONE OF THE FOLLOWING:

  • Upgrade to cluster-api-provider-incus v0.8.3 or newer (in progress).
  • Manually update your cluster manifests to use a custom image, or the new image server (see below).

What Is Going To Happen?

Starting from 2025-12-19, the cluster-api-provider-incus project will switch to the following two simplestreams servers:

Starting from 2025-12-31, the old simplestreams servers will be shut down.

Why Is This Happening?

Infrastructure costs for the CloudFront distributions that server our project image servers are kindly subsidized by the National Technical University Of Athens.

So far, these were:

Unfortunately, the owner project has not been renewed, and is shutting down unexpectedly and with very little notice, effective immediately. It is not clear when the distributions will be terminated, but starting from 2025-12-19 they are expected to start becoming unavailable. Image pulls will fail.

This has left the project with very limited options about how to proceed.

Will Existing Clusters Be Affected?

NO, EXISTING CLUSTERS WILL NOT BE AFFECTED.

Since the image is already pulled, any instances will keep running without issues, using the local copy of the image.

However, it will likely be impossible to spin up new instances with these images, as the image reference will not be available.

What Steps Is The Project Taking?

Apologies

While all this is in line with the project's disclaimer and support policy with the image server, we want to say that we are deeply sorry for the inconvience this is expect to cause. We are users of open source projects as well, but issues like this are not always preventable.

That said, we want to share that we are trying to address the arising situation as gracefully as possible, always within the confines of our limited resources.

With that, we want to echo the recommendation from our documentation page:

It is recommended that production environments build their own custom images instead.

Questions

For any questions, recommendations or help, please comment on this issue.

Metadata

Metadata

Assignees

Labels

No labels
No labels

Type

No type

Projects

No projects

Milestone

No milestone

Relationships

None yet

Development

No branches or pull requests

Issue actions