I'm currently setting up Kubernetes version 1.35, and I'm facing some serious speed issues. The package downloads are crawling along, barely hitting 100 kbps. When I try to run `kubeadm init`, the image downloads are just as frustratingly slow. It's taking forever—over 45 minutes and I'm still stuck at the download stage for several images, including the kube-apiserver and kube-controller-manager. I really need some help figuring out what's going wrong here!
2 Answers
To improve your experience, consider setting up something like Harbor on a Docker container. This way, you only need to download the images once, which speeds everything up and protects you from potential issues if a repository goes offline.
Have you checked whether this issue is happening only on the specific system you're using to set up Kubernetes, or is it also affecting other devices on your network? Sometimes, specific hops can cause delays. Is the slowness only occurring when pulling images, or are you noticing it with other sizeable downloads too?
It seems like the problem only happens with pulls from registry.k8s.io. I've heard that the redirector for registry.k8s.io isn't a true CDN, which could mean variable speeds based on your location. Some backends in Europe can get overloaded or might be down for maintenance, and you might end up redirected to a US server instead. Plus, containerd doesn't have any mirrors set up by default, so you're stuck with just one endpoint.