minikube start --download-only downloads duplicate kubeadm/kubelet binaries

[tstromberg]

kubeadm & kubelet are already included in the preload tarball.

😄  minikube v1.12.3 on Darwin 10.15.5
✨  Automatically selected the docker driver. Other choices: hyperkit, vmware, parallels, virtualbox, vmwarefusion
👍  Starting control plane node minikube in cluster minikube
💾  Downloading Kubernetes v1.18.3 preload ...
    > preloaded-images-k8s-v5-v1.18.3-docker-overlay2-amd64.tar.lz4: 510.91 MiB
    > kubelet.sha256: 65 B / 65 B [--------------------------] 100.00% ? p/s 0s
    > kubeadm.sha256: 65 B / 65 B [--------------------------] 100.00% ? p/s 0s
    > kubectl.sha256: 65 B / 65 B [--------------------------] 100.00% ? p/s 0s
    > kubectl: 41.99 MiB / 41.99 MiB [---------------] 100.00% 3.03 MiB p/s 14s
    > kubeadm: 37.97 MiB / 37.97 MiB [---------------] 100.00% 2.23 MiB p/s 17s
    > kubelet: 108.04 MiB / 108.04 MiB [-------------] 100.00% 3.21 MiB p/s 33s
    > kubectl.sha256: 65 B / 65 B [--------------------------] 100.00% ? p/s 0s
    > kubectl: 47.84 MiB / 47.84 MiB [---------------] 100.00% 4.10 MiB p/s 12s
✅  Download complete!

[n0npax]

I'll try to fix this.

/assign n0npax

[afbjorklund]

It also includes kubectl for linux (for uploading to the node, at least)

If you want to download kubectl for the host, it might need a download.

[n0npax]

I'll unassign, as seems it requires a little bit better understanding of the project.

[afbjorklund]

I'll unassign, as seems it requires a little bit better understanding of the project.

I think it was almost done, just that we missed that kubectl was also in the preload.

i.e. there is one kubectl for linux, that is in the preload and that we install to the node
then there is one optional kubectl client for host OS, that we download on demand
in this case, we want to skip downloading the kubectl linux that is already in the preload
should also make the code easier, since it doesn't have to special-case like before ?

[priyawadhwa]

Help wanted on this.

[fejta-bot]

Issues go stale after 90d of inactivity.
Mark the issue as fresh with /remove-lifecycle stale.
Stale issues rot after an additional 30d of inactivity and eventually close.

If this issue is safe to close now please do so with /close.

Send feedback to sig-testing, kubernetes/test-infra and/or fejta.
/lifecycle stale

[sharifelgamal]

This issue is still outstanding

[medyagh]

CC @andriyDev

[andriyDev]

This issue is still a problem. I'll look into it more deeply.

[andriyDev]

/assign andriyDev