Hi,

I also have a preference for minikube – even if Kind has the ability of
launching multi-node clusters, which minikube can't do.

Historically, Kind was used in one single place: the run.sh script at the
root of the repo. I think it was chosen so that we can launch a local
Docker registry (cf. the companion kind-registry.sh script) to upload
Docker images for Polaris. But minikube has a much easier way of achieving
that [1], so my take is that we don't need Kind at all at this point.

Thanks,
Alex

[1]:
https://minikube.sigs.k8s.io/docs/handbook/pushing/#1-pushing-directly-to-the-in-cluster-docker-daemon-docker-env

On Mon, Jul 7, 2025 at 11:57 PM Yufei Gu <flyrain...@gmail.com> wrote:

> +1 on picking one, instead of having both. I don't have any preference
> though.
>
> Yufei
>
>
> On Mon, Jul 7, 2025 at 1:37 PM Dmitri Bourlatchkov <di...@apache.org>
> wrote:
>
> > My personal preference is minikube.
> >
> > Cheers,
> > Dmitri.
> >
> > On Sat, Jul 5, 2025 at 8:22 PM Yong Zheng <yzh...@apache.org> wrote:
> >
> > > Hello,
> > >
> > > Currently, we provide Kubernetes references for both Kind (
> > > https://kind.sigs.k8s.io/) and Minikube (
> > > https://minikube.sigs.k8s.io/docs/). In terms of tooling, Kind is very
> > > lightweight, whereas Minikube is much more feature-rich and works out
> of
> > > the box. (Some workarounds or additional steps are required to make
> Kind
> > do
> > > the same, but those details are out of scope for this discussion.)
> > >
> > > In terms of the features and documentation we offer to others, either
> > > option will work. However, I think the overhead of maintaining support
> > for
> > > both tools is unnecessary. I don’t have a strong preference for which
> one
> > > we choose, but I believe it will be easier to support additional
> > automation
> > > if we stick to a single tool.
> > >
> > > Thanks,
> > > Yong Zheng
> > >
> >
>

Reply via email to