Minikube: Do not depend on localkube when released

[luxas]

[dlorenc]

Can you explain this a little more? What's wrong with using localkube?

[luxas]

I think we should have the logic in our own code.
So we have control of both the server and the client.

If localkube wants to donate a draft, it would be a great starting point though

[dlorenc]

/cc @vishh

[vishh]

cc @mfburnett

[mfburnett]

Actually, we're happy to donate all of localkube to this project - from a clean abstraction standpoint, though, I think it makes sense to still maintain it as a separate component, which we're also happy to do. Does that address your concerns, @luxas?

[luxas]

Thanks a lot @mfburnett!

[vishh]

I think localkube needs to be tested and released as part of kubernetes release. We should add it to core kubernetes, similar to hyperkube...

[luxas]

Works fine for me

[vishh]

@luxas @mfburnett Will either of you be able to post a patch to add localkube to core kubernetes?

[mfburnett]

@vishh yup, we can do that

[luxas]

@mfburnett Ping me and I'll review

[mfburnett]

@luxas great, thanks

[sebgoa]

if you guys are interested by an alternative for lib machine provisioning, we used the single node deployment here:

https://github.com/skippbox/kmachine/blob/kmachine/libmachine/provision/generic.go

happy to contribute that.

[dlorenc]

We could also move localkube into this repo instead of kubernetes/kubernetes.

[luxas]

We thought about that initially, but @vishh said this

I think localkube needs to be tested and released as part of kubernetes release. We should add it to core kubernetes, similar to hyperkube...

I think I agree

[vishh]

I originally thought that we might want to support upgrades to existing
local clusters by updating just the localkube image, pulled from a reliable
source. But @dlorenc mentioned that localkube would be bundled into
minikube, so there is less reason to move localkube to core kubernetes.
Sorry for the confusion..

On Wed, Apr 20, 2016 at 12:08 PM, Lucas Käldström [email protected]
wrote:

We thought about that initially, but @vishh https://github.com/vishh
said this

I think localkube needs to be tested and released as part of kubernetes
release. We should add it to core kubernetes, similar to hyperkube...

I think I agree

—
You are receiving this because you were mentioned.
Reply to this email directly or view it on GitHub
https://github.com/kubernetes/minikube/issues/7#issuecomment-212561438

[luxas]

@mfburnett Did you notice that @vishh wants it in this repo now?
Just to make sure you don't work on a PR for main Kubernetes :smile:

Or what's the final call @vishh?

[ethernetdan]

@luxas @vishh starting work on this PR tomorrow evening using this repo as the base unless we decide otherwise.

I do see value in testing localkube with Kubernetes releases so that it's kept closely in-sync with development by immediately surfacing breaking changes.

[vishh]

The testing requirements can be met by testing minikube against Kubernetes HEAD.

[dlorenc]

@ethernetdan Thanks for the help. This repo sounds good for now. I'll be on kubernetes slack all day today so feel free to DM me for help getting this merged.

[luxas]

Any update on this getting localkube merged into minikube @ethernetdan and @mfburnett? :smile:

[ethernetdan]

Sorry got caught up moving to SF but am finally settled. Sitting down with @dlorenc tomorrow to hash out details. I'll then get something preliminary out and we'll iterate from there.