On Mar 13, 2015, at 8:27 PM, Brian Grant notifications@github.com wrote:

We should start creating PRs to flip to v1beta3, so that we can start testing and merge when ready:

enable v1beta3 by default in apiserver
change to v1beta3 by default in kubectl
kubelet to use v1beta3 on apiserver
kubelet to expose v1beta3 and use v1beta3 for local pods
kube-proxy to use v1beta3 on apiserver
replication controller controller to use v1beta3
node controller to use v1beta3
endpoint controller to use v1beta3
use v1beta3 for etcd storage
There's also the issue of whether to provide a v1beta1->v1beta3 storage translation tool.

At a minimum, being able to test the cluster runs correctly when content is upgraded is important. I have my doubts.
ref. #1519
cc @smarterclayton @lavalamp

—
Reply to this email directly or view it on GitHub.

@smarterclayton You have your doubts that it will work correctly or that it's possible to test whether it works correctly? :-)

Yes :)

On Mar 13, 2015, at 8:31 PM, Brian Grant notifications@github.com wrote:

@smarterclayton You have your doubts that it will work correctly or that it's possible to test whether it works correctly? :-)

—
Reply to this email directly or view it on GitHub.

Since all clients use latest.Version to talk to a server, most internal code can be changed centrally.

@smarterclayton That's convenient, assuming everything works. :-)

Starting with the first task: "enable v1beta3 by default in apiserver"

Verified that e2e tests pass even when master stops exposing v1beta1 URLs.

I take that back. I didnt realize that my e2e tests were running on a different branch.
I re-ran them but it looks like a bunch of tests are flaky on head: #7548

@bgrant0607 says this can be subdivided and farmed out.

Elimination of v1beta1 and v1beta2: #8087

Any other pieces you think this can be divided into? Or the remaining work is reasonable for one person (namely @nikhiljindal )?

I'm sure this, #5799, #6584, and #8087 could each be broken up more. There are still 150 references to v1beta1 in the repo.

I don't think there's an issue for porting e2e, but @brendandburns is working on it.

The remaining issues are #6583, #6584, and #5799. Closing this since there are more specific issues filed.