AppController - statefull app deployments about enhancements HOT 19 CLOSED

Generally we prefer to only open features once we have rough consensus from one or more SiGs and a workable proposal. Since you've already started an issue, we can converse and discuss the feature proposal there and then come back and reopen this later. Adding a proposal describing your use cases to the issue is a good next step.

Thanks!

from enhancements.

@smarterclayton I was instructed by @idvoretskyi (sig-openstack lead) to do it that way.

from enhancements.

@smarterclayton we have discussed that with @pigmej and I've proposed him to open the new feature as the soft deadline for 1.4 features is today.

I feel this feature useful in the scope of Kubernetes-OpenStack integration, so I'm advocating to reopen it to continue working on it. Other SIG's are welcome for further discussion.

cc @philips @aronchick to discuss it.

from enhancements.

If the SIG is good, then I'm good. @smarterclayton - were there some
specific concerns?

On Fri, Jul 22, 2016 at 8:08 AM, Ihor Dvoretskyi [email protected]
wrote:

@smarterclayton https://github.com/smarterclayton we have discussed
that with @pigmej https://github.com/pigmej and I've proposed him to
open the new feature as the soft deadline for 1.4 features is today.

I feel this feature useful in the scope of Kubernetes-OpenStack
integration, so I'm advocating to reopen it to continue working on it.
Other SIG's are welcome for further discussion.

cc @philips https://github.com/philips @aronchick
https://github.com/aronchick to discuss it.

—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
#42 (comment),
or mute the thread
https://github.com/notifications/unsubscribe-auth/AADIdZ4qDqJES9WS-O342oMK1ymrtogSks5qYNz2gaJpZM4JSzwl
.

from enhancements.

@idvoretskyi @pigmej Thinking through the spirit of the consensus building I don't think this feature has met the bar. I think generally SIG consensus can be documented a few different means:

• Mailing list discussion
• Notes from a SIG meeting
• Issue or proposal that gets buy in

Right now it appears we have one other person from the SIG and a very young issue (kubernetes/kubernetes#29453) talking about this feature. I agree with closing this issue on those grounds. Lets try and get a bit more documentation on what this feature is and the problems it might be solving for folks before proceeding with a new feature.

from enhancements.

@philips, thank you for clarifying. We will handle it within the SIG-OpenStack and I'm asking @kubernetes/sig-apps to support it.

I'm expecting to reopen the issue later, after clarifying within the SIG(s). Also, I'd like to reserve an exception for adding it to 1.4 after the criteria will be met.

from enhancements.

@philips @aronchick I'd like to propose to define the criteria for submitting the "feature" with more details, I suppose, not all the Kubernetes community members are completely aware of it.

from enhancements.

Since this is discussing the meta-issue of how features should be used, I'll comment here.

Please be aware that SIG buy-in may not be sufficient for all features. For example, API changes require approval of API owners, and entirely new APIs must meet a higher bar.

I also commented on this specific proposal on kubernetes/kubernetes#29453.

from enhancements.

I can't tell if you are saying that you have someone in line to implement
it, or if you are asking sig-apps to find someone to implement it.
People should file feature issues once they have high confidence they have
found people who will implement (and support for some period) the feature.

On Fri, Jul 22, 2016 at 9:22 AM, Brian Grant [email protected]
wrote:

Since this is discussing the meta-issue of how features should be used,
I'll comment here.

Please be aware that SIG buy-in may not be sufficient for all features.
For example, API changes require approval of API owners, and entirely new
APIs must meet a higher bar.

I also commented on this specific proposal on kubernetes/kubernetes#29453
kubernetes/kubernetes#29453.

—
You are receiving this because you are subscribed to this thread.
Reply to this email directly, view it on GitHub
#42 (comment),
or mute the thread
https://github.com/notifications/unsubscribe-auth/AHuudiUBbZ75E5pmkb55CSnAwUL94r_uks5qYO47gaJpZM4JSzwl
.

from enhancements.

@erictune We ( @kubernetes/mirantis ) do have team to implement and support that. This is part of our contribution to k8s world.

from enhancements.

Great, thanks for the clarification.

from enhancements.

@erictune sorry that it wasn't clear from the start.

from enhancements.

Reopening the feature request.

@kubernetes/mirantis have presented the feature demo at the @kubernetes/sig-apps meeting, have received the positive feedback and a "green light" to move forward.

@pigmej please, update and fill the necessary fields in the Progress tracker.

from enhancements.

per the feature burndown meeting today: this will land in kubernetes-incubator first and be promoted based on the criteria in - https://github.com/kubernetes/community/blob/master/incubator.md

from enhancements.

@sarahnovotny we have discussed the ability to make this feature available in Kubernetes codebase and have decided to continue working with the feature implementing. To extend the feature functionality and make it tightly bound with Kubernetes codebase we have decided to continue the development of AppController in Kubernetes 1.5 scope.

from enhancements.

Is the goal to replace kubernetes/helm with whatever is worked on here? Or is this merely orthogonal/related to what tiller does?

from enhancements.

I had a user ask on slack today whether he/she should wait for AppController or use something else, after seeing this issue,

I think that shows there is a cost to having issues in the feature repo for ideas which are still at the concept stage.

from enhancements.

Speaking about this issue, I'm closing this one, because for now we're not joinning k8s core. AppController development will be continued at https://github.com/Mirantis/k8s-AppController, we will prepare design docs etc to discuss it with Kubernetes audience (we had some meetings on sig-apps). AppController will have 3week release cycle (we're going to form more formal roadmap etc in near future, and publish it on our repository)

@numbsafari (sorry for late reply), it's not going to replace helm, we're currently looking and searching for possible areas of cooperation between AppController and Helm.

from enhancements.

I forgot to click "close and comment" instead of "comment".

from enhancements.