search

skippbox / kubeadm-centos

An Ansible playbook to create a k8s cluster on CentOS 7.1 using kubeadm bootstrapping.
Apache License 2.0
24 stars 10 forks source link

interest on merging efforts? #2

Open v1k0d3n opened 7 years ago

v1k0d3n commented 7 years ago

@sebgoa would you be interested in merging efforts with a project that i've worked on for my community development team?

the repo is: https://github.com/att-comdev/halcyon-kubernetes

merging would probably be good. my thought is to merge the playbook efforts with a global_var declaration for centos/ubuntu (and other os's that become supported), and then just use the repo as a submodule for a vagrant deployment that can be multi-provider.

what are your thoughts? i could have our team help maintain both, and we can use travis or something else to maintain builds, if you're interested. merging could capture a broader audience.

sebgoa commented 7 years ago

@v1k0d3n why not ! though honestly this was just a quick proof of concept, we are not intending to develop this into a full production playbook. Definitely we will stick to CentOS

Are you guys going to Kubecon, we could meet there.

errordeveloper commented 7 years ago

@davidkbainbridge has also shared an Ubuntu playbook with Vagrant: https://github.com/davidkbainbridge/k8s-playground.

v1k0d3n commented 7 years ago

@errordeveloper this is a good recommendation too. would be fine with either, but would like to be able to declare multiple nodes and deploy to cloud (which would be better outside of the roles anyway; i was going to separate this this soon). do you know the owner of that repo? i think it would be good to merge these efforts for developers who really want multi-node deployment into a local environment.

davidkbainbridge commented 7 years ago

Happy to merge. My plan was to migrate the https://github.com/davidkbainbridge/k8s-playground repo so that the ansible script was a bit more separate from the vagrant bits. This way you could provide an inventory file and use kubeadm to get a cluster working on physical or virtual nodes.

On Mon, Oct 24, 2016 at 7:35 AM Brandon B. Jozsa notifications@github.com wrote:

@errordeveloper https://github.com/errordeveloper this is a good recommendation too. would be fine with either, but would like to be able to declare multiple nodes and deploy to cloud (which would be better outside of the roles anyway; i was going to separate this this soon). do you know the owner of that repo? i think it would be good to merge these efforts for developers who really want multi-node deployment into a local environment.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/skippbox/kubeadm-centos/issues/2#issuecomment-255758407, or mute the thread https://github.com/notifications/unsubscribe-auth/AB3fzU_jRt_SSLPjj32CmEaNjBe1qq7Iks5q3MI-gaJpZM4KWfYQ .

v1k0d3n commented 7 years ago

This is great news! Do you guys want to get together on a hangout to discuss merging these efforts? I'll be back in town after the Openstack Summit; so say...next Friday?

davidkbainbridge commented 7 years ago

Next Friday could work for me. I am in the America/Los_Angeles time zone (PDT).

errordeveloper commented 7 years ago

I've added CentOS support to my fork of @davidkbainbridge repo (https://github.com/errordeveloper/kubernetes-ansible-vagrant), and I managed to do so without disabling firewalld and SELinux.

v1k0d3n commented 7 years ago

@errordeveloper could we combine some deployment strategies with the halcyon repo and this one? this is why I wanted to schedule a discussion, but i'm currently traveling. we may have a slight opinion on this, because we have larger deployment PoC effort. i would hate to keep/maintain two repos that do the same thing (now we have three). once we combine, maybe this would be good to incubate multimode/Vagrant solution for development?

v1k0d3n commented 7 years ago

@davidkbainbridge / @errordeveloper btw, is everyone going to kubecon? maybe we can just talk about it there?

sebgoa commented 7 years ago

Yes, that might be easier to talk at kubecon. Fwiw, kargo which is going to be incubated, will contain kubeadm support.

v1k0d3n commented 7 years ago

definitely a fan of kargo getting incubated (because it's needed), and we support this. the approach to deploying kubeadm is different though, it will eventually remove some of the need for CAPS, is a bit closer to a cloud-native approach; which is the whole reason we started looking at. do we need the whole Ansible playbook model in kargo just to get a development multi-node vagrant deployment? (just a hypothetical question i've been pondering lately).

v1k0d3n commented 7 years ago

@davidkbainbridge @sebgoa can you email me so we can get together at Kubecon or started with merging some of these "vagrant" developmental efforts? my email is bjozsa@jinkit.com.

davidkbainbridge commented 7 years ago

I will send an email, but right now I am not planning to attend Kubecon. I check the site, but see it is sold out at this time. It also happens to be next week and I am in Paris this week for an ONOS event, so being home for a week isn't a bad thing. If there is a way to participate remotely I would be happy to do so.