Begin to separate out the etcd cluster and the master clusters

[mirthy]

Do not merge yet, testing separating out the master/controller from etcd.

[wellsie]

Tack started out that way ;) I merged etcd with k8s masters a while back -
mainly to cut down on under utilized instances. I also liked the idea of
the apiservers being really close to etcd.

It will be nice to have both options available.
On Wed, Aug 24, 2016 at 8:56 PM Jeff Kwan [email protected] wrote:

Do not merge yet, testing separating out the master/controller from etcd.

You can view, comment on, or merge this pull request online at:

#50
Commit Summary

• Begin to separate out the etcd cluster and the master clusters

File Changes

• M io.tf https://github.com/kz8s/tack/pull/50/files#diff-0 (2)
• M modules.tf https://github.com/kz8s/tack/pull/50/files#diff-1
  (31)
• M modules/etcd/cloud-config.tf
  https://github.com/kz8s/tack/pull/50/files#diff-2 (80)
• M modules/etcd/io.tf
  https://github.com/kz8s/tack/pull/50/files#diff-3 (1)
• A modules/master/cloud-config.tf
  https://github.com/kz8s/tack/pull/50/files#diff-4 (147)
• A modules/master/ec2.tf
  https://github.com/kz8s/tack/pull/50/files#diff-5 (34)
• R modules/master/elb.tf
  https://github.com/kz8s/tack/pull/50/files#diff-6 (2)
• A modules/master/io.tf
  https://github.com/kz8s/tack/pull/50/files#diff-7 (25)
• A modules/master/route53.tf
  https://github.com/kz8s/tack/pull/50/files#diff-8 (19)
• M modules/route53/route53.tf
  https://github.com/kz8s/tack/pull/50/files#diff-9 (8)
• M modules/security/io.tf
  https://github.com/kz8s/tack/pull/50/files#diff-10 (1)
• M modules/security/security.tf
  https://github.com/kz8s/tack/pull/50/files#diff-11 (32)

Patch Links:

• https://github.com/kz8s/tack/pull/50.patch
• https://github.com/kz8s/tack/pull/50.diff

—
You are receiving this because you are subscribed to this thread.
Reply to this email directly, view it on GitHub
#50, or mute the thread
https://github.com/notifications/unsubscribe-auth/AE6-kY6Je7tkeWuwNkE5pn2zCtvrj37hks5qjRJhgaJpZM4JsrBf
.

[mirthy]

I ran into this thinking about how to handle upgrades on the api server side without rebuilding the cluster entirely.