Move from the old m3.medium instance type to the marginally better t2.medium

[mumoshu]

ref #181

[codecov-io]

Current coverage is 69.11% (diff: 100%)

Merging #184 into master will not change coverage

@@             master       #184   diff @@
==========================================
  Files             4          4          
  Lines          1130       1130          
  Methods           0          0          
  Messages          0          0          
  Branches          0          0          
==========================================
  Hits            781        781          
  Misses          261        261          
  Partials         88         88          

Powered by Codecov. Last update 639829c...0a925c8

[pieterlange]

-1 Lets not use t2 for controllers by default - you should really know what you're doing when using this instance class.

Also, 4GB of memory seems a bit on the low side when you start actually using the cluster. My not keep the m3.medium and suggest to use m4.large or t2.large?

[mumoshu]

@pieterlange Thanks for your feedback here!

I believe setting good defaults and moving away from the old m3.medium instance type are separate things. So:

• Would you please read my comment about that? To be more clear, It's true that I began to believe t2.medium is better than m3.medium but I'm not opposed to make m4.large or t2.large by default.

• Also, if you have any concrete advice on choosing which instance type(s) as default(s), I'd appreciate it if you could read about my suggestions and reply with your suggestions on the instance type for etcd, controller and worker nodes 👍

[mumoshu]

Changing the subject according to what we've discussed in #181 (comment)

[mumoshu]

Things seems to be left up in the air but I'm going to merge this to get the ball rolling 🙇

Let's discuss further about better default(s) in #181