Automatically scale the default memory allocation

[tstromberg]

• If it's KIC, allow minikube to grow up until the containers maximum footprint, minus 128MB. This safety margin is to avoid unexpected results if someone else has a small docker image running.

• If it's a VM, allocate 25% of the available memory, up to 6GB. Never allocate less than 2.2GB by default. This balances the safety of a well-configured VM up-front, and the ability to run multiple local clusters.

• Specifically allocate memory ranges for most integration tests, so that they can coexist in parallel on a single machine.

Fixes #5021

[k8s-ci-robot]

[APPROVALNOTIFIER] This PR is APPROVED

This pull-request has been approved by: tstromberg

The full list of commands accepted by this bot can be found here.

The pull request process is described here

Needs approval from an approver in each of these files:

• OWNERS [tstromberg]

Approvers can indicate their approval by writing /approve in a comment
Approvers can cancel approval by writing /approve cancel in a comment

[tstromberg]

/ok-to-test

[codecov-io]

Codecov Report

Merging #6900 into master will increase coverage by 0.03%.
The diff coverage is 34.88%.

@@            Coverage Diff             @@
##           master    #6900      +/-   ##
==========================================
+ Coverage   37.11%   37.15%   +0.03%     
==========================================
  Files         144      144              
  Lines        8955     8979      +24     
==========================================
+ Hits         3324     3336      +12     
- Misses       5212     5220       +8     
- Partials      419      423       +4

Impacted Files	Coverage Δ
cmd/minikube/cmd/start.go	32.96% <34.88%> (+0.79%)	⬆️

[minikube-pr-bot]

All Times minikube: [ 93.192010 92.873971 90.675578]
All Times Minikube (PR 6900): [ 94.886672 95.094361 92.362530]

Average minikube: 92.247186
Average Minikube (PR 6900): 94.114521

Averages Time Per Log

+----------------------+-----------+--------------------+
|         LOG          | MINIKUBE  | MINIKUBE (PR 6900) |
+----------------------+-----------+--------------------+
| minikube v           |  0.212472 |           0.209749 |
| Creating kvm2        | 20.543964 |          20.157274 |
| Preparing Kubernetes | 49.096021 |          50.736928 |
| Pulling images       |           |                    |
| Launching Kubernetes | 20.818497 |          21.619912 |
| Waiting for cluster  |  0.058036 |           0.062175 |
+----------------------+-----------+--------------------+

[minikube-pr-bot]

All Times minikube: [ 92.365465 92.052297 93.063564]
All Times Minikube (PR 6900): [ 92.463622 94.669592 92.191196]

Average minikube: 92.493776
Average Minikube (PR 6900): 93.108137

Averages Time Per Log

+----------------------+-----------+--------------------+
|         LOG          | MINIKUBE  | MINIKUBE (PR 6900) |
+----------------------+-----------+--------------------+
| minikube v           |  0.199841 |           0.207232 |
| Creating kvm2        | 20.422828 |          20.085935 |
| Preparing Kubernetes | 50.094833 |          49.921209 |
| Pulling images       |           |                    |
| Launching Kubernetes | 20.444812 |          21.618871 |
| Waiting for cluster  |  0.059486 |           0.062300 |
+----------------------+-----------+--------------------+

[medyagh]

tried it locally with my mac looks good

Password: 
💿  Downloading VM boot image ...
    > minikube-v1.7.3.iso.sha256: 65 B / 65 B [--------------] 100.00% ? p/s 0s
    > minikube-v1.7.3.iso: 167.39 MiB / 167.39 MiB [] 100.00% 16.85 MiB p/s 10s
🔥  Creating hyperkit VM (CPUs=2, Memory=4096MB, Disk=20000MB) 

[afbjorklund]

Did you feedback this to kubeadm, that it no longer works with the recommended 2048M ?

https://kubernetes.io/docs/setup/production-environment/tools/kubeadm/install-kubeadm/

• 2 GB or more of RAM per machine (any less will leave little room for your apps)
• 2 CPUs or more

Or is the problem that it needs 2Gi free - just for Kubernetes, rather than 2Gi total ?

The wording of "per machine" is a little ambiguous. Elsewhere it says "per node"