Removed dependency on localkube to wait for the network to be up

[aaron-prindle]

Ref: #1277

[r2d4]

I think we have to split up the commands before we remove this dependency. I could be wrong though

[codecov-io]

Codecov Report

Merging #1298 into master will increase coverage by 0.04%.
The diff coverage is n/a.

@@            Coverage Diff            @@
##           master   #1298      +/-   ##
=========================================
+ Coverage   43.06%   43.1%   +0.04%     
=========================================
  Files          48      48              
  Lines        2271    2271              
=========================================
+ Hits          978     979       +1     
+ Misses       1133    1132       -1     
  Partials      160     160

Impacted Files	Coverage Δ
pkg/minikube/cluster/commands.go	61.79% <ø> (ø)	⬆️
pkg/minikube/kubeconfig/config.go	56.92% <0%> (+1.53%)	⬆️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 9750a7e...a11b471. Read the comment docs.

[dlorenc]

I think we have to split up the commands before we remove this dependency. I could be wrong though

I don't think so. IIRC we might have to investigate some flakiness inside of localkube itself.

[dlorenc]

Aaron, can you run the tests like 10 times and see if there are any new flakes?

[aaron-prindle]

test 1
1 failing test (Linux-KVM)
test 2
1 failing test (Linux-KVM)
test 3
2 failing tests (Linux-KVM and Linux-Virtualbox)
test 4
2 failing tests (Linux-KVM and OSX-Virtualbox)
test 5
2 failing tests (Linux-KVM and OSX-Virtualbox)

[aaron-prindle]

@minikube-bot retest this please

[aaron-prindle]

@minikube-bot retest this please

[r2d4]

I'm a little concerned that KVM hasn't passed. I'm pretty sure his was the issue that we saw before

[aaron-prindle]

We could make waiting for the network be driver specific to KVM. Or we could wait for a custom network check that avoids the delay issues mentioned in #1277

[r2d4]

Maybe we can do some manual testing to see how much faster this gets us

[dlorenc]

@minikube-bot test this please

[dlorenc]

@r2d4 the error looks different than before, it's actually related to SSH'ing in. I just tested this with KVM locally a bunch of times and didn't see any issues.

[dlorenc]

I ran some completely non-scientific benchmarks. With this change (linux/kvm):

1m7.390s
1m3.078s
0m57.739s
0m58.045s
0m58.333s
1m1.243s
0m56.404s
1m0.327s
0m58.563s
0m56.925s

Without it:

2m57.594s
3m2.195s
2m57.785s
2m57.945s
2m57.570s
2m56.606s
2m57.460s
2m57.988s
3m0.304s
2m56.295s

This looks pretty significant.

[r2d4]

I'm fine merging this in once we fix the KVM test. I haven't gotten a chance to try this out, but I'll look at it today and try to look at the issue.

[dlorenc]

@aaron-prindle could you rebase and re-run the tests?

[aaron-prindle]

rebased, re-running tests

[dlorenc]

@minikube-bot test this please