Windows + minikube + skaffold + non-elevated shell = fail

[balopat]

Expected behavior

When skaffold detects that minikube's docker daemon can't be resolved (as the warning shows as well) it should either:

• fail when build.local.push = false is setup
• force build.local.push =true when it is not setup

Actual behavior

Skaffold fails to find the minikube docker daemon and hence creates a deployment inside minikube that relies on a non-existent image (as it is not pushed):

C:\Users\balintp\dev\proj\skaffold\examples\getting-started>skaffold dev
time="2019-10-16T15:16:04-07:00" level=warning msg="Could not get minikube docker env, falling back to local docker daemon: getting minikube env: Running [minikube docker-env --shell none]: stdout , stderr: X The docker host is currently not running\n, err: exit st
atus 69: exit status 69"
Generating tags...
 - gcr.io/k8s-skaffold/skaffold-example -> gcr.io/k8s-skaffold/skaffold-example:v0.40.0-58-gd4bd5978-dirty
Tags generated in 413.9981ms
Starting build...
Found [minikube] context, using local docker daemon.
Building [gcr.io/k8s-skaffold/skaffold-example]...
Sending build context to Docker daemon  3.072kB
Step 1/6 : FROM golang:1.12.9-alpine3.10 as builder
 ---> e0d646523991
Step 2/6 : COPY main.go .
 ---> Using cache
 ---> 2015d9353f14
Step 3/6 : RUN go build -o /app main.go
 ---> Using cache
 ---> 0597b5dcee83
Step 4/6 : FROM alpine:3.10
 ---> 961769676411
Step 5/6 : CMD ["./app"]
 ---> Using cache
 ---> a42bf125a1f1
Step 6/6 : COPY --from=builder /app .
 ---> Using cache
 ---> 1cf52c839515
Successfully built 1cf52c839515
Successfully tagged gcr.io/k8s-skaffold/skaffold-example:v0.40.0-58-gd4bd5978-dirty
Build complete in 840.0648ms
Starting test...
Test complete in 1.0005ms
Starting deploy...
kubectl client version: 1.14
pod/getting-started created
Deploy complete in 2.2762852s
Watching for changes...
rpc error: code = Unknown desc = Error response from daemon: manifest for gcr.io/k8s-skaffold/skaffold-example:1cf52c83951577b49c247f19583544e0686629c14245777fd91502cf6488c2e0 not found
Back-off pulling image "gcr.io/k8s-skaffold/skaffold-example:1cf52c83951577b49c247f19583544e0686629c14245777fd91502cf6488c2e0"
rpc error: code = Unknown desc = Error response from daemon: manifest for gcr.io/k8s-skaffold/skaffold-example:1cf52c83951577b49c247f19583544e0686629c14245777fd91502cf6488c2e0 not found

Information

• Skaffold version: master (0.40+)
• Operating system: Windows 10 Enterprise
• Contents of skaffold.yaml: getting started example

Steps to reproduce the behavior

Run Skaffold in a non-admin powershell window against a hyperv minikube instance (that needs admin rights):

1. getting-started example
2. skaffold dev

[nkubala]

I wonder if this is still an issue 🤔

[dgageot]

I'm going to close this one. Not sure this is still an issue. Since then, Skaffold now fails on a failed minikube docker-env

[survivant]

I have the same problem today

PS C:\windows\system32> minikube version
minikube version: v1.12.3
commit: 2243b4b97c131e3244c5f014faedca0d846599f5-dirty

minikube start --vm-driver=hyperv

skaffold build -p dev
creating runner: creating builder: getting docker client: getting minikube env: running [minikube docker-env --shell none]

• stdout: "* The control plane node "" does not exist.\n - To fix this, run: "minikube start"\n"
• stderr: "E0901 08:14:22.467557 24084 main.go:111] libmachine: [stderr =====>] : Hyper-V\Get-VM : Vous ne disposez pas de l'autorisation requise pour effectuer cette t\x83che. Contactez l'administrateur de la strat\x82gie d'autorisation pour l'ordinateur\xffBID00112.\r\nAu caract\x8are Ligne:1 : 3\r\n+ ( H
  yper-V\Get-VM minikube ).state\r\n+ ~~~~~~~~~~~~~~~~~~~~~~~\r\n + CategoryInfo : NotSpecified: (:) [Get-VM], VirtualizationException\r\n + FullyQualifiedErrorId : Unspecified,Microsoft.HyperV.PowerShell.Commands.GetVM\r\n \r\n\n"
• cause: exit status 69

works if I launch skaffold from a "Administrator Powershell"

[briandealwis]

@gsquared94 this seems a motivator to make our KubeContext into a proper struct and move the minikube docker-env into the minikube detection. That way if minikube docker-env fails, then we downgrade the a normal non-local cluster KubeContext. WDYT?