Fix docker installation for CI

[feiskyer]

Fix docker installation for CI and upgrade docker version to 18.06 17.03.

Update: older version 17.03 is used because of two reasons:

• Fix docker installation for CI #445 (comment).
• Fix docker installation for CI #445 (comment)

[k8s-ci-robot]

[APPROVALNOTIFIER] This PR is APPROVED

This pull-request has been approved by: feiskyer

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 [feiskyer]

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

[Random-Liu]

There is a behavior change in the new docker.

For Docker 17.03, exec process inherits additional group id from the init process:

$ docker run --group-add 1234 busybox id -G
0 10 1234
lantaol@e2e-test-lantaol-minion-group-cq8p ~ $ docker run --group-add 1234 -d busybox sleep 1000
87e52362a4962267f408d6646c3de4482b5f6247de5362103c309bbeadccc175
lantaol@e2e-test-lantaol-minion-group-cq8p ~ $ docker exec 87 id -G
0 10 1234

For Docker 18.09, exec process doesn't inherit additional group id from the init process:

$ docker run --group-add 1234 busybox id -G
0 10 1234
$ docker run --group-add 1234 -d busybox sleep 1000
ded77fec7e0bc07542037ad4a669b7ffed3211d2762fc1d3381fa84af67f28c1
$ docker exec de id -G
0

I'm investigating why the change was made. Tell me if you have more information about this. :)

[Random-Liu]

OK. I think this is a docker bug introduced after switching to containerd 1.0.

Will file a docker issue and link here.

[Random-Liu]

See moby/moby#38865

[feiskyer]

@Random-Liu Thanks for the tip, didn't notice this bug. Let me switch to 18.03 then.

[feiskyer]

Dockershim panic now with errors:

panic: runtime error: invalid memory address or nil pointer dereference
[signal SIGSEGV: segmentation violation code=0x1 addr=0x18 pc=0x32e2d7f]
goroutine 3526 [running]:
k8s.io/kubernetes/pkg/kubelet/dockershim.(*dockerService).RemoveImage(0xc0006f1290, 0x48af4a0, 0xc002393b90, 0xc001b54c78, 0xc0006f1290, 0xc002393b00, 0x3a95260)
	/home/travis/gopath/src/k8s.io/kubernetes/_output/local/go/src/k8s.io/kubernetes/pkg/kubelet/dockershim/docker_image.go:134 +0x9f
k8s.io/kubernetes/pkg/kubelet/apis/cri/runtime/v1alpha2._ImageService_RemoveImage_Handler(0x42b7520, 0xc0006f1290, 0x48af4a0, 0xc002393b90, 0xc00016ebd0, 0x0, 0x0, 0x0, 0x22666669642f6239, 0x69446b726f57222c)
	/home/travis/gopath/src/k8s.io/kubernetes/_output/local/go/src/k8s.io/kubernetes/pkg/kubelet/apis/cri/runtime/v1alpha2/api.pb.go:4807 +0x23e
k8s.io/kubernetes/vendor/google.golang.org/grpc.(*Server).processUnaryRPC(0xc000228700, 0x48ce9a0, 0xc0022d6e00, 0xc0023d4d00, 0xc0005ea180, 0x76184a8, 0x0, 0x0, 0x0)

Filed kubernetes/kubernetes#75367 to fix this issue.

[feiskyer]

For workarounds, the tests for triggering dockershim panic have been skipped.

Self applied lgtm since it also blocks other PRs.