Is this a BUG REPORT or FEATURE REQUEST?:

/kind bug

/sig testing

What happened:

Tests from test/e2e/common/init_container.go are spawning pods with memory limits (30 MB). Running the tests using Windows nodes will cause the Windows nodes to end up in NotReady state. This happens because the Docker service couldn't start that container (because of the low memory limit) and it hangs. This is a known issue. [1]

Increasing the memory limit to a more reasonable value (50 MB) solves the issue.

[1] moby/moby#37429

What you expected to happen:

The Docker service shouldn't hang just because of the low memory limit imposed by a test.

How to reproduce it (as minimally and precisely as possible):

1. Create a Kubernetes deployment with Windows nodes (Windows Server 1803)
2. Run the tests, or just spawn the following pod:
3. Check the node status: kubectl get nodes
4. On the Windows node, check Docker's responsiveness by issuing a docker CLI command (docker ps -a). Observe that the CLI will not receive any answer and it will hang.

Anything else we need to know?:

Environment:

• Kubernetes version (use kubectl version): v1.11.1
• Cloud provider or hardware configuration: N/A
• OS (e.g. from /etc/os-release): Windows nodes (Windows Server 1803)
• Kernel (e.g. uname -a): N/A
• Install tools: N/A
• Others: