engine: docker healthcheck logLength index fix

[adnxn]

Summary

This change addresses #1237 by adding a zero check before accessing the Health.Log slice. Also adds corresponding test.

The issue was surfaced when the agent recieved a Docker health check message indicating the container was unhealthy but sent an empty []HealthCheck in Health.Log.

We have not reproduced the issue, but are handling this case by doing a length check on Health.Log.

Implementation details

Testing

• Builds on Linux (make release)
• Builds on Windows (go build -out amazon-ecs-agent.exe ./agent)
• Unit tests on Linux (make test) pass
• Unit tests on Windows (go test -timeout=25s ./agent/...) pass
• Integration tests on Linux (make run-integ-tests) pass
• Integration tests on Windows (.\scripts\run-integ-tests.ps1) pass
• Functional tests on Linux (make run-functional-tests) pass
• Functional tests on Windows (.\scripts\run-functional-tests.ps1) pass

New tests cover the changes:

Description for the changelog

Licensing

This contribution is under the terms of the Apache 2.0 License:

[jahkeup]

I was able to reproduce this state of the health check using this Dockerfile:

FROM debian:latest
HEALTHCHECK --interval=3s --timeout=1s --start-period=56s CMD /bin/true
CMD ["/bin/sleep", "60"]

And running it using the following commands:

docker build -t healthchecked .
docker run -d --name healthchecked healthchecked
# wait for container to start
sleep 2 
# check the container's state for health checks observed:
docker inspect -f '{{.State.Health}}' healthchecked

Which shows the health state with no checks performed and still in the starting state:

{starting 0 []}

[aaithal]

@jahkeup in your example, the container's health status is starting and not actually unhealthy right?

[jahkeup]

Yep, I see where we're checking the logs now. Let me try this out with the Agent running to see if there's some other way that the logs could be empty.

[jahkeup]

Doesn't appear that we could easily get into this state, the Docker daemon always appends to its Log field when handling a check result and this is also where it's setting the unhealthy/healthy status.

[aaithal]

@adnxn don't forget to add an entry in the chanelog before merging this.