SystemOOMs not reported for containers

[dashpole]

What happened:

A SystemOOM event was not reported even when the memory limit hit was the root cgroup memory limit.

What you expected to happen:

A SystemOOM event should have been reported.

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

Do not set --system-reserved or --kube-reserved, and set --eviction-hard= to disable memory eviction on the kubelet. Run a pod without memory limits which slowly consumes memory, such as with k8s.gcr.io/stress:v1 from https://github.com/vishh/stress. Use arguments -mem-alloc-size", "100Mi", "-mem-alloc-sleep", "10s". This should trigger an OOM on the root cgroup, but no event will be generated.

Anything else we need to know?:

cAdvisor populates ContainerName and VictimContainerName from matching the regexp: Task in (.*) killed as a result of limit of (.*). A SystemOOM should mean VictimContainerName == "/", as we are looking for OOMs that are "killed as a results of limit of /". However, we incorrectly check ContainerName instead in the kubelet.

cc @kubernetes/sig-node-bugs @derekwaynecarr @sjenning @dchen1107

[dashpole]

/sig node
/priority important-soon