Set apiserver oom_adj to -10 to avoid OOMing before other pods

[tstromberg]

This will allow users to be able to continue to run kubectl to diagnose a cluster that has OOM'd.

I'm open to other suggestions: apparently kubeadm doesn't let me set resource limits on pods in order to get placed into the guaranteed QoS class.

For reference, kubelet is configured for -16.

Example log messages:

I0517 07:50:08.225437   41021 ssh_runner.go:137] Run with output: cat /proc/$(pgrep kube-apiserver)/oom_adj
I0517 07:50:08.234262   41021 utils.go:240] > 16
I0517 07:50:08.234291   41021 kubeadm.go:234] apiserver oom_adj: 16
I0517 07:50:08.234299   41021 kubeadm.go:239] adjusting apiserver oom_adj to -10
I0517 07:50:08.234306   41021 ssh_runner.go:101] SSH: echo -10 | sudo tee /proc/$(pgrep kube-apiserver)/oom_adj
I0517 07:50:08.248105   41021 utils.go:240] > -10

[k8s-ci-robot]

[APPROVALNOTIFIER] This PR is APPROVED

This pull-request has been approved by: tstromberg

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 [tstromberg]

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