origin: OpenShift builds cannot start due to cgroup error (when cgroups are not enabled)

On the default “jessie” deployment for Kubernetes (under aws), I finally have OpenShift builds working (thank you @liggitt and @ncdc)

It seems now that issue #7297 has resurfaced.

Contents of /sys/fs/cgroup/cpu/:

admin@ip-172-20-0-9:/sys/fs/cgroup/cpu$ ls
cgroup.clone_children  cgroup.sane_behavior  cpuacct.usage         cpu.shares  notify_on_release  tasks
cgroup.procs           cpuacct.stat          cpuacct.usage_percpu  docker      release_agent

Error output from build:

F0316 21:46:24.866361       1 builder.go:204] Error: failed to retrieve cgroup limits: cannot determine cgroup limits: open /sys/fs/cgroup/cpu/cpu.cfs_quota_us: no such file or directory

I would probably assume that this was an issue with the privileged flag not working correctly (perhaps due to Kubelets being potentially configured to not allow privileged pods) but the path does not exist in the host system, so I think this might be a feature that’s not enabled in the kernel or something of this nature.