After performing several reboots, SNO fails to come back into a running state. It can sometimes be recovered with additional reboots (may require more than one to recover). However, in at least one instance the node never recovered after 10+ reboots.

kubelet and crio services start appropriately, however, pods fail to deploy and the following error is observed:

rpc error: code = Unknown desc = failed to pin namespaces [0xc000177b90 0xc000177bc0 0xc000177bf0]: taskset: failed to set pid 10248's affinity: Invalid argument

Not easily reproducible. Issue may occur after any given reboot.

Node tuning may also be a factor in reproducing

1. Reboot SNO
2.
3.    

etcd and other critical pods fail to deploy siting a namespace pinning / affinity error

All pods deploy and enter a running state