Node exporter not configured to use correct proc location in some circumstaces

[GrantFleming]

What happened?

In the node-exporter daemonsets:

• The host filesystem is mounted at /host/root
• The procfs location is not set, so it uses the default of /proc
• The procfs location should probably be set to /host/root/proc
• The difference between /proc and /host/root/proc probably doesn't matter much in most cases, but there are some where it does
• If your Kubernetes node is an LXC container, only /host/root/proc shows the correct stats for the node, /proc shows status for the host on which the node is running

Did you expect to see some different?

I expected node-exporter to be configured to use /host/root/proc for appropriate stats

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

Create a Kuberetes cluster where at least one node is an LXC container. Install the quick-start stack. Observe that the wrong cpu/mem values are displayed for the LXC node due to these values being taken from /proc rather than /host/root/proc