Reboot node pod when p2p too few connections is encountered

[orcutt989]

Instead of printing infinite lines of p2p too few connections the pod should restart itself. Oftentimes this is due to DNS problem or network problem that is resolved at the cloud compute level. When networking is updated at the cloud-compute level the pod needs to be restarted manually.

[nicolasochem]

one way is to have a liveness probe, which performs the same check than readiness probe, but takes longer to trigger (maybe 10 mins)