Idea #8236
Updated by Brett Smith over 8 years ago
* Add code to _check_poll_freshness that triggers the existing on_failure code to cause Node Manager to die if any of the lists are hopelessly stale.
* There should be a new configuration knob to decide when the poll is hopelessly stale.
* It should default to some multiple of the configured freshness time.
* Ops can decide what the default multiplier is, and maybe the name of the configuration value.