Where should alerts go when a node status is UNKNOWN ?

cgregors

We are having an internal discussion about this question.

My position is that if a node / vol / interface / app goes into an unknown state it needs to be investigated from an Orion point of view.

Alerts should not go to the support groups when the status is unknown. There is typically nothing that they can do to resolve the issue since it is most likely an Orion issue.

Your thoughs / comments ?

steve.audette

cgregors,

Unknown does not necessarily imply that it's a problem with Orion. In the context of interfaces, apps or volumes, if the node goes down, the associated interfaces and volumes will register unknown status. For example, if your core router goes down and the WAN port that's on it becomes unknown, that's probably something you'd want an alert on. A good way around this scenario is to add in a condition on the alert that specifically says Node Status != Down.

Of course, a lot of this depends on what your other alerts are already monitoring and what the responsibilities of the alerted parties are.