NAV has no concept of an "unknown" status

[Morten Brekkevold]

Before a status test has actually been performed (ping, service check, snmp query, etc.) NAV doesn't really know the status of the component. The default is, in most cases, to assume everything is OK until there is an actual alert saying otherwise.

In most cases, a test will be performed as soon as the component is inserted into or discovered by NAV, but if things break down the tests may not be running.

One might also imagine a concept where if a test hasn't been run within a given time period, the status of the component should revert to unknown.

[Morten Brekkevold (mbrekkevold)]

Some suggestions from offline discussions:

Today we have the "Good / not good" dimension. We need a new "Known / Unknown" dimension. Today we have alerting on state transitions between "good / not good". Do we need alerting on state transitions in the other dimension?

Suggest to use a receipt table. Anything that will be checked in NAV must have an entry in the receipt table. Each receipt has a timestamp (or no timestamp). When a timestamp is NULL or has become stale, the current state is "Unknown".

Who is responsible for inserting receipt rows? Who is responsible for updating the timestamps? Who is responsible for removing the receipt rows when no longer applicable?

[Morten Brekkevold (mbrekkevold)]

A preliminary decision from this spring was that a fix bug 1062298 will satisfy this request to a large degree. Therefore we are closing this one as wontfix.