Martin, Matthew wrote:
> 1) It seems awkward to me that during a defined escalation, the
> escalation contact_groups **replace** the contact_groups defined in the
> host or service check. It would make more sense (to me at least), if the
> escalation contact_groups were **added** to the list of notifications.
I agree. However it seems to be the correct behavior. Please read #2 to
find the way to fix it.
> 2) I also think it would make more sense if increasing severity of
> escalations could be defined, with additional contacts added at each
> level. For example a service defined with “HelpDesk” as a contact for
> **all** notifications; a “Severity 1” escalation being defined as
> beginning with the 3^rd unhandled notification and adding the contact
> group “Supervisors”; a “Severity 2” escalations being defined as
> beginning with the 6^th unhandled notification and adding the contact
> group “Managers”, etc. On the 7^th notification all three groups would
> be getting notified without having to define “HelpDesk” as part of
> “Severity 1”, or both “HelpDesk” and “Supervisors” in “Severity 2”.
If you use Nagios3, you should use "inheritance":
> 3) Different escalations definable for different severity of alert,
> Warning, Critical, Unknown, etc.
You should create 2 hostescalation/serviceescalation, each of them
having different escalation_options.
> 4) Apparently notifications and escalations do not work the same
> for passive checks as active checks. I have defined passive checks with
> “notification_interval 60” and escalations with “first_notification 3”.
> My expectation was that this would behave exactly the same as an active
> check would, notifications hourly to the group defined in the check,
> with escalations after 3 hours to the group(s) in the escalation. In
> fact, since the passive check only happens once each day, I only get one
> notification to the contact group defined in the service on each of the
> first two days, and then notifications to the contact groups defined in
> the escalations thereafter. Not what I expected.
Notifications seem to be triggered when Nagios is receiving new results
from the "check result reaper" for the specific host. Nagios won't
escalate the problem by itself if there's no new results.
Although this FAQ entry does not specify it explicitly, the same logic
applies for your case: