There is a race condition in which items with active escalations keep getting escalated, even though the action is disabled.
I do not yet have hard data (exact SQL sequences) to back this up, but we've had it consistenly a few times like:
- A host has a trigger in the PROBLEM state
- At the very moment that the web-dashboard is giving the notice of "in progress" for the trigger actions, we disable the action that is causing it
As a result: the current problem state will keep getting escalated, even if the trigger returns back to an "OK" state afterwards, because the action was seemingly disabled.
We have yet to upgrade our 1.8.9 to the latest 1.8.11 (it's in testing now), so I must apologize if this has already been fixed in 1.8.11.