Open lausser opened 1 year ago
Sounds like a bug to me
Disabling enable_flap_detection
does not reset all existing flapping flags. It simply prevents new flapping flags from being set.
But i agree, it should not prevent notifications from being sent out.
Isn't the flapping flag removed on the next check execution? If not, that would probably be sensible imo.
I don't think this is done aleady, but yes, that might be the way to go.
Thinking about it, simply resetting the flag might not be a good idea. The global enable_flap_detection
flag can be (temporarily) changed by an external command.
So this would result in lots of "starting to flap" notifications again.
One way would be to check the global flag at least in the notifications logic. Better than nothing, but this would still show the host/service as flapping in the UI unless the UI checks the global flag as well.
Naemon simply cannot predict whether the enable_flap_detection
is just temporarily disabled or forever.
Besides the quick fix from #452 we could think about slowly letting grow out the state history in case the host/service flapping flag is set, even if flapping detection is disabled. For example, each time a check result arrives, update the state history used for flapping detection with an OK untill the list is cleared. (But silently ignoring flapping stop notification)
I have a naemon 1.4.1 (OMD 5.21.20230729-labs-edition) where i disabled flap detection globally. But there is one service which is still shown as flapping (also in the logfile with [1690795047] SERVICE NOTIFICATION SUPPRESSED: snclient;os_windows_svc_check_Server;Notification blocked because the object is currently flapping.
In the retention date one can clearly see the two different settings:
Also Thruk shows the flapping symbol. Shouldn't everything be 0 when enable_flap_detection in the naemon.cfg was disabled?