NMON alerting duration calculation not as expected
The duration of something being above the configure threshold is currently handled with the range(_time) function.
| stats latest(_time) as _time range(_time) as duration
octamis / metricator-for-nmon / metricator-for-nmon / default / macros.conf — Bitbucket
Duration is used to only trigger an alert when memory / disk full is above a threshold for a longer period.
This period can be configured with alert_fs_min_time_seconds
See image included: Assume two events occur in which the threshold (red) is exceeded for two periods in time (green line above threshold line).
Current behavior: The duration of the event from macro perspective is now the purple arrow because with the range(_time) command we take the maximum and minimum value in which the threshold was exceeded. So it will trigger when two events are more then alert_fs_min_time_seconds
apart (correct me if I’m wrong)
Expectation: I would expect that it only triggers when expected duration 1 (blue arrow below green line), or expected duration 2 (blue arrow below green line) is larger then alert_fs_min_time_seconds
.
Question: Can this be fixed or is my assumption / current behavior incorrect?
Regards Rob van de Voort
Comments (2)
-
-
- changed status to resolved
→ <<cset ec532455cd03>>
- Log in to comment
Hi Rob!
Apologise for the abnormally long delay to respond, I will review your comments and proceed to changes accordingly.
Many thanks for taking the time to write this down, and again, sorry for the delay.
Kind regards,
Guilhem