Adjusting "Periods Over Threshold" to prevent unwanted APM Alarms
Article ID: 220389
Updated On:
Products
Issue/Introduction
The below alert configuration is causing unwanted APM alarms, the resolution value of 6 minutes is the length of 1 period; therefore, a caution alert will be issued if the caution threshold has been below 30% for 30 consecutive minutes within a 90-minute rolling window.
From Metric View:
However, the below unwanted Caution alert was generated:
2021-07-23T13:36:00+0000
Message: The alert Capacity Kubernetes CPU Request Deviation - Namespaces has breached the MAJOR threshold of 30
Metric Name: SuperDomain|HOSTNAME|ClusterDeployment|Infrastructure Agent|Kubernetes|<example>:CPU Request Deviation
Metric Value: 8
Severity: major
...
Environment
- DX APM 2*
- DX O2
Resolution
An alert will be issued if the caution threshold has been below 30% for any 5 periods below threshold during the last 15 periods (6 minutes is the length of 1 period).
If the purpose is to only get an alert if there is consecutive threshold breach past 30 minutes, then set the alarm configuration as below
- "Periods Over Threshold" 5
- "Observed Periods" 5.
Additional Information
Feedback
