Alarms for Anomaly Detections
Overview
CloudWatch allows you to create alarms on log anomaly detectors to monitor log groups for anomalies. These alarms trigger actions based on the number and priority of detected anomalies during a specified time.
Steps to Create an Alarm for a Log Anomaly Detector:
Access the CloudWatch Console:
Navigate to Logs > Log Anomalies.
Select the Anomaly Detector:
Choose the radio button for the anomaly detector you want to monitor.
Click Create alarm.
Configure Alarm Settings:
Priority Filters (Optional):
Use HIGH to trigger only on high-priority anomalies.
Use MEDIUM to trigger on high- and medium-priority anomalies.
Threshold Type:
Choose Static (a fixed value) or Anomaly detection (CloudWatch-calculated threshold).
Set Alarm Conditions:
Define how the alarm triggers (e.g., greater than, lower than).
Set the period (time frame) for evaluating anomalies.
Additional Configuration:
Datapoints to Alarm:
Specify how many consecutive periods must breach the threshold to trigger the alarm.
Create M out of N alarms by setting the first value (M) lower than the second (N).
Missing Data Treatment:
Configure how the alarm behaves when data points are missing.
Set Notifications (Optional):
Add Amazon SNS notifications for ALARM, OK, or INSUFFICIENT_DATA states.
(Optional) Configure actions for Auto Scaling, EC2, or Systems Manager when the alarm triggers.
Name and Describe the Alarm:
Enter a name and description (supports markdown for adding runbook links or internal resources).
Review and Create the Alarm:
Confirm the alarm configuration and click Create alarm.
Tips and Recommendations:
Set the alarm to notify on INSUFFICIENT_DATA state to capture potential Lambda or data source issues.
Use Systems Manager actions only for alarms in the ALARM state.
Last updated