Skip to main content

CloudWatch

A page for viewing AWS CloudWatch alarm states at a glance and examining detailed configurations.

CloudWatch

Key Features​

Alarm State Summary​

  • OK: Number of alarms in normal state (green)
  • ALARM: Number of triggered alarms (red)
  • INSUFFICIENT_DATA: Number of alarms with insufficient data (gray)

Visualizations​

  • Alarm State Distribution: Pie chart showing alarm distribution by state (semantic colors: OK green · ALARM red · INSUFFICIENT_DATA gray)
  • The alarm table sorts worst-first by default (ALARM → INSUFFICIENT_DATA → OK, newest state change first); a column-header sort overrides it
  • Alarms by Namespace: Bar chart showing alarm count by namespace

Alarm List​

ColumnDescription
Alarm NameName of the alarm
NamespaceAWS service namespace (AWS/EC2, AWS/RDS, etc.)
MetricMetric being monitored
StateCurrent state (OK, ALARM, INSUFFICIENT_DATA)
ReasonReason for state change
ActionsWhether actions are enabled

Alarm Details​

Click on an alarm row to view details in the slide panel:

  • Alarm: Name, ARN, state, state reason
  • Configuration: Comparison operator, threshold, evaluation periods, statistic
  • Actions: Actions executed on alarm/OK/insufficient data (SNS, Lambda, etc.)

How to Use​

  1. Filter by State: Click StatsCards at the top to filter by alarm state
  2. Check Namespace: Identify services with many alarms from the bar chart
  3. View Details: Click an alarm row to view configuration and actions
  4. Refresh: Click the button in the top right to fetch latest state
Alarm State Meanings
  • OK: Metric is within threshold
  • ALARM: Metric exceeds/falls below threshold (depending on configuration)
  • INSUFFICIENT_DATA: Insufficient metric data or alarm just created

Usage Tips​

Check ALARM State Immediately​

If the red "ALARM" StatsCard shows "Active alarms!", immediate attention is required.

Verify Action Configuration​

If Actions Enabled shows "No" in alarm details, no notifications will be sent when the alarm triggers. Verify that an SNS topic or Lambda function is connected.

Resolving INSUFFICIENT_DATA​

  • Newly created alarms: Wait for metric collection (up to 5-10 minutes)
  • Existing alarms: Check metric source (EC2 stopped, Lambda inactive, etc.)
Alarm Evaluation Periods

For an alarm to enter ALARM state, the threshold must be exceeded for consecutive evaluation periods. Example: Period 300s, Eval Periods 3 = alarm triggers after 15 minutes of continuous threshold breach.

AI Analysis Tips​

Example questions using the Monitoring Gateway in AI Assistant:

  • "Analyze the common causes of alarms in ALARM state"
  • "Show alarm state change history for the last 24 hours"
  • "Analyze whether this alarm threshold is appropriate"
  • "Would it be better to use SNS instead of Lambda for alarm actions?"