如何配置CloudWatch Alarm在EC2实例uptime超5小时时触发告警？

Hey there! Since you're new to AWS CloudWatch, let's break this down into simple, actionable steps. The catch here is that AWS doesn't provide a native "uptime" metric for EC2 instances out of the box, so we'll need to either collect system-level uptime data or calculate it using the instance's launch time. Let's cover both approaches:

Approach 1: Collect Uptime via CloudWatch Agent (Direct System Metric) #

This method uses the CloudWatch Agent to send your EC2 instance's actual system uptime to CloudWatch, then builds an alarm around that metric.

Step 1: Install and Configure the CloudWatch Agent #

• Connect to your EC2 instance via SSH (or RDP for Windows instances).
• For Linux-based instances, run the appropriate install command:

  # For Amazon Linux/RHEL
  sudo yum install amazon-cloudwatch-agent -y
  # For Ubuntu/Debian
  sudo apt-get install amazon-cloudwatch-agent -y
  

• Launch the configuration wizard to set up metric collection:

  sudo /opt/aws/amazon-cloudwatch-agent/bin/amazon-cloudwatch-agent-config-wizard
  

• Follow the wizard prompts:
  • Select EC2 as the host type.
  • Choose to store the configuration in AWS Systems Manager Parameter Store (simpler for future management).
  • When selecting metrics to collect, enable System metrics and ensure the uptime metric is included (under Linux/Windows system metrics sections).
  • Complete the wizard, then start the agent:

    sudo /opt/aws/amazon-cloudwatch-agent/bin/amazon-cloudwatch-agent-ctl -a fetch-config -m ec2 -c ssm:your-parameter-store-name -s
    

Step 2: Create the CloudWatch Alarm #

• Navigate to the AWS CloudWatch Console → Alarms → All alarms → Create alarm.
• Select Select metric → Search for the CWAgent namespace → Choose your instance's dimension → Select the uptime metric.
• Set the Statistic to Maximum (since uptime is a cumulative value) and Period to 5 minutes (adjust based on how frequently you want to check).
• Under Conditions:
  • Threshold type: Static
  • Whenever uptime is: Greater than
  • Threshold value: 18000 (5 hours converted to seconds: 56060)
• Configure notifications: Select an existing SNS topic or create a new one to send alerts (email, SMS, etc.).
• Name your alarm (e.g., "EC2-Uptime-Exceeds-5Hours"), add a clear description, then click Create alarm.

Approach 2: Calculate Uptime Using Instance Launch Time (No Agent Required) #

If you prefer not to install the CloudWatch Agent, you can use CloudWatch Metric Math to calculate uptime based on the instance's launch time metadata.

Step 1: Build a Metric Math Alarm #

• Go to CloudWatch Console → Alarms → Create alarm → Select metric.
• Under All metrics, navigate to EC2 → Per-Instance Metrics → Select any metric tied to your instance (e.g., CPUUtilization—we just need the instance dimension).
• Click Add math → Start with empty expression.
• Enter this Metric Math expression (replace i-1234567890abcdef0 with your actual instance ID):

  TIME_SINCE(GET_METRIC_DATA(SEARCH('{AWS/EC2,InstanceId} InstanceId="i-1234567890abcdef0"', 'Maximum', 300)), 'SECONDS')
  

  This formula calculates the time elapsed since the instance launched.
• Set the Period to 5 minutes.
• Under Conditions:
  • Threshold type: Static
  • Whenever math expression is: Greater than
  • Threshold value: 18000 (5 hours in seconds)
• Configure your notification preferences, name the alarm, and click Create alarm.

Key Notes #

• For Approach 2: The calculation resets if you stop and start the instance (since a stopped/started instance gets a new launch time).
• For Approach 1: Ensure your EC2 instance has the AmazonCloudWatchAgentServerPolicy managed IAM policy to send metrics to CloudWatch.

内容的提问来源于stack exchange，提问作者user12417145