100-Days-Of-DevOps-Challenge-KodeKloud

Create and Configure Alarm Using CloudWatch Using Terraform

The Nautilus DevOps team has been tasked with setting up an EC2 instance for their application. To ensure the application performs optimally, they also need to create a CloudWatch alarm to monitor the instance’s CPU utilization. The alarm should trigger if the CPU utilization exceeds 90% for one consecutive 5-minute period. To send notifications, use the SNS topic named nautilus-sns-topic, which is already created.

1. Launch EC2 Instance: Create an EC2 instance named nautilus-ec2 using any appropriate Ubuntu AMI (you can use AMI ami-0c02fb55956c7d316).

2. Create CloudWatch Alarm: Create a CloudWatch alarm named nautilus-alarm with the following specifications:

   • Statistic: Average
   • Metric: CPU Utilization
   • Threshold: >= 90% for 1 consecutive 5-minute period
   • Alarm Actions: Send a notification to the nautilus-sns-topic SNS topic.

3. Update the main.tf file (do not create a separate .tf file) to create a EC2 Instance and CloudWatch Alarm.

4. Create an outputs.tf file to output the following values:

   • KKE_instance_name for the EC2 instance name.
   • KKE_alarm_name for the CloudWatch alarm name.

Notes:

1. The Terraform working directory is /home/bob/terraform.

2. Right-click under the EXPLORER section in VS Code and select Open in Integrated Terminal to launch the terminal.

3. Before submitting the task, ensure that terraform plan returns No changes. Your infrastructure matches the configuration.

Steps

1. Create a variables.tf file with these contents:

    variable "prefix" {
        default = "nautilus"
    }
   

   Replace nautilus according to your task description

2. Create a main.tf file with these contents:

    resource "aws_sns_topic" "sns_topic" {
        name = "${var.prefix}-sns-topic"
    }
   
    resource "aws_instance" "ec2_instance" {
        ami = "ami-0c02fb55956c7d316"
        instance_type = "t2.micro"
   
        tags = {
            Name = "${var.prefix}-ec2"
        }
    }
   
    resource "aws_cloudwatch_metric_alarm" "cw_alarm" {
        alarm_name = "${var.prefix}-alarm"
        comparison_operator = "GreaterThanOrEqualToThreshold"
        evaluation_periods = 1
        metric_name = "CPUUtilization"
        namespace = "AWS/EC2"
        period = 300
        statistic = "Average"
        threshold = 90
   
        dimensions = {
            InstanceId = aws_instance.ec2_instance.id
        }
   
        alarm_actions = [aws_sns_topic.sns_topic.arn]
    }
   

   There is a complete terraform file: main.tf with vpc, subnet, igw, route table, security group, ec2 key pair, sns topic and cloudwatch alarm. You can give it a look.

3. Create an outputs.tf file with these contents:

    output "KKE_instance_name" {
        value = aws_instance.ec2_instance.tags["Name"]
    }
   
    output "KKE_alarm_name" {
        value = aws_cloudwatch_metric_alarm.cw_alarm.alarm_name
    }
   

4. Run the terraform commands:

    terraform init
    terraform plan
    terraform apply -auto-approve
   

Youtube Video

• Watch Youtube Video: https://youtu.be/UED-vxjxXuM

Referrence

• Cloud Watch Alarm
• Click to Start your 100 Days Challenge

Good To Know

CloudWatch Alarm Configuration

• Period vs Evaluation: Period (300s = 5 minutes) is data collection interval, evaluation_periods (1) is how many periods to check
• Threshold Logic: GreaterThanOrEqualToThreshold with threshold 90 means CPU >= 90%
• Statistic Types: Average, Maximum, Minimum, Sum, SampleCount - Average is most common for CPU
• Dimensions: Must match exact case - use InstanceId not instanceId for EC2 instances

EC2 Instance Monitoring

• Default Metrics: Basic monitoring (5-minute intervals) is free, detailed monitoring (1-minute) costs extra
• Common Metrics: CPUUtilization, NetworkIn, NetworkOut, DiskReadOps, DiskWriteOps
• Instance Requirements: CloudWatch agent needed for memory, disk space, and custom metrics
• AMI Selection: Use region-appropriate AMIs - ami-0c02fb55956c7d316 is for us-east-1

SNS Topic Integration

• Alarm Actions: Can trigger SNS, Auto Scaling, EC2 actions, or Systems Manager actions
• Topic ARN: Use aws_sns_topic.name.arn to reference topic ARN in alarm actions
• Notification Types: Email, SMS, HTTP/HTTPS endpoints, Lambda functions
• Multiple Actions: Single alarm can trigger multiple SNS topics or actions

Terraform Best Practices

• Resource Naming: Use consistent naming with variables for reusability
• Default VPC: EC2 instances use default VPC/subnet when not specified
• Output Values: Reference resource attributes like tags["Name"] for dynamic outputs
• Variable Defaults: Set sensible defaults in variables.tf for easier reuse

Monitoring Thresholds

• CPU Utilization: 90% is high threshold for alerts, consider 70-80% for warnings
• Evaluation Periods: Single period (1) for immediate alerts, multiple periods (2-3) to avoid false positives
• Time Windows: 5-minute periods balance responsiveness with stability
• Alarm States: OK, ALARM, INSUFFICIENT_DATA - plan for all states

Common Issues & Solutions

• Missing Metrics: Ensure EC2 instance is running and has proper IAM permissions
• Dimension Mismatch: Verify InstanceId matches exactly (case-sensitive)
• SNS Permissions: CloudWatch needs permission to publish to SNS topic
• Region Consistency: All resources must be in same AWS region

Production Considerations

• Alarm Fatigue: Set appropriate thresholds to avoid too many false alarms
• Escalation: Consider multiple alarm levels (warning, critical, emergency)
• Documentation: Tag resources and alarms for easy identification
• Testing: Verify alarm triggers work by temporarily lowering thresholds

This site is open source. Improve this page.