This project was built to investigate a client issue. Their CloudWatch alarms stayed stuck in the ALARM state. Even after the triggering condition had passed. It especially happened with low-threshold alarms. The goal was to replicate this behavior and explore how ECS Fargate and CloudWatch interact.
The setup uses ECS Fargate to run a simple containerized app. The service is deployed into a basic VPC using Terraform. Everything is managed in code. i.e. the VPC, the ECS service, IAM roles, and logging.
CloudWatch is used for both logging and monitoring. The container sends its logs to a log group, and CloudWatch Container Insights is enabled. This collects metrics like CPU and memory usage directly from the ECS tasks.
This project is useful if you are trying to troubleshoot CloudWatch alarm behaviour, or if you want a simple, real-world example of ECS Fargate working with CloudWatch. It is also a great reference for setting up ECS Fargate infrastructure using Terraform.
All infrastructure is lightweight. It is designed for quick testing and low-cost monitoring. The default setup uses an Amazon Linux base image with stress testing enabled, so it simulates activity for the alarms to monitor.
By deploying this, you get hands-on insight into how alarms behave with short-lived or fluctuating workloads. You can also practice tuning CloudWatch alarms and checking how quickly state transitions happen in real time.
Overall, this is a small but effective example of observability in action. It shows how ECS and CloudWatch work together and gives you a place to experiment and learn before applying these techniques to production environments.