Started my Project on Energy-Efficient DevOps: Auto-Suspending Idle AWS Resources using ML

Week 2– Automation Phase (Auto-Suspend Script Deployment)

Objective:

To automate EC2 instance suspension when the machine learning model predicts an idle state, using boto3, instance metadata, and cron jobs.

---

What I Did Today:

Transferred Trained Model to EC2

• Uploaded model.pkl to Node-1 using scp.

• Ensured it’s readable by the predict_and_suspend.py script.

Wrote & Deployed Prediction Script

A Python script (predict_and_suspend.py) that:

• Reads the latest CPU usage from cpu_log.csv

• Loads the trained ML model (model.pkl)

• Predicts if the system is idle (CPU < threshold)

• If idle, stops the instance using boto3 and the EC2 metadata endpoint to fetch the instance ID

Solved IAM Permissions

• Attached an IAM Role with AmazonEC2FullAccess to allow the instance to stop itself using boto3.

• Verified no need to store or configure ~/.aws/credentials manually.

Scheduled the Script Using Cron

• Used crontab -e to schedule predict_and_suspend.py every 15 minutes.

• Verified that the script shuts down the instance when idle is predicted.

Fixed Metadata Access Error

• Instance metadata was initially blocked.

• Enabled it via EC2 > Actions > Modify metadata options

• Set IMDSv2 to “Required” and allowed ip access.

  All Commands Used Today and Their Purpose

---

Problems Faced + Solutions

1. Instance Metadata Access Failed

Cause: Metadata service was blocked or IMDSv2 required
Solution: Enabled IMDSv2 via EC2 settings → Instance Settings → Modify metadata options

2. boto3: NoCredentialsError

Cause: No IAM Role attached to instance
Solution: Created IAM Role with EC2 permissions and attached to instance

3. Instance ID '' is malformed

Cause: Couldn't fetch instance ID from metadata
Solution: Enabled metadata access + confirmed that request returns correct value

4. DeprecationWarning: datetime.utcnow()

Cause: Python is moving to timezone-aware datetime
Solution: Replaced with datetime.datetime.now(datetime.UTC) (resolved warning)

---

Setup Status

• predict_and_suspend.py working

• Cron job running every 15 mins

• Instances now self-terminate if idle

• IAM Role + metadata access properly configured

---

Screenshots:

Tags

#AWS #DevOps #MachineLearning #CloudComputing #Python #EnergyEfficiency #Boto3 #CronJobs #Automation #BuildInPublic

---

Want to Follow Along?

I’ll be sharing weekly progress — issues, logs, architecture, and ML models.

If you've solved similar problems (like automated cloud optimization), I’d love to hear your insight.