EC2 Instance Freeze

Based on the symptoms you've described, there are several possible causes for your EC2 instance freeze and several approaches to troubleshooting and resolving this issue.

When an EC2 instance becomes completely unresponsive with no obvious errors in system logs and normal resource utilization metrics, it could indeed point to an underlying host or hypervisor-level issue. Here are some steps you can take to diagnose and address this problem:

1. Use EC2 Serial Console: This tool allows you to access your instance even when it's unresponsive to network connections. It can help you diagnose issues when the instance is frozen by providing direct console access to see what's happening at the system level.

2. Stop and Start the Instance (not just reboot): While you mentioned that a reboot resolved the issue temporarily, a full stop and start would move your instance to different underlying hardware. This can help determine if the issue is related to the physical host. Unlike a reboot, which keeps the instance on the same host, a stop/start operation migrates your instance to new hardware.

3. Create a Replacement Instance: If the issue persists, you can create an Amazon Machine Image (AMI) from your instance and launch a new instance using this AMI. This approach is particularly useful if the problem is specific to the current instance rather than your configuration or application.

4. Enhanced Monitoring: Set up CloudWatch Enhanced Monitoring at a more frequent interval (e.g., 1-minute) to capture detailed metrics for memory utilization and disk I/O. Standard CPU utilization metrics might not show the complete picture.

5. Check for Steal Time: When the system is responsive, run the top command and check the "st" value. High steal time (above 5%) can indicate resource contention at the hypervisor level, which might explain the freezes.

For future prevention:

1. Implement more comprehensive monitoring to catch early warning signs.
2. Consider setting up automated recovery actions based on status checks.
3. Design your architecture with redundancy to minimize impact if a single instance becomes unresponsive.

While it's possible the issue could be related to an AWS outage from the previous day, without specific information about that outage and its affected services, it's difficult to establish a direct connection. If you continue to experience these issues, you may want to open a support case with AWS for further investigation of the underlying infrastructure.
Sources
Rocky 9 Instance c7i.4xlarge freezes randomly | AWS re:Post
Troubleshoot Amazon EC2 instance stop issues - Amazon Elastic Compute Cloud
We noticed sudded spike in CPU usage (1% to 71%) and the machine stopped responding. | AWS re:Post
Troubleshoot Amazon EC2 instance termination issues - Amazon Elastic Compute Cloud

Hi,

Have a look at last posts of https://health.aws.amazon.com/health/status

You're probably right: your issue was probably generated by the us-east-1 failure that severely affected EC2 service

Didier