EC2 t2.micro实例（Ubuntu16.04.3）自行断连，重启后复现故障求助

Environment Details #

• EC2 Instance Type: t2.micro
• Operating System: Ubuntu 16.04.3
• Configuration Tool: Webmin

Problem Description #

My EC2 instance shows as "running" in the console, but I can't connect to it properly. When I right-click the instance → "Instance State" → "Stop"/"Start" to restart it, I can connect normally again. However, after some time, the issue recurs.

When trying to connect, I get this error:

This site can’t be reached 18.217.7.112 refused to connect.

Restarting fixes it temporarily, but the failure happens again. I'm not sure where to start troubleshooting or what to check. Any help would be greatly appreciated!

Hey Marius, sorry to hear you're stuck in this frustrating cycle with your EC2 instance. Let's walk through some targeted checks to figure out why this keeps happening:

1. Monitor CPU Credits and Resource Usage #

t2.micro instances rely on burstable CPU credits—once those run out, the instance can become unresponsive or slow to the point of being unreachable:

• Head to the EC2 Console, select your instance, and go to the Monitoring tab. Look at the CPU Credit Balance metric over the period when the instance stopped responding. If it drops to near zero, that's a clear sign of credit exhaustion.
• While the instance is working, run top or htop in the terminal to identify processes hogging CPU or memory. Webmin itself might not be the issue, but a service it manages (like a web server, database, etc.) could be leaking resources over time.

2. Verify Network and Firewall Rules #

Sometimes connections get blocked due to misconfigured or changing network rules:

• Double-check your EC2 Security Group rules: ensure ports 22 (SSH) and Webmin's default port 10000 are open to your IP address (or the necessary range). Security groups don't change automatically, but it's worth confirming no accidental edits were made.
• On the Ubuntu instance, run sudo ufw status to check the local firewall. If enabled, make sure the required ports are allowed. Also, check if any cron jobs or scripts are modifying firewall rules periodically (look in /etc/cron.d/ or user crontabs).

3. Dig Into Logs for Errors #

Logs will give you specific clues about what's failing:

• Webmin logs are typically stored in /var/webmin/miniserv.log or /var/log/webmin/. Look for error messages around the time the connection failed—this could show if Webmin crashed or lost network access.
• Check system-wide logs with sudo journalctl -xe to spot out-of-memory kills, service crashes, or system-level issues that might be taking down SSH or network services.

4. Check if Critical Processes Are Running #

If Webmin or SSH crashes and doesn't auto-restart, you'll lose connectivity:

• When the instance is unresponsive (after restarting and waiting for the issue to hit), try connecting via EC2 Instance Connect (if enabled) to run ps aux | grep webmin and ps aux | grep sshd. If either process is missing, set up a systemd service to auto-restart them on failure.
• For Webmin, you can create a simple systemd unit file to ensure it restarts automatically if it crashes.

5. Address Ubuntu 16.04's End-of-Life Status #

Ubuntu 16.04 reached end-of-life in April 2021, meaning it no longer receives security updates or bug fixes. This can lead to unexpected instability:

• If possible, consider upgrading to a newer LTS version like 20.04 or 22.04. Old, unsupported distros often have unresolved bugs that cause intermittent issues.
• If upgrading isn't an option, you can enable the Ubuntu old-repositories to apply any remaining fixes (note: this isn't ideal for security, but might resolve stability issues). Run sudo sed -i 's/archive.ubuntu.com/old-releases.ubuntu.com/g' /etc/apt/sources.list then sudo apt update && sudo apt upgrade.

6. Check EC2 Instance Status Checks #

Occasionally, the underlying hardware can cause issues:

• In the EC2 Console, check the Status Checks section for your instance. If it shows any system status failures, restarting the instance moves it to new hardware, but if the issue recurs, you might need to launch a new instance from an AMI of your current setup.

Hopefully one of these steps helps you pinpoint the root cause. If you find any unusual logs or metric spikes, feel free to share more details!

内容的提问来源于stack exchange，提问作者Marius