AWS Knowledge Hub

English

Tiếng Việt

English

Tiếng Việt

Appearance

EC2 Troubleshooting ​

Summary ​

  • Most EC2 issues fall into a handful of categories: launch failures, connectivity, performance, storage/network, cost, and status check failures.
  • Effective troubleshooting starts with a standard checklist (state, network/Security Group/route, logs, CloudWatch metrics) before making configuration changes.
  • Strong monitoring and logging (CloudWatch, VPC Flow Logs, Systems Manager) dramatically reduce mean time to resolution.

Troubleshooting flow ​

mermaid
flowchart TD
  A[EC2 issue] --> B{Category?}
  B --> C[Launch failure]
  B --> D[Cannot connect]
  B --> E[Performance problem]
  B --> F[Storage/Network issue]
  B --> G[Cost anomaly]

  C --> C1[Check limits, AZ capacity, instance type]
  D --> D1[Check SG/NACL/Route, IP, key/user]
  E --> E1[Check CPU/Mem/IO; right-size/ASG]
  F --> F1[Check EBS type, IOPS, network metrics]
  G --> G1[Use Cost Explorer; find idle resources & data transfer]

Best Practices ​

  • Always verify the basics first: instance state, correct Security Group rules, route tables, NACLs, DNS/IP, and correct SSH/RDP user and key.
  • Use CloudWatch metrics and logs to pinpoint root causes (high CPU, I/O bottlenecks, network throttling, status check failures) instead of guessing.
  • For status check failures:
    • System status → underlying host issue → stop/start to move to a new host, check AWS Health.
    • Instance status → OS/app/network issue inside the instance → inspect logs and configuration.
  • For performance issues, prioritize right‑sizing, code/DB optimization, appropriate EBS volume and instance type selection, and enhanced networking/placement groups where applicable.
  • For cost anomalies, look for always‑on instances, unattached EBS volumes, idle Elastic IPs, old snapshots, and unexpected data transfer patterns.
  • Document root causes and fixes, then improve monitoring/alerting or configuration to prevent recurrence.

Exam Notes ​

  • Many questions describe “cannot SSH/RDP” or “no internet”; the expected approach is to walk through the basic connectivity checklist (state, SG, NACL, route, IP, key/user).
  • Know the difference between System status checks (host issues, fixed by stop/start) and Instance status checks (OS/app/network issues, fixed by configuration changes).
  • Cost‑related troubleshooting often involves identifying idle resources (running instances, unattached EBS, unused Elastic IPs, old snapshots) and high data transfer patterns.
  • Key tools to remember: CloudWatch (metrics/logs), VPC Flow Logs, Systems Manager, EC2 Instance Connect, AWS Health Dashboard, Cost Explorer, AWS Budgets.

AWS documentation ​