Solutions:

1. Check Service Quotas:

   # In AWS Console, check:
   Service Quotas → Amazon VPC → VPCs per region (need at least 1 available)
   Service Quotas → Amazon EC2 → NAT gateways per AZ (need 2 for HA)
   Service Quotas → Amazon EC2 → Elastic IPs (need at least 2)

2. Verify IAM Permissions: Ensure the IAM user has AmazonEKSClusterPolicy and AmazonEKSVPCResourceController
3. Try Different Region: Some regions have limited capacity. Try us-east-1, us-west-2, or eu-west-1
4. Check Logs: View cluster creation logs in SeaGit for specific error messages

Solutions:

1. Verify Image Exists:

   # Test pull locally
   docker pull your-image:tag
   
   # Common mistakes:
   nginx:latest        ✓ Correct
   nginx:1.21          ✓ Correct  
   nginx               ✗ Missing tag
   nginxinc/nginx      ✗ Wrong org
   nginx:latests       ✗ Typo in tag

2. Add Registry Credentials: If using private registry:
   • • Navigate to Providers → Add Docker Registry provider
   • • Enter username and password/token
   • • SeaGit will create imagePullSecret automatically
3. Docker Hub Rate Limit: Use authenticated pulls:

   # Add Docker Hub credentials (gets 200 pulls/6hrs)
   # OR use private registry/mirror to avoid limits

4. Check Image Pull Logs:

   kubectl describe pod <pod-name>
   # Look at Events section for specific error

Solutions:

1. Check Container Logs:

   # In SeaGit, view deployment logs
   # OR use kubectl:
   kubectl logs <pod-name> --previous
   # --previous shows logs from crashed container

2. Common Fixes by Error Type:

   Error: "Cannot connect to database"

   Check DATABASE_URL variable and database connectivity

   Error: "Missing required environment variable"

   Add the variable at application or instance level

   Error: "Port 3000 already in use"

   Ensure container port matches PORT env variable

   Error: "ENOENT: no such file"

   Check file paths - may need absolute paths or working directory

3. Test Locally: Run same image locally with same env variables to reproduce issue
4. Add Startup Delay: If app needs time to initialize, increase health check initialDelaySeconds

Solutions:

1. Increase Memory Limit:
   • • Go to Application → Edit → Resources
   • • Increase memory limit (e.g., 512Mi → 1Gi)
   • • Redeploy application
2. Profile Memory Usage:

   # Check current usage
   kubectl top pods
   
   # If usage is near limit, definitely need increase
   # If usage is low but still OOMKilled, may be spike/leak

3. Check for Memory Leaks:
   • • Use profiling tools (Node.js: --inspect, Python: memory_profiler)
   • • Look for growing memory over time
   • • Common causes: unclosed connections, large caches, circular references
4. Enable Horizontal Pod Autoscaling: Spread load across more pods instead of increasing single pod memory

Solutions:

1. Check Pod Health:

   kubectl get pods
   # All pods should show Running and 1/1 ready
   
   kubectl logs <pod-name>
   # Check for application errors

2. Fix Health Check Endpoint:
   • • Ensure /health endpoint returns 200 status
   • • Check health endpoint doesn't depend on external services
   • • Test: curl http://<pod-ip>:<port>/health
3. Increase Timeouts: If application is slow:
   • • Edit application configuration
   • • Increase health check timeout and interval
   • • Consider increasing ALB idle timeout (default 60s)
4. Check Target Group:
   • • In AWS Console, go to EC2 → Target Groups
   • • Find target group for your ingress
   • • Check targets tab - should show healthy targets

Solutions:

1. Check Build Logs: View detailed error in deployment logs
2. Common Issues by Language:

   Node.js:

   • • Missing package.json or package-lock.json
   • • Node version mismatch (specify in package.json: "engines": {"node": ">=18"})
   • • npm install failures (check dependencies)

   Python:

   • • Missing requirements.txt
   • • Dependency conflicts or unavailable packages
   • • Python version issues (use runtime.txt to specify)

   Go:

   • • Missing go.mod file
   • • Build errors (fix in code)
   • • Private module dependencies (need git credentials)
3. Use Dockerfile Instead: If buildpack auto-detection fails, create Dockerfile:

   # Add Dockerfile to repo root
   FROM node:18
   WORKDIR /app
   COPY package*.json ./
   RUN npm install
   COPY . .
   EXPOSE 3000
   CMD ["npm", "start"]