非Web应用服务器的AWS ALB简易健康检查方案咨询

Alright, let's tackle your two questions step by step—first configuring AWS ALB health checks for your backend servers, then a simple solution for those non-web servers without HTTP services.

If your backend servers run web services (HTTP/HTTPS), here's how to set up reliable health checks via the AWS Console or CLI:

Console Setup #

• Navigate to the EC2 Dashboard → Target Groups and select the target group linked to your ALB.
• Click Edit health check to adjust these key settings:
  • Protocol: Pick HTTP or HTTPS to match your server's service.
  • Path: Use a dedicated health endpoint like /health (preferred) or a simple path like /—ensure your server returns a 200-399 status code for this path.
  • Port: Choose "Traffic port" to use the same port as your application, or specify a fixed port if needed.
  • Thresholds: Set Healthy threshold count (e.g., 2 successful checks to mark as healthy) and Unhealthy threshold count (e.g., 3 failed checks to mark as unhealthy).
  • Timings: Adjust Health check timeout (e.g., 5 seconds) and Health check interval (e.g., 10 seconds) based on your server's response speed.
  • Success codes: Define valid HTTP status codes (default is 200, but you can expand to 200-399 if needed).
• Save the changes, and the ALB will start periodically verifying your servers' health—unhealthy instances get removed from traffic rotation automatically.

CLI Setup #

If you prefer command-line configuration, use this example (replace placeholders with your actual values):

aws elbv2 modify-target-group \
  --target-group-arn arn:aws:elasticloadbalancing:us-east-1:123456789012:targetgroup/my-target-group/abc123 \
  --health-check-protocol HTTP \
  --health-check-port traffic-port \
  --health-check-path /health \
  --healthy-threshold-count 2 \
  --unhealthy-threshold-count 3 \
  --health-check-timeout-seconds 5 \
  --health-check-interval-seconds 10 \
  --matcher HttpCode=200

Since your servers don't have HTTP services, we'll focus on lightweight, no-fuss options:

Option 1: TCP Port Health Check (Easiest) #

ALB supports TCP-based health checks, which only require your server to listen on a specific port—no extra software needed.

• On your non-web server, set up a persistent TCP listener. For example, use nc (netcat) with a systemd service to keep it running:
  1. Create a service file /etc/systemd/system/tcp-healthcheck.service:

     [Unit]
     Description=TCP Health Check Listener
     After=network.target
     
     [Service]
     ExecStart=/usr/bin/nc -l 10000
     Restart=always
     
     [Install]
     WantedBy=multi-user.target
     

  2. Enable and start the service:

     sudo systemctl daemon-reload
     sudo systemctl enable --now tcp-healthcheck.service
     

• Back in your ALB target group, edit the health check to use:
  • Protocol: TCP
  • Port: 10000 (or whatever port you chose)
  • Adjust thresholds and timings as needed.
• The ALB will attempt to establish a TCP connection to this port—successful connections mean the server is healthy; failed connections trigger removal from the target group.

Option 2: Custom Health Checks (For Deep Verification) #

If you need to check more than just an open port (e.g., if a specific process is running, disk space is sufficient), use a script + CloudWatch:

1. Create a simple health check script (/usr/local/bin/health-check.sh):

   #!/bin/bash
   # Check if your non-web service process is running
   if pgrep "my-critical-service" > /dev/null; then
       # Report healthy status to CloudWatch
       aws cloudwatch put-metric-data --namespace "NonWebServers" --metric-name "HealthStatus" --value 1 --dimensions InstanceId=$(curl -s http://169.254.169.254/latest/meta-data/instance-id)
       exit 0
   else
       # Report unhealthy status
       aws cloudwatch put-metric-data --namespace "NonWebServers" --metric-name "HealthStatus" --value 0 --dimensions InstanceId=$(curl -s http://169.254.169.254/latest/meta-data/instance-id)
       exit 1
   fi
   

2. Make it executable: sudo chmod +x /usr/local/bin/health-check.sh
3. Set up a cron job to run it every minute:

   * * * * * /usr/local/bin/health-check.sh
   

4. Create a CloudWatch Alarm that triggers when HealthStatus is 0 for 2 consecutive minutes. Link this alarm to a Lambda function that removes the unhealthy instance from your ALB target group.

This approach gives you more control over what defines a "healthy" server, beyond just an open port.

内容的提问来源于stack exchange，提问作者Rohit Sharma