Alibaba Cloud DNS:TCP health check

Overview

Global Traffic Manager (GTM) uses TCP health checks to monitor target IP addresses for network reachability, port availability, and latency. When an IP address or port becomes unhealthy, GTM automatically stops routing traffic to it and resumes when it recovers.

Procedure

1. Log in to the Alibaba Cloud DNS - Global Traffic Manager console.

2. On the instance list page, find the target instance and click Actions in the Configure column.

3. On the Basic Configuration page, click the Address Pool Configuration tab. Click the + icon next to the target address pool, and then click Add next to Health Check.

4. In the Modify Health Check dialog box, set the health check protocol, health check interval, health check port, timeout, consecutive failures, failure ratio, and monitoring nodes, and click Confirm.

   Example: Set health check protocol to TCP, health check interval to 1 minute, health check port to 80, timeout to 5 seconds, consecutive failures to 2, and failure ratio to 50%. For monitoring nodes, select the BGP nodes in China (Zhangjiakou), China (Qingdao), China (Hangzhou), China (Shanghai), China (Hohhot), China (Shenzhen), and China (Beijing).

Parameters

1. Health check protocol

   Select TCP. GTM monitors the target IP address for network reachability, port availability, and latency over TCP.

2. Health check interval

   The interval between health checks. Default: 1 minute. Ultimate Edition supports a minimum interval of 15 seconds.

3. Health check port

   The port to monitor on the target IP address. GTM establishes a TCP connection to this port. A successful connection means the service is healthy; a failed connection means it is unhealthy.

4. Timeout

   The maximum wait time for a health check response. If no response is received within this period, the check times out. Valid values: 2, 3, 5, or 10 seconds.

5. Consecutive failures

   The number of consecutive failures before GTM marks a service as unhealthy. Prevents false alarms from transient network issues. Valid values: 1, 2, or 3.

• 1: Marks the service as unhealthy after one alert.

• 2: Marks the service as unhealthy after two consecutive alerts.

• 3: Marks the service as unhealthy after three consecutive alerts.

6. Failure ratio

   The percentage of monitoring nodes that must report a failure before the service is marked unhealthy. Valid values: 20%, 50%, 80%, and 100%.

7. Monitoring node

   The nodes that perform TCP health checks. The following default nodes are available.