Alibaba Cloud DNS:Ping health check

What is a ping health check?

A ping health check is a method that Global Traffic Manager (GTM) uses to check the health of your application services. This method is based on the Internet Control Message Protocol (ICMP). When you configure an address pool, you can create a ping health check to monitor each IP address in the pool. If an IP address becomes unhealthy, GTM automatically stops routing traffic to it. When the IP address becomes healthy again, GTM resumes routing traffic to it.

Configuration parameters

1. Template Name: The name of the health check template. For easy identification, use a name that indicates the health check protocol.

2. Probe Type: The IP address type of the probe nodes. Valid values are IPv4 and IPv6.

3. Probe Protocol

   Select Ping to monitor metrics such as network reachability, packet loss rate, and latency for the target IP address.

4. Number Of ICMP Packets To Send

   The number of ICMP data packets to send for each ping health check. This value is used to calculate the packet loss rate and is fixed at 20.

5. Packet Loss Rate Threshold

   The packet loss rate is calculated for the ICMP data packets sent during each ping health check. The formula is: Packet loss rate = (Lost data packets / Total ICMP data packets sent) × 100%. An alarm is triggered if the packet loss rate reaches this threshold. Valid values: 10%, 30%, 80%, 90%, and 100%.

6. Health Check Interval

   The interval between ping health checks. The default interval is 1 minute. The minimum interval is 15 seconds and is available only for Ultimate Edition instances.

7. Timeout

   For each ping health check, the system calculates the response time for the sent ICMP data packets. If a packet does not return within the specified timeout period, the health check is considered to have timed out. Valid values: 2 seconds, 3 seconds, 5 seconds, and 10 seconds.

8. Retry Count

   The number of consecutive failed health checks before the application service is considered abnormal. This setting prevents false alarms caused by transient network jitter. Valid values: 1, 2, and 3.

   • 1: The application service is considered abnormal after one failed health check.

   • 2: The application service is considered abnormal after two consecutive failed health checks.

   • 3: The application service is considered abnormal after three consecutive failed health checks.

     Important

     If the target address is disconnected from the network or returns an `ICMP destination unreachable` message, the check is not counted as a failure and no alarm is triggered. You must fix the network issue for the address or switch to an HTTP health check task.

7. Probe Nodes

   The geographic locations of the nodes that perform the ping health checks. The system provides the following default probe nodes:

   Node type	Geographic location
   BGP node	Qingdao, Zhangjiakou, Beijing, Hohhot, Shenzhen, Shanghai, Hangzhou
   International node	Malaysia, Germany, China (Hong Kong), Singapore, California, Japan
   Carrier node	Changsha Telecom, Xi'an Telecom, Nanjing Unicom, Zhengzhou Telecom, Tianjin Unicom, Nanjing Mobile, Dalian Unicom, Shenzhen Mobile, Dalian Mobile, Qingdao Telecom

   Important

   • If all addresses in the address pool are Alibaba Cloud addresses and you use a blackhole filtering policy for fault testing, select carrier nodes as the probe nodes. This is because blackhole filtering is an access control list (ACL) policy that takes effect on the Internet between the Alibaba Cloud network and carrier networks. Traffic between Alibaba Cloud IP addresses mostly flows within the Alibaba Cloud network, which reduces the effectiveness of the probes.

   • Difference between BGP nodes and carrier nodes: Border Gateway Protocol (BGP) nodes automatically select the optimal carrier network line. For example, if you use a BGP node in Shanghai and the China Mobile line fails, the probe node may switch to the China Telecom line. This switch occurs unless all lines fail. In contrast, a carrier node uses only its specified carrier network and does not have this optimization mechanism.

   • To configure a whitelist for access sources on your server, click View Probe Node IP Information to obtain the IP addresses of the probe nodes.

   • If the IP address to be probed is located outside mainland China, select international nodes.

8. Probe Failure Rate Of Selected Nodes

   The percentage of selected probe nodes that must report a health check failure before the application service is considered abnormal. Valid values: 20%, 50%, 80%, and 100%.

Configuration

Create a probe template

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

2. Click the Probe Templates tab, and then click Create Probe Template. In the dialog box that appears, configure the parameters and click OK.

   Note

   To create more probe templates, click Add And Continue.