Hi Zhengyu, I'm not sure I understand your question, so I'm going to take a guess. Correct me if I am answering the wrong question. First question I have is are you using the EOL neutron-lbaas or Octavia? I will assume you are using Octavia. When you add a member server (backend web server for example), you have a few options: 1. If you create a member without the "subnet_id" option, the load balancer will attempt to route to the member IP address over the VIP subnet. Health monitor checks will also follow this route. 2. If, when you create the member, you specify the "subnet_id" option to a valid neutron subnet, the load balancer will be attached to that subnet and will route to the member IP address. If you do not specify a "monitor_address", health monitoring will follow the same route as the member IP address. 3. If you create a member, with the "monitor_address" specified, traffic will be routed to the member IP address, but health monitoring checks will be directed to the "monitor_address". To give an example: Say you have a neutron network 436f58c2-0454-49dc-888e-eaafdd178577 with a subnet of e6e46e02-7768-4ae4-89c6-314c34557b5d with CIDR 100.64.0.0/14 on it. When creating the pool member is created you would specify something like: openstack loadbalancer member create --address 100.64.100.5 --subnet-id e6e46e02-7768-4ae4-89c6-314c34557b5d --protocol-port 80 <pool ID> This will attach the neutron network 436f58c2-0454-49dc-888e-eaafdd178577 to the load balancer and allocate an IP address on 100.64.0.0/14 that will be used to contact the member server address of 100.64.100.5. Health monitor checks will also follow this same path to the member server. We have some documentation for this in the cookbook here: https://docs.openstack.org/octavia/latest/user/guides/basic-cookbook.html#de... I hope this helps clarify, Michael On Tue, Jul 7, 2020 at 12:45 AM Zhengyu Pan <zhengyupann@163.com> wrote:
There are some private cloud or public cloud introduction: They use 100.64.0.0/14 network segments to check vm's health status in load balancer. In Region supporting VPC, load balancing private network IP and health check IP will be switched to 100 network segment. I can't understand how to implement it. How to do it?
--