We have api gateway setup on Azure Cloud and we are currently using Azure Internal Load Balancer for distributing the traffic across api gateways. We want to setup a health check probe which will take the unhealthy gateway node out of rotation automatically. Azure ILB provides tcp and http probe. I would like to know which healthcheck probe would be better choice.
A tcp probe will only give you a very basic gateway up/down check. It doesn't show if the gateway is actually processing messages. So I would go for the http probe. I have posted a sample policy you can use for an http probe here: Load Balancer Status Service
I would go with MichielHelder's advice. You have some more control about taking down separate nodes and message processing is checked. The PING URI test (also HTTP) from docops is more basic and does not necessarily indicate normal message processing.
This is helpful. Many thanks.