I need some help on understanding the way azure Load balancer forwards requests to backend VMs that have failed responding to health probes BUT have recovered after sometime (say after 30,60 minutes).
The standard Azure load balancer has support for health probes using TCP, HTTP and HTTPS protocols. I understand that it has support for minimum tries and duration after which if a VM does NOT respond it is marked as unresponsive. My understanding is that once a VM is declared Unresponsive it does NOT forward ny new flows however allows existing esablished connections to pass through.
But i am not sure whether Load balancer keeps checking on an iterative mode on whether a failed VM is up and running. I like to know how the Azure Load balancer detects whether the backend VM is ONCE AGAIN up or running ?
Is there a way to setup any probe to keep checking whether the backend VM is responding for a extended duration of 30,60 minutes ?
it sends probes all the time (even if the vm is down\stopped\etc) as soon as the vm starts responding to probes it will be marked as healthy and it will start forwarding traffic to it.