Hi all,
We have two servers, a Primary and a Secondary, which work in a failover scenario. We run on the primary and the secondary is ready to take over in the event of a failure.
We have other servers which send traffic to these servers. Some of them are only capable of sending traffic to a single IP address, so to avoid manual configuration in a failover event, we have configured NLB to send this traffic to a single address. This is configured in Single Host mode where primary has priority 1 and secondary has priority 2.
My question is, when traffic is sent to the NLB cluster address, will it look at the individual port on the primary server to see if it is open and if not, send it to the secondary, or will it only look to see if the primary is online.
Eg if we had multiple ports, say 8000 and 9000, that we were sending data on, and the primary stopped listening on 8000, but was still listening on port 9000, should it direct the traffic on 8000 to the secondary while the traffic on 9000 continued to go to the primary?
Or another example, lets say the server stopped listening on all ports because the services stopped, but the server itself was still running so the NLB address was still available, would it just try to send the traffic to the Primary as it still thinks it is online, even though it is not listening on those ports?
We had this set up on some Server 2012 R2 servers and I am sure this was tested and working where traffic would go to the secondary if the primary was not able to receive it, even if it was online. However we have just upgraded to new Server 2019 servers and I am finding that the NLB is only failing over to the secondary if I actually disable the NLB interface on the primary so the NLB cluster sees it as offline. It is possible I am mistaken as to how it was working on the old servers, as it is very rare that we need to failover.
Is this how it is supposed to work? Or should traffic go to the first server that can receive it, not just the first server that is online in the cluster?