I’ve recently started to deploy a new server infrastructure which has been in planning for a considerable amount of time. One aspect I failed to adequately research is DNS - or more specifically, the way browsers handle multi-IP DNS records.

My scenario:
Each domain name has a designated primary web server which should be used for all requests for that domain by default. If the primary web server IP is unreachable, a designated secondary server IP should be used instead.

In the past I assumed this was as simple as listing the primary server IP in the DNS record first, the secondary IP second, etc. But I’ve now found out (the hard way) that this isn’t the case - web browsers randomly pick IPs from the DNS records.

Is there anyway to accomplish IP priorities easily? At the moment I’m considering the following less-than-ideal workarounds:

a.) Set up a dedicated name-server for each web-server that returns the IP of the respective web-server. Set the name-servers for each domain in the order that prioritizes the respective web-server.

or b.) As soon as the primary web-server dies, dynamically replace the primary IP in the DNS record with the secondary web server IP.