I recently needed to add a second IP address to a server using a previously unused Ethernet adapter. When I did this, suddenly no workstations could resolve the hostname of that server. When I undid my change, hostname resolution resumed, at least once I reset the first adapter.
Both adapters are using DHCP, and we then use DHCP reservations to set the IP. This rather than static. This isn't fully my choice, it is how it was set up by a previous person and seems to be largely preferred within our network. The second IP address exists on a separate subnet from the first, though there is currently routing between them, both subnets can communicate to each other via their gateways.
I do not have any DNS server at this time, though it's possible that our Ubiquiti router may be performing some local dns resolution. It is certainly set as the name server within my dhcp settings. We have configured a domain name on the Ubiquiti, and typically we use a FQDN to contact the server in question. I'm aware that I probably wouldn't have this issue if I had a DNS server. It's planned but not implemented at this time. Never the less, both hostname and fqdn resolution broke when I implemented the second IP address.
It does also occur to me that strictly speaking, I'm unclear as to how hostname's are being resolved at all for this server, and for other Ubuntu servers on my network. I'm unsure how that information is conveyed and what I can do to influence it.
Secondary bit of information: I repeated this same condition on an older server running Ubuntu 16.04. I'm aware that it's out of support though, and I'm not asking for a resolution applicable to it. Simply I used it as a test for this behavior in a different server.
