Lingering bad DNS traffic to our authoritative DNS server

I recently wrote about how getting out of being people's secondary authoritative DNS server is hard . In the process of that I said that there was a background Internet radiation of external machines throwing random DNS queries at us. Now that we've reduced the number of DNS zones that were improperly still pointing at us for historical reasons, I think I can finally see enough from our public authoritative DNS server's traffic to say something about that.

The rejected DNS queries we're seeing so far are a mixture of three types of queries. The first sort of query is for one of those DNS zones that used to be pointing to us but haven't been for long enough that people's DNS caches should have timed out by now. My best guess is that some systems simply hold on to DNS nameserver information for well over any listed TTLs for it. The amount of these queries has been going down for some time so it seems that eventually people do refresh their DNS information and stop poking us.

The second sort of query is for more or less random DNS names that have definitely never pointed at us, not infrequently in well known domains such as 'google.com' or 'googleapis.com', or a well known name like 'chrome.cloudflare-dns.com'. The source IPs for these queries are all over and they're generally low volume. Some IPs may be probing to see if we have any sort of open recursive resolver behavior, but others seem much more random, enough so that I wonder if the remote machines are experiencing some sort of corruption in the DNS server IP that they want to query (or perhaps their DNS lookup software or resolving DNS software is copying the NS record from one entry over to another).

(Sometimes people even make queries for things in the RFC 1918 portion of in-addr.arpa.)

The third and largest source of bad traffic is queries for what look like internal domains within at least one top level domain (and I'm going to name it, it's koenigmetall.com). On spot checking so far, all of the queries come from IP addresses that seem to be located in Romania. What I suspect here is a version of people using our 128.100/16 as internal IP address space . Our public authoritative DNS server is at the IP address 128.100.1.1, which is a very attractive IP to put something important on if you're using 128.100/16 internally. So I suspect that if someone were to inspect the internal DNS of the company in question, they'd find errant DNS NS and A records that said an internal DNS server for these internal zones was found at 128.100.1.1. Then queries theoretically to that internal DNS server are leaking onto the public Internet and reaching us, likely in a process similar to how people keep sending dynamic DNS updates to us (that entry is from 2021 but it's all still going on).