The registrar points to our nameservers, and our nameservers point to themselves. If you query either the registrar or our own nameservers, you get the same responses. But they have different TTLs. Which TTL will clients honour?
$ dig dotser.ie ns @b.iedr.ie
; <<>> DiG 9.8.1-P1 <<>> dotser.ie ns @b.iedr.ie
;; global options: +cmd
;; Got answer:
;; ->>HEADER<<- opcode: QUERY, status: NOERROR, id: 1393
;; flags: qr rd; QUERY: 1, ANSWER: 0, AUTHORITY: 3, ADDITIONAL: 0
;; WARNING: recursion requested but not available
;; QUESTION SECTION:
;dotser.ie. IN NS
;; AUTHORITY SECTION:
dotser.ie. 172800 IN NS ns0.reg365.net.
dotser.ie. 172800 IN NS ns2.reg365.net.
dotser.ie. 172800 IN NS ns1.reg365.net.
;; Query time: 27 msec
;; SERVER: 77.72.72.34#53(77.72.72.34)
;; WHEN: Thu Sep 11 13:46:12 2014
;; MSG SIZE rcvd: 91
$ dig dotser.ie ns @ns0.reg365.net
; <<>> DiG 9.8.1-P1 <<>> dotser.ie ns @ns0.reg365.net
;; global options: +cmd
;; Got answer:
;; ->>HEADER<<- opcode: QUERY, status: NOERROR, id: 64285
;; flags: qr aa rd; QUERY: 1, ANSWER: 3, AUTHORITY: 0, ADDITIONAL: 0
;; WARNING: recursion requested but not available
;; QUESTION SECTION:
;dotser.ie. IN NS
;; ANSWER SECTION:
dotser.ie. 86400 IN NS ns2.reg365.net.
dotser.ie. 86400 IN NS ns1.reg365.net.
dotser.ie. 86400 IN NS ns0.reg365.net.
;; Query time: 41 msec
;; SERVER: 85.233.160.79#53(85.233.160.79)
;; WHEN: Thu Sep 11 13:46:28 2014
;; MSG SIZE rcvd: 91
A very similar question has already been asked, but its accepted (and only) answer suggests a workaround to the specific problem rather than actually answering the theoretical question.
Another similar question is not actually quite the same: it asks whether the nameservers will be cached for longer than the A records. (Anyway, none of the answers there seems to answer my question.)