After switching a domain's nameservers at the registrar, a resolver that looked the domain up before the switch can keep sending queries to the old nameservers for up to the delegation TTL (48 hours for .com), so the site looks down from that network while it is up everywhere else. Test through other resolvers before concluding the new setup is broken.
Situation: nameservers moved from the registrar's default DNS to another provider. The registry showed the new delegation at once and public resolvers returned the new address, but one ISP's resolver kept returning the registrar's old parking address, and kept refreshing it each time its 30-minute record TTL ran out, because the old nameservers were still answering for the zone.
How to tell this apart from a real outage:
dig +norec +noall +authority example.com @a.gtld-servers.net # what the registry says
dig +short example.com @1.1.1.1 # a resolver with no stale cache
dig +short example.com # your own resolver
curl --resolve example.com:443:<new address> https://example.com/ # the site itself
If the registry and public resolvers agree and only your resolver differs, the setup is fine. Only networks that cached the old delegation are affected, which for a new domain is usually just your own. Either wait for the TTL to run out, or point the machine at a public resolver.
Lowering a TTL beforehand does not help: the delegation TTL is set by the registry, not by you. What does help is not visiting the domain from your own network before the switch, so that its resolver has nothing cached.
This post
- supersedes another post
Replies and references
Nothing yet.