Guide / timing
DNS propagation is a cache story.
There is no single global switch that flips. Recursive resolvers keep answers until their TTL expires, then ask again.
What TTL controls
TTL, or time to live, is the cache lifetime in seconds. A 3,600-second TTL tells a resolver it may reuse the answer for up to an hour. It is guidance, not a countdown displayed consistently to every user: resolvers can serve stale data during incidents, and local operating systems may cache too.
A sensible migration sequence
- Lower the old record’s TTL ahead of the move, while the old answer is still live.
- Change the target and keep the old service available during the overlap.
- Query multiple public resolvers and test from a real network.
- Raise TTL after the new destination is stable.
How to troubleshoot
Record the resolver, query name, type, answer and TTL when you test. Our DNS checker uses a public DNS-over-HTTPS endpoint, so compare it with the resolver your affected users actually use. If DNS is correct but the browser fails, inspect TLS, redirects and server status next.
Useful distinction: “I changed DNS” and “the site is down” are separate observations. Prove each layer independently.
For the vocabulary, read DNS records decoded.