Skip to main content

Does DNS TTL Control Propagation Time?

TTL limits how long a cached DNS answer may be reused, but it does not guarantee a worldwide propagation deadline. Prior cache state, negative answers, delegation, resolver behavior, and application caches all affect when a change is observed.

A safer planned address change

T-48h  lower TTL from 86400 to 300
T-0    change the address
T+?    verify authoritative and recursive answers
T+24h  restore an appropriate steady-state TTL

The lead time must cover the previous TTL, not merely the new one. Exact timings depend on the existing zone and operational requirements.

TTL starts when a resolver caches an answer

The authoritative record publishes a TTL in seconds. A recursive resolver that retrieves the answer can reuse it while that timer counts down. If you lower the TTL only at the same moment you change the value, resolvers that cached the older answer may continue using the older, longer TTL. Planned migrations lower TTL far enough in advance for previous caches to expire.

Different resolvers query at different moments, so their cache expiry times are not synchronized. A resolver with no cached answer can see the new data immediately while another continues serving a valid old answer. Some clients, operating systems, browsers, proxies, and applications add their own caching behavior beyond the recursive DNS layer.

Not every DNS change is one record change

Nameserver delegation and DS records live in the parent zone, while application records live in the child zone. Negative responses can be cached when a name did not exist. Provider control planes can also take time to publish a committed change to all authoritative servers. Check the right authority and record type before calling every delay “propagation.”

Relate TTL to the cache that observed it

Observation pointWhat it can tell youWhat it cannot prove
Authoritative nameserverWhat that server currently publishesWhat recursive caches still hold
Recursive resolverIts current cached or fresh answerEvery user’s answer
Local machineThe answer used by that clientWhether the web service is healthy
HTTP requestApplication reachability at one addressDNS state for all networks

Guarantees outside TTL control

A zero or very low TTL increases authoritative query load and does not bypass every downstream cache.

Flushing one laptop’s cache cannot flush an ISP or public recursive resolver.

Changing nameservers involves parent delegation and should not be diagnosed only with a query to the old child authority.

Prepare a DNS change around real cache behavior

  • Record the current value and TTL before changing anything.
  • Lower TTL in advance by at least the duration of the old TTL for a planned migration.
  • Query each authoritative server directly, then compare several recursive resolvers.
  • Keep the old endpoint available during the overlap when the change permits it.

TTL and propagation questions

What does a TTL of 300 mean?

It normally permits a cached answer to be reused for up to 300 seconds from the time that cache obtained it; it is not a promise that every observer updates in five minutes.

Why do some users see NXDOMAIN after I created the record?

A previous negative response may still be cached, or the resolvers may be reaching inconsistent authoritative servers or delegation data.

Put the concept into practice

Try the DNS Lookup with harmless sample data. Then connect the idea to A, AAAA, and CNAME records.

Definitions and standards behind the answer