DNS propagation explained: why changes take time and how to check them
DNS changes do not spread, they expire from caches. What propagation really is, how long it takes, how to check it with dig, and how to make a move go smoothly.
· 6 min read · By the Spot Downtime team
You change a DNS record, reload the site and still see the old server. A colleague on another network already sees the new one. This is what people call DNS propagation, and it is the most common reason a migration seems to half work.
The word is slightly misleading. Nothing is pushed out across the internet. What you are waiting for is old cached answers to expire.
What is actually happening
Your records live on your domain's authoritative name servers. When someone visits your site, their device asks a recursive resolver, usually run by their ISP or a public service. The resolver asks the authoritative servers, then keeps the answer in a cache for the length of the record's TTL.
- The moment you save a change, the authoritative servers answer with the new value.
- Each resolver that cached the old answer keeps serving it until its copy expires.
- Browsers and operating systems add their own short caches on top.
So at any moment, different people can legitimately see different answers. That window is the propagation delay.
How long does it take?
The upper bound for most record changes is the old TTL. If the record had a TTL of 3600 seconds, a resolver that cached it one second before your change can keep serving it for about an hour. A TTL of 86400 means up to a day.
Some changes take longer than the TTL suggests:
- Name server changes. Changing the NS records at your registrar updates the delegation held by the TLD servers, and those records have their own, often longer, TTLs. This is the case that can take up to a day or two.
- Negative caching. If someone looked up a name before it existed, the "does not exist" answer is cached too, for the period set in the zone's SOA record.
- Resolvers that ignore TTLs. A few apply their own minimum, so the old value may linger longer than it should.
Check where the change has reached
Ask the authoritative servers directly
This tells you whether your change is saved correctly, ignoring all caches:
dig NS example.com +short
dig @ns1.yourdns.example example.com A +noall +answerIf the authoritative answer is wrong, the problem is not propagation. Fix the record at your DNS provider first.
Ask public resolvers
dig @1.1.1.1 example.com A +noall +answer
dig @8.8.8.8 example.com A +noall +answer
dig @9.9.9.9 example.com A +noall +answerThe number before the hostname in the answer is the remaining TTL in seconds. Run the same command twice and you will see it count down. When it reaches zero, the resolver fetches the new value.
Check many resolvers at once
The DNS propagation checker queries 12 public resolvers for a record and shows what each returns, which makes it quick to see whether a change has reached everyone or only some. For a single lookup, use the DNS record lookup.
Clear your own caches
If only your machine is stale, flush its cache. On macOS and Linux with systemd-resolved:
# macOS
sudo dscacheutil -flushcache && sudo killall -HUP mDNSResponder
# Linux (systemd-resolved)
resolvectl flush-caches
# Windows
ipconfig /flushdnsBrowsers keep a cache too. Restarting the browser or testing in a private window usually clears it.
Make the next change painless
- Lower the TTL in advance. A day or more before a planned move, set the TTL to something short like 300 seconds. Wait out the old TTL so every cache has the short one.
- Make the change. Resolvers now refresh within minutes.
- Keep the old server running. Leave it live and serving the same site until traffic has moved, so users on a stale answer do not hit an error.
- Verify, then raise the TTL again. Once you are confident, restore a longer TTL to cut lookups and make you less dependent on the DNS provider being fast.
Moving email?
Common mistakes
- Editing the wrong zone. After changing name servers, the old provider's zone no longer matters. Edits there do nothing.
- Forgetting a record. Moving the A record but leaving
wwwas a CNAME to the old host, or missing AAAA records so IPv6 visitors still reach the old server. - Testing only from one place. Your office resolver may be stale while the rest of the world is fine, or the other way round.
- Panicking and changing it again. Every extra change restarts the clock for caches that have not yet refreshed.
Watching DNS after a change
After a migration, it is useful to confirm that the records stay as you expect. Spot Downtime's DNS monitoring resolves your records on a schedule and alerts you if a lookup fails or an expected value goes missing, covering A, AAAA, CNAME, MX, TXT and NS records. Pair it with website monitoring so you also know the site itself answers.
DNS propagation checkerSee what 12 public resolvers return for your record right now.Get new guides by email
Practical posts on monitoring, outages, error fixes and on-call, when they're published. No spam.
Keep reading
- DNS · Troubleshooting · IncidentsWhat to do when your DNS provider goes downWhen your DNS host has an outage, your site disappears even though your servers are fine. How to confirm it, what you can do during it, and how to prepare.November 12, 2026 · 6 min read
- DNS · TroubleshootingDNS SERVFAIL: what it means and how to fix itSERVFAIL means a DNS resolver couldn't get an answer for your domain. The usual causes, DNSSEC and broken name servers, and the dig commands that find the fault.November 11, 2026 · 7 min read