DNS propagation taking 72 hours—what am I missing?

Changed my NS records at my registrar 3 days ago and still seeing intermittent resolution to old servers

Actually docker is just orchestrated disappointment but this is worse I have verified at registry level that glue is updated dig against authoritative shows correct records yet multiple users report old IPs TTL was 300 seconds should have propagated in minutes not days

My local resolver (systemd-resolved actually which is just dnsmasq with extra steps and journald spam) shows correct records but users on ISP resolvers still stale

Is this just ISP cache disrespect or am I missing something about parent zone update timing þetta er óþolandi

If previous was 86400 then resolver cache that long regardless new value also check if parent zone (.com/.net) actually updated registrar portal say updated but whois server may lag

The TTL and the parent zone update is two separate thing XD

Old TTL is what resolver will cache, that is correct~ but also the registry whois and the TLD server have different refresh time, this is the "propagation" people talk about but actually wrong word

Will show you the check command:

dig +trace yourdomain.com
dig @a.gtld-servers.net yourdomain.com NS

First one show if TLD server updated, second show if registry database updated, sometimes one is YES and other is NO, very annoying two server XD

Your 72 hour maybe because TLD server slow, not because resolver cache

Fighting! Kk

Your registrar maybe not push to registry yet
Check EPP status
Some registrar batch update
Once per day
Not realtime
Very annoying

Also ISP in korea ignore TTL under 1800
ALL RESOLVERS DO THIS
Not just bad ones
"low TTL abuse" they call it
Stupid

Right then, proper nightmare this one

We've had customers blame us for "slow propagation" when it's their ISP playing silly buggers. Honest truth: we push to registry inside 15 minutes at Hostinger, but we've seen ISPs cache for a week against all reason. One engineer at a major UK ISP told me off the record they run unbound with min-TTL 3600 regardless of record. Saves them query volume, apparently. "customer experience" my arse.

Cheers lads, check your TLD authoritative directly before blaming yourself

I am thinking The Vultr Has The Same Issue With The Vps Panel

The Registry Update Is Separate From The Registrar Display So You Must Check The Root Serwer Not Just The Portal

I Vould Suggest Using Dig With The +trace Option To See Vhere The Chain Breaks

Also Some Resolver Ignore Low Ttl As Policy Not A Bug

It Is Very Frustrating I Am Agreeing