Why did a DNS change not take effect?

Usually caching rather than a fault: resolvers that already answered hold the previous value for the whole of its TTL, so publishing a new value does not un-cache the old one. If the authoritative nameservers are serving the new value, the change is correct and the wait is the TTL of the old record. Where the provider accepted the write but is not serving it, some providers require an explicit zone refresh.

How do I diagnose a DNS problem quickly?

Establish three things in order: what the authoritative nameservers are actually serving, what a public resolver returns, and whether anything changed recently. Query the authoritative servers directly to separate a zone fault from a caching one, run dig +trace to check the delegation at the parent, and check the change history before altering anything — changing records to see whether it helps is what turns a ten-minute incident into a day-long one.

Why is mail not arriving after a DNS change?

The most common cause is an MX target that does not resolve — valid DNS pointing at a hostname that does not exist, which produces no error anywhere. Sending servers queue and retry for up to five days rather than bouncing, so the fault is often days old before anyone reports it. The next most common causes are a second SPF record, which removes SPF entirely, and a DMARC policy at reject with an unaligned sender.

Troubleshooter

Something is broken. Let's narrow it down.

How do I diagnose a DNS problem?

Establish three things in order: what the authoritative nameservers are actually serving, what a public resolver returns, and whether anything changed recently. Querying the authoritative servers directly separates a zone fault from a caching one, and checking the change history before altering anything prevents a ten-minute incident becoming a day-long one.

Question 1

The guided diagnostic needs JavaScript. The same ground is covered in A client is offline and you think it is DNS and A change applied but is not resolving.

Still stuck after this? Raise a ticket from inside the platform — a client currently offline is treated as urgent on every plan, including trials.

Last updated 23 August 2026