Network

Compare saved DNS responses after changing an A record

·

After an address change, compare what each recorded observation returned for the same DNS name and type. In the sample below, one capture shows the planned address, one shows an older address and one records a timeout. Each calls for a different next check.

DNS Capture Comparison

Compare five saved IPv6 rollout observations

Choose Load IPv6 rollout example and Compare captures. The report shows two comparable captures in a 15-second UTC window. London recorded 2001:db8::20 with a TTL of 600 seconds. Dublin recorded 2001:db8::10 with a TTL of 41 seconds, using expanded IPv6 notation in the saved answer.

Choose Dublin under Reference capture. The two values trade positions in the difference report, while saved timestamps and TTLs stay with their original rows. Madrid records a timeout, Paris records REFUSED with Extended DNS Error code 18, and Helsinki recorded NOERROR without a direct AAAA answer.

Capture labels describe the supplied observation points and resolvers. Compare the report with the addresses planned for your change and keep separate records of any later observations.

Download the five-observation IPv6 rollout bundle (JSON)RFC 8914 Extended DNS ErrorsRFC 2181, resource record sets and time to live

Find which test point recorded the older address

Open the sample capture bundle or select Load example in the tool. Click Compare captures. The sample uses documentation addresses and fictional observation points. London is the first complete capture, so it becomes the reference for the planned address, 192.0.2.20.

Dublin is marked Different. Look for 192.0.2.10 under Only here and 192.0.2.20 under Only in reference. The 41-second TTL belongs to Dublin’s capture at 14:00:04 UTC. Madrid is Unavailable because it recorded a timeout. Its timestamp, 15:00:05+01:00, is 14:00:05 UTC.

Find which test point recorded the older address
Observation pointAnswer and TTLExpected result
Office · London192.0.2.20, 300 sReference
Test VM · Dublin192.0.2.10, 41 sDifferent
Test VM · MadridTimed outUnavailable

Download the three-observation DNS bundle (JSON)

Check whether the address or only the TTL changed

In the sample editor, change Dublin’s 192.0.2.10 to 192.0.2.20 and keep the TTL at 41. Compare again. Both locations now report the same address, so Dublin shows Matches while keeping its own TTL.

For a real change, retain both timestamps and resolver identities. A difference can be consistent with cached data. Resolver policy, split DNS and geographically targeted or rotating answers can also explain different values. Check the planned record and repeat the observation with the same query and resolver before attributing a difference to caching.

RFC 2181, resource record sets and time to liveRFC 8767, serving stale DNS data

Prepare and export comparable observations

Use one query name and record type at the top of the JSON bundle. For each capture, give the observation point, the resolver used by that point, and the capture time with seconds and a UTC offset. Put a saved DNS response in response with its original Question. To preserve a recorded failure, use an error string and optionally errorKind of timeout, policy or transport. For a saved DNS response with an Extended DNS Error, use extendedError with its numeric code and optional text.

Direct answer values are compared as sets, so order and repeated values leave the match unchanged. Addresses and supported DNS names are normalized. TXT uses the exact saved data string, including case and quoted segments. A CNAME can explain why an address for a different owner appears in the response. Query that owner separately if you need to compare its direct addresses.

Download the report and check that the intended reference and original capture offsets appear beside the expected answer sets. Keep the original DNS bundle with the comparison report. Each report covers the captured observations and their stated scope.

Google Public DNS JSON response formatRFC 2181, resource record sets and time to liveRFC 4343, DNS case insensitivity

Common questions

What if one point reports NXDOMAIN and another records a timeout?

NXDOMAIN is a DNS response code. A timeout means you have no DNS response from that attempt. Both rows are unavailable for a comparison of positive answer sets. Their separate statuses preserve the distinction.

Can I use the shortest TTL to predict when everyone will see the new record?

The reported TTL describes an answer at its capture time. A reliable update assessment needs more observations, including repeat captures from resolvers your users actually use.

Sources and sample

Put it into practice.

Related guides