Ask the same question at a known destination
Dig sends DNS queries and displays their responses. Fix the name, record type, destination resolver and observation time before comparing results. IPv4 and IPv6 answers can differ, so one successful A lookup does not establish every resolution path used by an application. Search suffixes and trailing-dot handling also matter. Success against the tool’s default server is different from proof that the application uses that same resolver. Browsers or services may use an independent cache or DNS-over-HTTPS path. Establish the actual destination and resolution mechanism before treating a convenient command-line result as a faithful reproduction of the affected client’s behavior.
Read response status and sections
The example asks for A and AAAA separately with bounded waiting and attempts. Inspect status, flags, QUESTION, ANSWER, AUTHORITY and server information. NOERROR can accompany an empty answer for the requested type and must be distinguished from NXDOMAIN. SERVFAIL must not be relabeled as proof that the name does not exist. Short output is convenient for reading values, but omits response codes or SOA information needed during diagnosis. Preserve a detailed initial observation rather than only the final address list. The commands demonstrate collection; they do not claim any particular current DNS value was measured or remains permanently configured.
dig example.com A +time=2 +tries=1
dig example.com AAAA +time=2 +tries=1Compare the used resolver with authority
When an update appears stale, ask the same name and type through the resolver used by the application and through the zone’s authoritative servers. Use the intended nonrecursive conditions when consulting authority and inspect indicators such as AA. A stale recursive result suggests TTLs, negative answers or layered caching, while disagreement among authorities suggests update or delegation questions. For negative answers, inspect the SOA and relevant negative lifetime rather than substituting the new A record’s TTL for an already cached NXDOMAIN. Record every destination change. A comparison is useful only when its different paths are explicit rather than silently assumed to represent the same cache or authority.
Bound the meaning of trace and DNSSEC output
The trace option follows delegation through iterative queries beginning from the root. It does not display the internal cache of the application’s recursive resolver. Direct outbound DNS can also be filtered differently from the normal resolver path. Requesting DNSSEC records with +dnssec does not mean dig independently validated the chain merely by displaying those records. Interpret AD as information from the responding validating service together with the trust placed in that service. UDP-versus-TCP differences can motivate investigation of size and path constraints, but one timeout does not diagnose the entire delegation. Match query form and network conditions before drawing conclusions from different responses.
Relate tool results to application resolution
Finally, compare returned addresses with the endpoint the application actually contacted and determine whether the failed request selected A or AAAA. TCP and TLS failures after resolution are different stages, so dig success is not HTTP success. After a repair, repeat the same resolver, name, type and relevant timing conditions, preserving TTL changes where useful. A success through an unrelated server is not sufficient recovery evidence for an affected user path. When sharing a reduced report, retain the question, destination, status and necessary answer or lifetime information. Those details make it possible to separate current authoritative data from a cached historical answer without inventing a diagnosis from an address alone.