Comparisons and explanations · Intermediate

Recursive versus authoritative DNS

Compare recursive resolvers and authoritative servers through delegation, caching, negative answers and practical troubleshooting.

Ask Logeia about this articleTry the three questionsReview your answers

Updated

On this page
  1. Separate lookup responsibility from zone responsibility
  2. Trace a lookup without useful cached information
  3. Explain differing answers before declaring a fault
  4. An authoritative answer is not an application health check
  5. Classify negative answers and temporary failures
  6. Choose the observation point for the question
  7. Ask Logeia about this articleCheck your understanding

Separate lookup responsibility from zone responsibility

A recursive resolver obtains information on a client’s behalf and returns the result. An authoritative server supplies records for a zone it serves. One implementation can support both capabilities, but this does not merge the responsibilities. A DNS server configured on a laptop does not necessarily administer the records of every company whose sites the user visits. The distinction directs troubleshooting toward different places when only office clients fail versus when the underlying authoritative data is incorrect. Count roles in a particular exchange, not physical machines: two functions can coexist, or several machines can collectively provide one service.

Trace a lookup without useful cached information

Consider an A lookup for api.example.com with no relevant cached information. A recursive resolver can follow referrals from the root to com and then to the servers responsible for example.com. The root is not a giant table containing every website’s final address. Delegation assigns responsibility for part of the naming tree to another set of servers. Real resolvers cache referrals and other intermediate information, so they need not repeat the same complete path for each user request. State exactly which caches are empty before predicting a query count. Otherwise an idealized explanatory trace can be mistaken for a mandatory packet sequence.

Explain differing answers before declaring a fault

An authoritative server can show a new address while a recursive resolver still returns an ordinarily valid cached old answer. For example, an answer cached at time zero with a six-hundred-second TTL can have four hundred seconds remaining at time two hundred, even if authoritative data changed at time one hundred. This is an illustrative single-cache timeline. Stale-serving policies, local application storage, and resolver selection can change observations. Align server, time, name, and record type before comparing results. Otherwise differences in locality or IPv4 versus IPv6 can be incorrectly diagnosed as broken caching rather than different questions or observation paths.

An authoritative answer is not an application health check

The AA flag provides information about authority in a DNS response, not a badge establishing that a web application is healthy. An authoritative A or AAAA answer says nothing by itself about an open service port, hostname-valid certificates, or a successful HTTP response. One answer also does not prove that every authoritative server for a zone has matching configuration. When inconsistent service replicas are suspected, record the responsible server set and which server actually answered. Connect DNS evidence to subsequent transport, TLS, and application checks. Keeping those stages separate makes the next investigation more precise rather than merely repeating the same successful DNS command.

Classify negative answers and temporary failures

NXDOMAIN for a nonexistent name differs from SERVFAIL or a timeout. An existing name without the requested record type is another distinct situation. Negative-cache rules can make a newly registered name temporarily invisible through a resolver that recently cached its absence. After checking the authoritative configuration, compare the recursive response code and associated caching information. Do not immediately delete and recreate records simply because an answer section is empty. Classification comes first: a naming error, missing type, validation failure, and unreachable server call for different remedies. Preserving the exact response makes this reasoning possible after an intermittent incident has disappeared.

Choose the observation point for the question

An authoritative service commonly publishes its assigned zone data, while recursive service is often restricted to authorized clients. Unrestricted recursion can let unrelated parties consume resources and cause operational problems. A small controlled dig investigation can specify the server explicitly and, when appropriate, avoid requesting recursion to distinguish roles. No large query flood against an unmanaged server is required to learn this difference. Choose whether the investigation needs the source of authoritative information or the answer a particular user currently receives. That choice determines the server, query flags, and interpretation, rather than one DNS command being universally sufficient.

Reading and practice are recorded separately in this browser, for each language.

Learn this in order

Understand the systems behind the web · STEP 10

Choose a learning path