DNS and Web Visibility

Internal preview

This test correlates what the TestCenter's authoritative DNS server sees from your resolver with what its web server sees from your browser. It reports both server-visible legs without claiming to observe the private path between your browser and resolver.

Conceptual paths from a browser or proxy through DNS resolvers to the TestCenter nameserver, and directly or through a proxy to the TestCenter webserver
DNS queries may come from the browser's configured resolver or from a proxy resolving on the browser's behalf. Components may be combined or bypassed; the TestCenter observes only the traffic arriving at its nameserver and webserver.

This service records browser and resolver network addresses and DNS/TLS metadata in operational logs for up to 35 days. See the Privacy Policy for details.

Probe hostname
DNS queries
0
Web requests
0
Valid until

Server-observed timeline

Time Δ SideEventDetails
Raw JSON result

          

How to read this

Each observation uses a new random probe hostname. Because that name has not been used before, it normally triggers fresh DNS queries; the same random identifier then links those queries to the web requests observed for this test.

The DNS address is the recursive resolver's egress address, not necessarily your device. The lookup may have been initiated by your browser or by a proxy resolving on its behalf; the authoritative nameserver cannot distinguish those client-side paths. An ECS entry means that the resolver forwarded a network prefix associated with the client. It is not the client's full address, and no ECS entry simply means that the resolver did not provide one. See EDNS Client Subnet (ECS) for how source prefixes, response scope, privacy and CDN routing fit together.

Time is shown in seconds since the observation started; Δ shows the gap from the preceding row. These values show when each event reached the TestCenter. Usually the initial DNS queries arrive before the first web request. A client or proxy may start connecting as soon as one usable address answer is available, while parallel DNS queries or retries continue; packet loss or delay can therefore make a DNS event appear after the first web request. The ordering is useful evidence, but does not prove the client's internal waiting or selection logic.

The timeline retains the first 50 DNS queries and the first five web requests. Total counters continue after those limits; an omitted count therefore always describes later events that are not shown. DNS type totals include every observed query, including omitted ones.

An HTTPS-record query followed by HTTP/3 is consistent with the browser or an intermediary using HTTPS-record information, but does not prove the browser's internal selection path. Only ECH accepted proves that ECH protected the ClientHello observed on the web connection. This test cannot determine whether the client-to-resolver leg used DoH, DoT or cleartext DNS. See Encrypted ClientHello (ECH) for the inner and outer ClientHello, acceptance, rejection, GREASE and inspecting gateways.

In the raw result, edns_payload_size is the maximum UDP DNS response size advertised by the resolver, not the size of the observed query. The dnssec_ok value reflects the DNSSEC OK (DO) bit—a request for DNSSEC records, not proof that the resolver validated the answer. A CD flag means “Checking Disabled”: it asks an upstream server not to enforce DNSSEC validation for that query. It has little practical effect at an authoritative nameserver and does not prove that DNSSEC validation was disabled for the user; the resolver may validate the response itself.