CloudSpex

Domain health guide

How to troubleshoot a DNS resolution failure

When DNS cannot resolve a domain, dependent website and certificate checks cannot reliably reach the target. Start with the response type and the team that controls the DNS zone.

Use the DNS Health Check

Türkçe

Separate the failure types

NXDOMAIN means the requested name does not exist in DNS. Check spelling, the DNS zone, and whether the record was published at the intended name.

SERVFAIL means a resolver could not complete the request. It can be temporary, but it can also indicate an authoritative nameserver, delegation, or DNSSEC-related issue. The response type tells you where to start; it does not replace investigation.

Check ownership before editing

Identify the registrar, DNS provider, and any team that manages CDN, hosting, or infrastructure automation. A record may be correct in one system but unpublished because the authoritative nameservers point elsewhere.

Avoid changing multiple unrelated records at once. A narrow change is easier to verify and easier to reverse if it creates a new problem.

A practical diagnostic sequence

Confirm the exact hostname first. Then confirm the authoritative nameservers, the intended record type, and the public result after the change. Allow for DNS propagation and caching, but do not use propagation as an explanation before the authoritative zone has been checked.

  1. Confirm the hostname and record type being queried.
  2. Check which nameservers are authoritative for the domain.
  3. Compare the intended zone record with the public response.
  4. Review delegation and DNS provider status when the response is SERVFAIL.
  5. Retest public resolution before treating SSL or HTTPS as separate failures.

Why dependent checks may be blocked

SSL and HTTPS checks need DNS to locate a public service first. When DNS resolution fails, a careful diagnostic treats those checks as not assessed because of the DNS issue.

CloudSpex follows that approach: DNS is the root cause when it prevents dependent checks, rather than duplicating the same condition as separate critical findings.

How a name is resolved

Troubleshooting is easier with the path in mind. A resolver that knows nothing about a name starts at the root servers, which refer it to the servers for the top-level domain, such as com. Those refer it to the domain's authoritative nameservers, which are the ones listed at the registrar. The authoritative servers give the final answer, and the resolver caches it for as long as the record's TTL allows.

A failure can occur at any link. The registry may not delegate the domain at all, the delegation may point to servers that do not answer for it, the authoritative servers may not hold the record, or a resolver along the way may be serving a stale answer. Each produces a different symptom.

Reading the response code

Every DNS answer carries a status. NOERROR with records means the name resolved. NOERROR with an empty answer means the name exists but has no record of the type you asked for, which is easy to mistake for a failure: asking for an AAAA record on a host that only has an A record is the usual case.

NXDOMAIN means the name does not exist at all. SERVFAIL means the resolver could not obtain a valid answer, which points at the authoritative servers, the delegation or DNSSEC. REFUSED means the server you asked declines to answer for that zone, typically because it is not configured to serve it. A timeout means nothing answered, which is a network or availability problem rather than a data problem.

Asking for the same name three ways
dig www.example.com A
dig www.example.com A @1.1.1.1
dig www.example.com A +trace

Compare the delegation with the zone

A large share of SERVFAIL and intermittent failures come from a mismatch between two lists of nameservers: the one at the registrar, which the parent zone hands out, and the one inside the zone itself. They should be identical.

  1. Look up the NS records the parent zone returns for the domain.
  2. Ask each of those servers directly for the domain's SOA record. Every one must answer with authority.
  3. Compare the serial numbers in the SOA answers. Different serials mean the servers are out of sync.
  4. Confirm that each nameserver name itself resolves to an address.
  5. Remove any nameserver from the registrar that no longer serves the zone.
Checking each authoritative server directly
dig example.com NS +short
dig example.com SOA @ns1.example-dns.net +norecurse
dig example.com SOA @ns2.example-dns.net +norecurse

Lame delegation

A delegation is called lame when a listed nameserver does not actually serve the zone. It happens after a migration between DNS providers when the registrar still lists the old servers, or lists a mixture of old and new. Resolvers choose among the listed servers, so the domain works for some queries and fails for others, and the pattern changes from one resolver to the next.

Intermittent failures that cannot be reproduced on demand are the typical sign. The fix is at the registrar, not in the zone: the nameserver list there must contain only servers that currently hold the zone.

DNSSEC failures

If a domain is signed with DNSSEC, a validating resolver rejects any answer whose signatures do not check out, and reports SERVFAIL. The data may be perfectly correct; it is the proof that is broken. The common causes are a DS record at the registrar that no longer matches the zone's keys, usually after moving DNS providers, and signatures that expired because signing stopped.

A quick test is to query a validating resolver with checking disabled. If the name resolves with checking disabled and fails without it, the problem is DNSSEC. Either correct the DS record at the registrar so that it matches the current key, or remove it and wait for the old one to expire from caches before re-enabling signing.

Separating a DNSSEC failure from a missing record
dig www.example.com A @1.1.1.1        ; SERVFAIL
dig www.example.com A @1.1.1.1 +cd    ; answers: validation is the problem

Caching and the meaning of propagation

DNS changes do not spread outward from a source. Each resolver keeps an answer until the TTL it was given runs out and then asks again. During that window some users see the old answer and some the new one, which is what people call propagation.

Negative answers are cached too. If a resolver asked for a name before the record existed, it remembers the NXDOMAIN for a period taken from the zone's SOA record, often between five minutes and an hour. A record that is correct on the authoritative servers can therefore still fail for a while at a resolver that asked too early. Query the authoritative server directly to see the truth, and lower the TTL a day ahead of any planned change.

Mistakes inside the zone

When the delegation is healthy and a specific name still fails, the cause is usually in how the record was entered.

  1. The domain was appended twice, creating www.example.com.example.com, because the panel adds the zone name to whatever is typed.
  2. A CNAME was created at a name that also has other records. A CNAME must be the only record at its name.
  3. A CNAME was placed at the zone apex, which is not permitted alongside the SOA and NS records. Use the provider's alias or flattening feature instead.
  4. The record exists in a different zone from the one the nameservers serve, typically an old copy at a previous provider.
  5. A wildcard record answers for names it was not meant to, hiding the fact that a specific record is missing.

When the domain itself is the problem

If the top-level domain's servers return NXDOMAIN for the whole domain, nothing in the zone matters. The registration has lapsed, the domain is on hold at the registry, or the nameservers were removed. Check the registration status and expiry date with the registrar first. A domain that expired recently is often restorable for a limited period, but the process is slower and more expensive than a renewal, and mail and web traffic stop in the meantime.

What an external check adds

Tests run from inside your own network can mislead, because a local resolver may have a cached answer, a private zone that overrides the public one, or a filter. A check from the public internet shows what customers and other mail servers see. The CloudSpex DNS Health Check resolves the domain from outside, reports which lookups succeeded, and marks dependent checks as not assessed rather than failed when DNS is the cause.

Frequently asked questions

Does DNS propagation always cause a resolution failure?

No. Propagation and caching can delay visibility, but an incorrect zone, delegation issue, or unavailable authoritative service can also cause failure.

Should I change SSL settings when DNS is failing?

Usually resolve the DNS problem first, because SSL and HTTPS checks may be unable to reach the intended service.

Does CloudSpex change DNS records?

No. The public check is read-only.

What is the difference between NXDOMAIN and SERVFAIL?

NXDOMAIN is a definite answer that the name does not exist. SERVFAIL means the resolver could not get a valid answer at all, which usually points at the authoritative servers, the delegation or DNSSEC.

Why does the domain work on one network and fail on another?

Different resolvers hold different cached answers and may pick different nameservers. Intermittent or location-dependent failures often indicate a lame delegation or nameservers that are out of sync.

Can I force DNS to update everywhere?

No. You can only wait for caches to expire. Some public resolvers offer a cache-flush page for a single name, which helps for that resolver only.

Use public checks as a starting point

CloudSpex reads limited public-facing signals. It does not change DNS, website, certificate, or email settings. Confirm the responsible provider and test relevant changes before applying a remediation.

Editorial details

Author and reviewer: CloudSpex. Sources: public technical standards and provider documentation relevant to the topic. Last reviewed: November 10, 2026. We update guides when the product behavior or the underlying guidance changes.