Why a domain returns SERVFAIL on validating resolvers after a DNS provider or registrar move, how to confirm a stale DS record with dig +cd, and how to fix it by updating or removing the DS record.
Home › Guides › DNSSEC SERVFAIL (broken chain)
Last updated 17 November 2026
If your domain suddenly returns SERVFAIL from validating resolvers such as 1.1.1.1 but resolves fine with validation turned off (dig +cd), DNSSEC is broken. The usual cause, per Cloudflare's troubleshooting docs, is that the authoritative DNS provider was changed without updating or removing the old DS record at the registrar. The parent zone (for example .com) still points at a key the new DNS host doesn't publish, so validators can't build a chain of trust and treat your answers as bogus.
Fix it at the registrar, one of two ways:
Either way, resolvers can keep the old DS cached until its TTL runs out. Run the DNSSEC Checker to see the DS, DNSKEY and key tags side by side.
Cloudflare's documented test compares a normal query with one that has checking disabled:
dig A example.com @1.0.0.1 ;; ->>HEADER<<- opcode: QUERY, status: SERVFAIL dig A example.com @1.0.0.1 +dnssec +cd +short 203.0.113.10
If the +cd query returns records but the normal one returns SERVFAIL, DNSSEC is misconfigured. Cloudflare notes this often happens when name servers change but DS records don't, and that it can also happen when an attacker tries to forge a response.
Then compare the parent's DS with the zone's keys:
dig +short DS example.com dig +short DNSKEY example.com
The first number in a DS record is the key tag of the DNSKEY it points to. If no DNSKEY has that key tag and algorithm, the chain is broken. dig doesn't print key tags for DNSKEY records in +short output; the DNSSEC Checker calculates them for you (RFC 4034 Appendix B) and also recomputes the DS digest. Microsoft gives the same check for the 4.7.324 error: review the child domain's RRSIG key IDs and make sure they match the key IDs in the DS records in the parent zone.
Resolvers that support RFC 8914 attach an Extended DNS Error (EDE) code to the SERVFAIL. Cloudflare returns it in the Comment field of its JSON DNS-over-HTTPS answers, and the DNSSEC Checker shows it. The DNSSEC-related codes are:
| EDE | Name | RFC 8914 meaning | Usual fix |
|---|---|---|---|
| 1 | Unsupported DNSKEY Algorithm | A DNSKEY RRset contained only unsupported DNSSEC algorithms. | Re-sign with a recommended algorithm such as 13 or 8. |
| 2 | Unsupported DS Digest Type | A DS RRset contained only unsupported digest types. | Publish a SHA-256 (type 2) DS. |
| 6 | DNSSEC Bogus | Validation ended in the Bogus state. | Check DS vs DNSKEY and the signatures. |
| 7 | Signature Expired | No signatures are presently valid and some (often all) are expired. | The DNS host must re-sign the zone. |
| 8 | Signature Not Yet Valid | No signatures are presently valid and at least some are not yet valid. | Check the signing server's clock. |
| 9 | DNSKEY Missing | A DS record existed at a parent, but no supported matching DNSKEY record could be found for the child. | Update or remove the DS at the registrar (the stale DS case). |
| 10 | RRSIGs Missing | No RRSIGs could be found for at least one RRset where RRSIGs were expected. | Make sure every name server serves the signed zone. |
| 11 | No Zone Key Bit Set | No Zone Key Bit was set in a DNSKEY. | Publish keys with flags 256 or 257. |
| 12 | NSEC Missing | The requested data was missing and a covering NSEC or NSEC3 was not provided. | Check the signer's NSEC/NSEC3 settings. |
RFC 4035 defines the states behind these codes. Bogus means the resolver expected to build a chain of trust and couldn't, because signatures failed or DNSSEC records said data should exist and it didn't. Insecure means there is no chain to build (no DS), which is not an error.
If the new DNS host supports DNSSEC, enable it there, copy the DS values it shows (key tag, algorithm, digest type, digest) and replace the old DS at the registrar. Cloudflare notes that if a registrar doesn't list algorithm 13, it may be called ECDSA Curve P-256 with SHA-256, and that Cloudflare Registrar (and .ch and .cz domains) get the DS added automatically.
If you just need the site and mail back, delete the DS record at the registrar. The zone becomes insecure and resolves normally once the cached DS expires.
After the change, check with dig DS example.com and the DNSSEC Checker until the result is SIGNED + VALIDATING or UNSIGNED, not BROKEN.
Cloudflare documents this order for moving a signed zone when the old provider can't do multi-signer DNSSEC. It leaves a short window without DNSSEC protection:
dig DS example.com; Cloudflare says this is commonly 24 to 48 hours for most TLDs.RFC 6781 section 4.3.5 covers the same choice. If the old and new operators cooperate, the move can be done with a pre-published ZSK and a double-signature KSK rollover, so the zone stays signed. If they don't, the only viable option is to let the zone go insecure: remove the DS, move the NS, and add the new DS once the new zone is signed. RFC 8901 describes multi-signer models, which Cloudflare supports when the old provider can publish external DNSKEY records.
CDS 0 0 0 0 and CDNSKEY 0 3 0 0. Cloudflare Registrar uses CDS/CDNSKEY scanning, which Cloudflare says can take one to two days.Exchange Online validates DNSSEC and DANE on outbound mail by default. Microsoft documents 4/5.7.324 dnssec-invalid: Destination domain returned invalid DNSSEC records, and notes that today a domain that signals DNSSEC support but fails the checks usually gets the generic 4/5.4.312 DNS query failed instead. If partners on Microsoft 365 report either code for mail to your domain, check your DNSSEC first. See DNSSEC and SMTP DANE in Exchange Online.
Check DS, DNSKEY, key tags and the AD flag with the DNSSEC Checker, look up NS and SOA records with DNS Lookup and compare resolvers with the DNS Propagation Checker. Once the domain resolves again, confirm mail routing with the MX Record Lookup and Email Deliverability Checker, and read any bounce in the Email Header Analyzer. Related guides: DNSSEC and SMTP DANE in Exchange Online, Exchange Online 4.4.7 message expired and MTA-STS and TLS-RPT setup.
Only validating resolvers check DNSSEC. Resolvers that validate (Cloudflare 1.1.1.1 is one) return SERVFAIL when the chain of trust is bogus, while resolvers that do not validate still return the records. That is why a broken DS can look like an outage for some users only.
It makes the zone insecure in DNSSEC terms: resolvers stop validating it, the same as any unsigned domain. That is far better than a broken chain, which makes validating resolvers refuse to answer at all. You can enable DNSSEC again at your DNS host and add a fresh DS record afterwards.
Until resolvers drop the cached DS record, which depends on the DS TTL set by the parent zone. Cloudflare says waiting for the DS TTL to expire is commonly 24 to 48 hours for most TLDs. Resolvers that never cached the old DS see the change straight away.
Yes, if both providers support it: with cooperating operators (RFC 6781 section 4.3.5.1) or a multi-signer setup (RFC 8901), the new host's keys are published before the switch, so the zone stays signed. If the old provider cannot add the new keys, remove the DS first and accept a short insecure window.
RFC 8914 defines it as: a DS record existed at a parent, but no supported matching DNSKEY record could be found for the child. In practice the DS at your registrar points at a key your current DNS host does not publish. Replace or remove the DS.
Runbooks, checklists and helpdesk templates: IT Admin Packs on Gumroad.