About this article: This article is reviewed against the cited sources. This is source research, not lab validation. Check the stated versions and your own requirements before making changes.
When Certbot reports DNS problem: NXDOMAIN looking up A for www.example.com, Let's Encrypt asked the DNS for the address of that name and was told the name does not exist. It is not a firewall or web server problem. Either the name in the request is wrong, the name has no records in the zone that the world sees, or the delegation points at a different zone from the one you edited.
What the error means
The Let's Encrypt validation software (Boulder) builds this message in bdns/problem.go. For a DNS answer that is not a success, it takes the response code name, then writes DNS problem: followed by that code, looking up, the record type, for, the name. For a name error the code adds - check that a DNS record exists for this domain. For a server failure it adds - the domain's nameservers may be malfunctioning.
NXDOMAIN is the name error. RFC 2308 section 2.1 says that in a name error "the domain referred to by the QNAME does not exist". RFC 8020 states that an NXDOMAIN answer means the name "AND ALL THE NAMES UNDER IT do not exist". A different case is NODATA, where the name exists but has no records of the type asked (RFC 2308 section 2.2, answer code NOERROR with no relevant answers). When a lookup succeeds but no usable address is left, va/dns.go builds different words (no valid A records found for and the name), which are covered below.
For HTTP-01 and TLS-ALPN-01 validation, Boulder looks up both A and AAAA records for the name (va/dns.go, called from va/http.go and va/tlsalpn.go). It returns an error only when both lookups fail, and the source joins the two messages with a semicolon, so the combined message reads DNS problem: NXDOMAIN looking up A for www.example.com - check that a DNS record exists for this domain; DNS problem: NXDOMAIN looking up AAAA for www.example.com - check that a DNS record exists for this domain. For DNS-01 the lookup is a TXT lookup at _acme-challenge followed by the name (the Challenge Types page gives _acme-challenge.<YOUR_DOMAIN>), and a failure there carries the same message with TXT.
If Let's Encrypt's resolver adds extended DNS error information, the message takes a different shape: DNS problem: looking up A for the name, then a colon and a summary such as the extended error's description. If you see that form, read the summary rather than the NXDOMAIN text above.
Every name in the request must validate. The Rate Limits page says "Before a certificate can be issued, all authorizations in the order must be successfully validated." One missing name therefore fails the whole certificate.
Likely causes
| Cause | What you see | How to confirm |
|---|---|---|
| A name in the request is misspelled | The failing name is not one you serve | Compare the name in the message with certbot certificates and your site names |
The name has no A or AAAA record (for example www was never created, or was removed) |
NXDOMAIN for a name you expect to exist | dig the name at the authoritative server |
| The registrar's name servers hold a different zone from the one you edited | Your provider's console shows the record, but public lookups do not | Compare the registrar's NS list with the zone you edit |
| A CNAME points to a name that does not exist | NXDOMAIN for a name that has a CNAME | dig the CNAME target |
| A recently added record is hidden by a cached negative answer | The record shows at the provider, but validation still reports NXDOMAIN straight after you add it | Read the SOA record's negative caching time (dig SOA example.com, RFC 2308 section 5), wait that long and retest |
| Private or reserved addresses only | no valid A records found for the name, not NXDOMAIN (va/dns.go skips reserved addresses) |
dig +short A returns only an internal or reserved address |
| The name exists but has no A or AAAA record (NODATA) | va/dns.go builds the no valid A records found for text whenever no usable address remains, so expect that text rather than NXDOMAIN |
dig for the name returns status NOERROR and no answer section |
Prove the cause before changing anything
All of these are read-only.
1. See which names the certificate asks for.
certbot certificates
The Certbot manual says this lists each certificate with its Domains and expiry. Check the failing name against that list and against what you really serve.
2. Ask the DNS for the failing name.
dig +short A www.example.com
dig +short AAAA www.example.com
The dig manual describes +short as a terse answer. Empty output with +short means nothing was returned, not why. A name error and a missing record look the same in short form. The manual says the default is the verbose form, so run the query once without +short: the reply carries a header with the response status (NXDOMAIN for a name error, NOERROR with an empty answer for a name with no record of that type). The dig manual's description of its exit status mentions "NXDOMAIN status", but it does not itself describe the reply header.
3. Find the authoritative servers and ask them directly.
dig +short NS example.com
dig @ns1.example.net www.example.com A
The manual shows the form dig @server name type. If the authoritative server returns the record but your own resolver does not, a cache or an old delegation is the cause. If it also says the name does not exist, the record is missing from that zone.
4. Follow the delegation from the root.
dig +trace www.example.com
The manual says +trace "makes iterative queries to resolve the name" and shows the answer from each server. Compare the name servers it ends at with the ones in your provider's console. A mismatch means you edited a zone the internet does not use.
5. Check a CNAME target. If the name has a CNAME, repeat step 2 for its target. RFC 2308 section 2.1 shows that a name error can come with a CNAME in the answer, so the missing name may be the target.
Fix
Fix the cause, then test without spending production limits. The Certbot manual says --dry-run uses the Let's Encrypt staging server, and describes it as a test of renew or certonly "without saving any certificates to disk". The Rate Limits page recommends the staging environment while troubleshooting and says it "has significantly higher limits". The same page says "Up to 5 authorization failures per identifier can be incurred by one account every hour", refilling at 1 per identifier every 12 minutes; check the current page.
Correct a misspelled or removed name
If the name should not be in the certificate, change the certificate's names. The "Changing a Certificate’s Domains" section of the Certbot manual says: "If certificate example.com previously contained example.com and www.example.com, it can be modified to only contain example.com by specifying only example.com". Its command is certbot certonly --cert-name example.com -d example.com. As advice, add the same options the certificate was first issued with (authenticator, installer, server and hooks), and list only the names that should stay.
First save the listing and a copy of the Certbot directory. The Certbot manual shows certificate files under /etc/letsencrypt/live/.
certbot certificates > /root/certbot-certificates-2026-11-20.txt
cp -a /etc/letsencrypt /root/letsencrypt-2026-11-20
Then rehearse the same command with --dry-run, which the Certbot manual says tests certonly against the staging server without saving any certificates to disk. Add --dry-run to the command above and read the result before you go on.
Disruptive step: the same command without --dry-run reissues the certificate with the new set of names. Any site still using a name you left out will show a name mismatch. The old files cannot be rebuilt by running certbot certonly again with the removed name, because that name still returns NXDOMAIN and validation fails again. When the dry run succeeds, run the command without --dry-run. To back out, put the saved directory back with these commands, then check and reload the web server:
mv /etc/letsencrypt /root/letsencrypt-failed-2026-11-20
cp -a /root/letsencrypt-2026-11-20 /etc/letsencrypt
After that, run your web server's own configuration check and reload it. The restored certificate still contains the removed name, so its next renewal fails in the same way until a DNS record for that name exists.
Create the missing record
First save what the zone holds now: export the zone, or copy the record list from your DNS console into a dated file. Risk: an A record that points at the wrong address sends visitors, and validation, to that address. Then add the A (and AAAA if you use IPv6) record in the zone that the registrar's NS records point to. Check the address before saving. To back out, remove the record you added.
Fix delegation
If dig +trace ends at different name servers from your console, first print and save the current delegation with dig +short NS example.com, and keep a dated copy of the registrar's NS list. Risk: changing NS records moves the whole domain, including mail, to another set of servers. Then change the records at the zone the world sees, or correct the NS records at the registrar. To back out, put the saved NS list back at the registrar. See "DNSSEC SERVFAIL after changing DNS provider: the stale DS record and the safe order to move a signed zone" before you change delegation for a signed domain.
Wait out a cached negative answer
RFC 2308 section 5 says a negative answer has a time to live taken from the SOA record, and "upon reaching zero (0) indicates the cached negative answer MUST NOT be used again". Let's Encrypt's pages do not say whether its own resolvers cache negative answers, so if the record is correct at the authoritative servers, wait for the SOA record's negative caching time and retest. Read it with dig SOA example.com; the form dig @server name type is in the dig manual.
Example (illustrative): the certificate for example.com was issued with -d example.com -d www.example.com. The marketing team removes www from DNS. At renewal, the log shows DNS problem: NXDOMAIN looking up A for www.example.com - check that a DNS record exists for this domain. certbot certificates lists www.example.com, and dig @ns1.example.net www.example.com A reports no such name. The administrator confirms www is no longer served, saves the listing and /etc/letsencrypt, and reissues with only example.com.
Confirm the fix
dig +short AandAAAAfor each name in the certificate return the addresses you expect, at the authoritative server and at a public resolver.certbot renew --dry-runsucceeds. The Certbot manual uses this command to test a changed renewal configuration.- A real renewal completes, and your web server serves the new certificate after it reloads.
What to watch afterwards
- When a name is removed from DNS, remove it from the certificate at the same time.
- Alert on renewal failures well before expiry.
- Keep the registrar's NS list and your DNS provider's zone in step after any move.
- If you use DNS-01, check that the
_acme-challengename resolves for the TXT type.
Read next
Related articles you may want next.
- HostingACME challenge types: choosing HTTP-01, DNS-01 or TLS-ALPN-01 for each certificateHow HTTP-01, DNS-01 and TLS-ALPN-01 validate a domain, what access each needs, and how to choose one per certificate so renewals keep working.Read article
- HostingCAA records: limiting which certificate authorities can issue for your domainWhat a CAA DNS record stops, what it cannot, and how a wrong record or DNS fault can block your own certificate renewals.Read article
- TroubleshootingDNSSEC SERVFAIL after changing DNS provider: the stale DS record and the safe order to move a signed zoneDomain gives SERVFAIL after changing DNS provider? Prove a stale DS record with dig +dnssec and +cd, then fix it at the registrar in the safe order.Read article
Risks and limits
- These commands have not been run or lab-tested.
- The message wording is from the current Boulder source. Certbot may add its own text around it, and Let's Encrypt can change the wording.
- The combined A and AAAA message is the one the source builds by joining the two errors with a semicolon; Certbot may show it with extra text around it.
- Limits on the Rate Limits page can change. Check the current page.
- DNSSEC failures give a different message and are a separate fault.
Sources
- Boulder: bdns/problem.go — Let's Encrypt, main branch, no date shown, accessed 20 November 2026
- Boulder: va/dns.go — Let's Encrypt, main branch, no date shown, accessed 20 November 2026
- Boulder: va/http.go — Let's Encrypt, main branch, no date shown, accessed 20 November 2026
- Boulder: va/tlsalpn.go — Let's Encrypt, main branch, no date shown, accessed 20 November 2026
- Let's Encrypt: Challenge Types — Let's Encrypt, last updated February 12, 2026, accessed 20 November 2026
- Let's Encrypt: Rate Limits — Let's Encrypt, last updated August 5, 2026, accessed 20 November 2026
- RFC 2308: Negative Caching of DNS Queries — IETF, March 1998, accessed 20 November 2026
- RFC 8020: NXDOMAIN: There Really Is Nothing Underneath — IETF, November 2016, accessed 20 November 2026
- Certbot documentation: User Guide — EFF Certbot, no date shown (stable), accessed 20 November 2026
- BIND 9 manual pages: dig — ISC, no date shown, accessed 20 November 2026
- cp(1) — GNU coreutils 9.11, April 2026, accessed 20 November 2026
This article was originally published on .