Search
Enter at least two characters without personal information.

    LibraryArticles & FAQ Your projectRequest a quote
    LanguageEN
    Practical guides

    Change DNS without breaking your website or email

    Moving a website does not always require changing registrar or DNS servers. Start by defining the operation: moving hosting, transferring registration or replacing the service publishing the zone. Dependencies and tests differ.

    Go to the method

    Cables and connection diagram beside a laptop
    AI-generated illustration

    Inventory the zone and responsibilities

    Identify the domain holder, registrar, DNS provider, website host and email host. Record renewal dates and the person authorised to approve the change. Keep a dated record inventory in the project workspace, without exposing access methods publicly.

    Cover the main domain, www, subdomains, A and AAAA addresses, CNAME aliases, email MX records and TXT records used for authentication or service verification. Check CAA and DNSSEC dependencies where present. A working homepage does not demonstrate that email and other services work.

    Prepare the destination first

    When replacing the DNS provider, recreate the required zone and compare it with the approved inventory. When moving only website hosting, limit changes to affected records. Ask service owners to confirm dependencies rather than deleting TXT records whose purpose is unclear.

    TTL describes how long a response may remain cached. Reducing it sufficiently before a change may limit persistence of the old value, but does not replace validation or recovery planning. Where DNSSEC is enabled, zone signing and the parent DS record must be coordinated; a mismatch can make the domain unavailable to validating resolvers.

    Fictional exercise, on one day in UTC: an A record’s TTL changes from 3,600 to 300 seconds at 10:00; its address changes at 10:10. A response received at 09:59 with the old TTL may still be cached at 10:30. Lowering the zone’s TTL does not rewrite answers already stored by resolvers.

    The table assumes ordinary caching without prefetch or serve-stale. A TTL is an ordinary retention limit, not a guaranteed minimum: an entry may be evicted earlier. A negative answer has its own TTL derived from SOA information; creating the name later does not purge caches that recorded its absence.

    Scroll the table horizontally to read every column. With a keyboard, focus the table area and use the arrow keys.

    Fictional exercise: four answers and their cache deadlines
    Answer received in UTCTheoretical expiry on the same dayWhat the observation can explain
    Old address · 09:59 · TTL 3,600 s09:59 + 01:00 = 10:59The old address may persist after the 10:10 switch.
    Old address · 10:04 · TTL 300 s10:04 + 00:05 = 10:09This entry expires before the planned switch.
    New address · 10:11 · TTL 300 s10:11 + 00:05 = 10:16This cache received the updated destination.
    NXDOMAIN · 10:08 · effective negative TTL 600 s10:08 + 00:10 = 10:18A name created at 10:10 may remain absent in this cache.
    • RFC 2181 · DNS TTL

      Sources consulted on 22 October 2026: RFC 2181 for TTL, RFC 2308 for negative caching and RFC 8767 for serve-stale. These times illustrate a model, not a universal completion time for a DNS change.

    • RFC 2308 · Negative caching

      Read the mechanism and its limits in this primary reference.

    • RFC 8767 · Serve stale

      Read the mechanism and its limits in this primary reference.

    Test the website, email and related services

    Define trials for the main domain and www over HTTPS, an internal page, external email sending and receiving, a contact form and important subdomains. Compare results across several resolvers and networks. Record times, returned values and discrepancies.

    Visitors may temporarily reach different destinations depending on cached responses. Plan how to prevent divergent writes if orders or enquiries continue during that period. A static presentation page and an active shop need different continuity arrangements.

    A SERVFAIL response after changing nameservers warrants checking the DNSSEC chain. The DS in the parent zone must match the expected signing key at the authoritative provider. Copying A, AAAA and MX records does not recreate that trust relationship.

    Compare an authoritative answer with one from a validating resolver, recording time, queried name and type. An answer obtained with validation disabled for diagnosis is not a repair. Have domain owners coordinate delegation, keys and their cache lifetimes.

    Scroll the table horizontally to read every column. With a keyboard, focus the table area and use the arrow keys.

    DNSSEC: separate observations before choosing a response
    ObservationPossible meaningEvidence to retain
    Correct direct answer, SERVFAIL at a validating resolverValidation may fail; this symptom alone does not identify the causeParent DS, DNSKEY and zone signatures
    Old DS after a provider changeDelegation may still refer to the previous keyMatch against the new provider’s keys
    Answer available without validationDNS data may be reachable despite a broken trust chainRestored validated resolution, then website and email tests

    Define recovery conditions

    Before changing anything, name the operator, the symptoms that trigger rollback and the old values to retain. DNS recovery is not instantaneous for every visitor; caching and TTL also belong in the plan.

    After stabilisation, check certificates, automated email and provider verification. Retire old dependencies only after confirming they are unnecessary. Registrar transfers may follow extension-specific rules: consult the registry or registrar rather than assuming identical timing for every domain.

    Recovery case: the website works but email stops

    Synthetic example: the new homepage opens, but a test mailbox no longer receives messages. Compare published MX records with the inventory and inspect TXT records used by the sending service. Establish whether nameservers changed or only web records were updated. A positive HTTP check does not resolve this finding.

    The change record should distinguish three decisions: fix a missing record, restore the old zone or keep services running in parallel during observation. For each, identify who can act, how web and mail are checked and what evidence closes the incident. Do not delete a DNSSEC dependency or an unexplained TXT record merely to simplify diagnosis.

    Content updated on

    Acceptance matrix to adapt to your project

    These proposed checks use synthetic cases. Decide the required behaviour with the team, record the result and assign unresolved gaps before release.

    Scroll the table horizontally to read every column. With a keyboard, focus the table area and use the arrow keys.

    Test cases, expected outcomes and useful evidence
    CaseExpected outcomeEvidence to retain
    Prepared zoneRequired services have their expected records.Inventory of required names and record types, expected values and answers before and after the change.
    HTTPS websiteMain domain, www and internal pages work.URLs, statuses and certificates.
    EmailSending, receiving and forms work.Test messages and authentication checks.
    RecoveryOld values and responsibilities are documented.Procedure approved before the change.

    Frequently asked questions

    Does moving hosting require transferring the domain?

    No. Registration may stay with the same registrar while DNS points to a different host. Registration transfer, DNS and hosting are separate operations.

    Why does the website work while email fails?

    MX or authentication records may have been lost when replacing the zone. Compare values with the email provider’s requirements and the original inventory.

    Does DNS propagation always take 48 hours?

    There is no universal duration for every response. TTLs, caches and the changes made affect the result. Observe values from several points and maintain service continuity.