Tailscale / Troubleshooting

Tailscale MagicDNS Name Fails While the Tailscale IP Works

Prove the Tailscale IP path first, then separate tailnet MagicDNS state, client DNS acceptance, and documented split-DNS or nameserver conflicts before changing any DNS setting.

By Andrey Tretyak · Sources last reviewed

Start with the peer IP, not the hostname. If the same peer is reachable by its Tailscale IP but not by its MagicDNS machine name or fully qualified tailnet name, the failure is narrow enough to investigate DNS. MagicDNS derives those names from the machine name and tailnet DNS name, and search domains support the short form.[1][4]

Then locate the DNS layer that differs: the tailnet MagicDNS state, the client preference for accepting Tailscale DNS, configured split DNS or nameservers, or a local resolver conflict. A general Internet DNS failure belongs to a different path.[2][3]

Sections

Applicability

Products
Tailscale
Scope
For a tailnet where MagicDNS is expected and a normal tailnet peer is reachable by its Tailscale IP address, but the same peer cannot be reached by its MagicDNS machine name or fully qualified tailnet name. This page is not for a peer that is unreachable by IP, general Internet DNS failure, DNS observed only through an exit-node path, subnet-route targets, Synology service-port diagnosis, or shared-device access that requires the full domain name.[1][2]
Last verified

Symptoms

  • The target peer responds over its Tailscale IP address, but the same peer cannot be reached by its MagicDNS machine name or its fully qualified tailnet name from the same client.[1][4]

Quick diagnosis

Prove the Tailscale IP path before choosing a branch. A bare name failure is not proof of a MagicDNS infrastructure problem.

Safe / low risk

Prove the Tailscale IP path to the peer

Run on the failing client after traffic to the intended peer, then confirm the peer entry and reachability by its Tailscale IP address. A peer that is unreachable by IP leaves this page to the connectivity diagnosis.[4]

tailscale status
tailscale ping <peer>
Safe / low risk

Compare the same peer by MagicDNS name

Test the same peer by its short machine name and by its fully qualified tailnet name without changing anything else. Record exactly which form fails; shared devices require the full domain name.[1]

Read-only

Confirm the observation is a plain tailnet name failure

Confirm the failure is not observed only through an exit-node path, not limited to a Synology service or port, and not general Internet DNS. Each of those contexts exits to its own diagnosis.[6][2]

Interactive troubleshooter

Your answers stay in this browser tab. The tool does not connect to your systems or send diagnostic results.

Guided check / step 1

Can the target peer be reached by its Tailscale IP?

Confirm Tailscale IP reachability from the failing client. Name evidence cannot isolate a MagicDNS problem without a working IP path.[4]

Read the complete diagnostic tree without using the controls
  1. Can the target peer be reached by its Tailscale IP?

    Confirm Tailscale IP reachability from the failing client. Name evidence cannot isolate a MagicDNS problem without a working IP path.[4]

    • Yes, the Tailscale IP worksDoes only the MagicDNS name fail?
    • No, the IP path failsUse the connectivity diagnosis instead
    • Cannot confirm the IP pathEstablish the Tailscale IP path
  2. Establish the Tailscale IP path

    Record whether the peer responds over its Tailscale IP address from the failing client.[4]

  3. What did the IP check show?

    Only a proven working IP path continues on this page.[4]

    • The Tailscale IP worksDoes only the MagicDNS name fail?
    • The IP path failsUse the connectivity diagnosis instead
  4. Use the connectivity diagnosis instead

    A peer unreachable by its Tailscale IP has a connectivity problem, not a MagicDNS problem. Prove the IP path first.[4]

  5. Does only the MagicDNS name fail?

    Compare the short machine name and the fully qualified tailnet name against the working IP result. Shared devices require the full domain name.[1]

    • Yes, only the name failsWhere is this name failure observed?
    • No, IP or general DNS also failsInvestigate the general connectivity or DNS problem
    • Cannot confirm the name scopeRecord the exact name results
  6. Record the exact name results

    Record the short-name, full-name, IP, and general Internet name results separately.[1][2]

  7. What did the name check show?

    Only a name-only failure continues toward MagicDNS layers.[1]

    • Only the MagicDNS name failsWhere is this name failure observed?
    • IP or general DNS also failsInvestigate the general connectivity or DNS problem
  8. Investigate the general connectivity or DNS problem

    A failure beyond the MagicDNS name is outside this page: prove the IP path or diagnose general DNS before returning here.[2][4]

  9. Where is this name failure observed?

    Exit-path, Synology-scoped, and shared-device observations each belong to their owning diagnosis. Only a plain tailnet name failure continues here.[6][1]

    • Plain tailnet peerIs MagicDNS enabled and does the machine record match?
    • Only through an exit nodeUse the exit-node diagnosis instead
    • Synology service or portUse the Synology NAS diagnosis instead
    • Device shared from another tailnetUse the full domain name for shared devices
  10. Use the exit-node diagnosis instead

    If the name failure exists only while traffic/DNS uses the selected exit node, stay in the exit-node diagnosis. If the same MagicDNS name also fails on the normal tailnet path, return to this MagicDNS page.[6]

  11. Use the Synology NAS diagnosis instead

    A Synology service or port questioned only by name belongs to the Synology NAS diagnosis, which owns its service and DNS branches.[1]

  12. Use the full domain name for shared devices

    Devices shared from another tailnet require the full domain name; the short machine name alone is not sufficient.[1]

  13. Is MagicDNS enabled and does the machine record match?

    Check the admin console DNS page for MagicDNS enablement and the tailnet DNS name state. Separately, check the target machine record for its current machine name and documented fully qualified tailnet name.[1]

    • Yes, enablement and name look correctDoes the fully qualified name work while the short name fails?
    • No, enablement or name is wrongHand tailnet DNS evidence to the tailnet admin
    • Cannot confirm the tailnet stateRecord the tailnet DNS state
  14. Record the tailnet DNS state

    Record the admin console DNS state (MagicDNS enablement, tailnet DNS name) and, separately, the machine record (machine name, fully qualified tailnet name) without changing shared settings.[1]

  15. What did the tailnet check show?

    Only a correct DNS page state plus a matching machine record continues toward client layers.[1]

    • Enablement and name look correctDoes the fully qualified name work while the short name fails?
    • Enablement or name is wrongHand tailnet DNS evidence to the tailnet admin
  16. Hand tailnet DNS evidence to the tailnet admin

    Preserve the admin console DNS evidence (enablement, tailnet DNS name) and the machine-record evidence (machine name, fully qualified name) for the admin; do not change shared DNS settings as a probe.[1]

  17. Does the fully qualified name work while the short name fails?

    Compare the short machine name against the fully qualified tailnet name. Search domains normally bridge the two.[1]

    • Yes, only the short name failsPreserve the search-domain evidence
    • No, both name forms failDoes the client accept and use Tailscale DNS settings?
    • Cannot confirm the comparisonRecord the short-name versus full-name results
  18. Record the short-name versus full-name results

    Record both name forms for the same peer without changing anything else.[1]

  19. What did the name-form check show?

    The short-versus-full comparison separates search-domain evidence from deeper DNS layers.[1]

    • Only the short name failsPreserve the search-domain evidence
    • Both name forms failDoes the client accept and use Tailscale DNS settings?
  20. Preserve the search-domain evidence

    With the full name working, record the short-name failure with client and tailnet DNS state for admin review instead of rewriting resolver configuration as a probe.[1]

  21. Does the client accept and use Tailscale DNS settings?

    Check the documented client DNS preference. Ignoring Tailscale DNS settings can prevent the client from using the tailnet DNS configuration and is consistent with a name-only failure, but alone does not prove the cause.[3]

    • Yes, the client uses Tailscale DNSIs there a documented split-DNS, nameserver, or local conflict?
    • No, the client ignores Tailscale DNSRestore the documented client DNS preference
    • Cannot confirm the preferenceRecord the client DNS preference
  22. Record the client DNS preference

    Record whether the client accepts Tailscale DNS settings through the documented menu or CLI state.[3]

  23. What did the preference check show?

    The client preference separates local configuration from tailnet DNS state.[3]

    • The client uses Tailscale DNSIs there a documented split-DNS, nameserver, or local conflict?
    • The client ignores Tailscale DNSRestore the documented client DNS preference
  24. Restore the documented client DNS preference

    Restore the client to accept Tailscale DNS settings through the documented flow, then re-test both name forms. Changing this preference affects all names on that client.[3]

  25. Is there a documented split-DNS, nameserver, or local conflict?

    Configured nameservers, split-DNS behavior, policy-limited nameserver access, or local overrides can each break MagicDNS names with a healthy IP path.[1][2]

    • Yes, a documented conflict appliesResolve only the proven documented conflict
    • No, no documented conflict appliesCollect a DNS evidence packet
  26. Resolve only the proven documented conflict

    Hand the exact nameserver, split-DNS, or local-override evidence to the owner of that layer. Do not broaden policy or rewrite resolvers blindly.[1][2]

  27. Collect a DNS evidence packet

    Preserve the peer IP result, both name-form results, tailnet DNS state, and client DNS preference with all identifiers redacted. Send the sanitized evidence for upstream review.[2]

Detailed diagnosis

Read-only

1. Prove the Tailscale IP path works

Establish that the peer is reachable by its Tailscale IP address from the failing client. Without a working IP path, no name evidence can isolate a MagicDNS problem.[4]

Read-only

2. Isolate a name-only failure

Compare the short machine name against the fully qualified tailnet name for the same peer. A failure that also affects numeric-IP connectivity or general Internet names exits this page.[1][2]

Read-only

3. Check the tailnet MagicDNS state

Check on the admin console DNS page whether MagicDNS is enabled and what the current tailnet DNS name is. Separately, check the target machine itself — via the Machines list and its device details — for its current machine name and its documented fully qualified tailnet name. Tailnets created on or after October 20, 2022 have MagicDNS enabled by default; older tailnets may not.[1]

Read-only

4. Check whether the client uses Tailscale DNS settings

Check whether the failing client accepts and uses the tailnet DNS settings. Clients can ignore Tailscale DNS through the documented menu option or CLI preference, which can cause a name-only failure with a healthy IP path.[3]

Read-only

5. Check documented split-DNS, nameserver, and local conflicts

Check configured nameservers and split-DNS behavior, whether policy limits nameserver access, and whether local overrides or resolver behavior explain the failure. Some diagnostic tools bypass system resolution and cannot test MagicDNS.[1][2]

Read-only

6. With no discriminating evidence, preserve the DNS state

Record the peer IP result, the short-name and full-name results, the tailnet DNS settings state, and the client DNS preference with secrets and identifiers redacted. Never convert a bare resolution failure into a proven infrastructure cause.[2]

Supported scenarios

These observations narrow the next investigation; they do not establish a unique cause.

The peer is not reachable by IP

A peer that fails over its Tailscale IP address has a connectivity problem, not a MagicDNS problem. Name evidence cannot discriminate anything until the IP path works.[4]

How to check: Confirm Tailscale IP reachability from the failing client before any DNS reasoning.[4]

The tailnet MagicDNS state does not cover the name

MagicDNS disabled for the tailnet, an unexpected machine name, or a stale fully qualified name each can cause a name-only failure with a healthy IP path.[1]

How to check: Confirm MagicDNS enablement and the tailnet DNS name on the admin console DNS page, and confirm the current machine name and fully qualified tailnet name on the target machine record.[1]

The client ignores Tailscale DNS settings

A client that does not accept Tailscale DNS settings resolves names through local configuration instead, so MagicDNS names can fail while Tailscale IPs keep working. This state is consistent with the symptom but does not prove it by itself.[3]

How to check: Confirm the client DNS preference through the documented menu or CLI state.[3]

A documented split-DNS, nameserver, or local conflict applies

Configured nameservers, split-DNS routing, policy-limited nameserver access, or local overrides can each break MagicDNS names without breaking the IP path. When MagicDNS is disabled, unqualified hostnames are forwarded to the configured nameservers instead.[1][2][5]

How to check: Compare the configured nameservers, split-DNS behavior, and local resolver state against the failing name.[1][2]

Next actions and procedure boundaries

Read-only

Tailnet DNS state fails: hand admin-console evidence to the tailnet admin

Hand the admin console DNS evidence (MagicDNS enablement, tailnet DNS name) and the machine-record evidence (expected machine name, full domain name) to the tailnet admin without changing shared DNS settings as a probe.[1]

Potentially disruptive

Client ignores Tailscale DNS: restore the documented DNS preference

Before changing anything

Risk
Changing the client DNS preference affects name resolution for every tailnet and local name on that client, not only the failing one.[3]
Safer check
Confirm the client currently ignores Tailscale DNS settings and record the current preference before changing it.[3]
Expected result
After the correction, re-test the short name and the fully qualified name; the client again uses tailnet DNS settings, but a persistent failure means the cause lies in another DNS layer.[3]
Backup / recovery access
Record the current DNS preference and any custom resolver configuration before changing the setting.[3]
Rollback
Re-apply the recorded prior DNS preference if resolution does not improve, then continue with the conflict evidence.[3]

Restore the client to accept and use Tailscale DNS settings through the documented menu option or CLI preference, then re-test the short name and the fully qualified name.[3]

tailscale set --accept-dns=true
Read-only

Exit-node, Synology, or shared-device context: use the owning diagnosis

Hand exit-path DNS observations to the exit-node diagnosis, Synology service observations to the Synology NAS diagnosis, and shared-device observations to the documented full-domain-name flow. Do not duplicate their trees here.[6][1]

Read-only

If no branch is proven: preserve the evidence

Record the peer IP result, short-name and full-name results, tailnet DNS state, and client DNS preference with all identifiers and secrets redacted. Send the sanitized evidence for upstream review; a bare resolution failure never becomes a proven cause.[2]

Warnings and boundaries

Never publish unredacted tailscale status output, node identifiers, tailnet or device names where sensitive, auth keys, or logs containing account metadata while diagnosing. Diagnostics redact nothing automatically.[2]
Do not disable firewalls, broaden tailnet policy or nameserver access, or rewrite local resolver configuration as a generic DNS test; first identify the failing DNS layer.[2]

Editorial verification

The citations beside technical claims show their evidence. HomeLabFix chooses the diagnostic order and states where the evidence stops supporting a conclusion.

Author and technical editor
Andrey Tretyak
Verification basis
The cited vendor or project documentation plus the page-specific reasoning above.
Hands-on status
Direct reproduction is claimed only when a page explicitly describes the tested environment and result. “Last verified” means the cited evidence was reviewed on that date; it does not mean every procedure was reproduced.
Creation process
Read how HomeLabFix uses source review, human editorial approval, automation, and AI-assisted tools.
Correction path
Report a source change, scope error, or reproducible contradiction.

Sources

Links were reviewed on 2026-09-13. Reachability and automated validation do not replace editorial verification of each claim.

  1. MagicDNS
    Tailscale · Tier A · accessed 2026-09-13
  2. Can't resolve domain names
    Tailscale · Tier A · accessed 2026-09-13
  3. Manage client preferences
    Tailscale · Tier A · accessed 2026-09-13
  4. Connection types
    Tailscale · Tier A · accessed 2026-09-13
  5. Tailscale changelog
    Tailscale Inc. · Tier A · accessed 2026-09-13
  6. Exit nodes (route all traffic)
    Tailscale Inc. · Tier A · accessed 2026-09-13

Still troubleshooting?