Australian cloud for business software — built and run in Sydney since 2002

NetLeverage
NetLeverage

VPN connected but internal names do not resolve: read the NRPT and the interface order

Technical article
Troubleshooting

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.

If a Windows VPN shows connected but app01.corp.example.com will not resolve, the DNS server is often fine. The client may be sending the query somewhere else. Windows decides where each query goes from a table of rules (the name resolution policy table, or NRPT), then from the DNS suffix on the preferred interface, then from the suffix search list.

There is no Windows error text or event for this fault. The symptoms are a failed name lookup, or an internal name that resolves to a public or wrong address. The steps below read what the client actually has.

What the error means

Microsoft's VPN name resolution page (Windows 10 and 11, updated 2025-01-28) gives the order:

  1. The network stack looks in the NRPT for a match and tries a resolution if one is found.
  2. If there is no match, the DNS suffix on the most preferred interface, based on interface metric, is appended to a short name, and the query goes to that preferred interface.
  3. If the query times out, the DNS suffix search list is used in order and queries are sent on all interfaces.

The NRPT "is the first place that the stack will look after the DNSCache", so a cached wrong answer comes before any rule.

In a VPN profile the NRPT is built from rows under DomainNameInformationList in the VPNv2 CSP. Each row has a DomainName, and DnsServers (a comma-separated list of DNS servers to use for that name). The CSP defines two match types:

  • FQDN: a DomainName with no leading dot applies only to that one fully qualified name.
  • Suffix: a DomainName with a leading dot, such as .corp.example.com, applies to the namespace, all records in it, and all subdomains.

A row with the name . in the CSP's own example is "for all other (any) traffic through DNS Servers." A Persistent row stays in effect "even when the VPN isn't connected." The separate DnsSuffix node holds one or more suffixes; the first becomes the primary connection-specific suffix for the VPN interface, and the whole list is added to the suffix search list. In Intune, the DNS rule fields Name, Servers and Proxy server map to DomainName, DnsServers and WebServers.

Likely causes

Cause What you see How to confirm
No NRPT row for the internal namespace Internal names fail or go to the home DNS Get-DnsClientNrptPolicy -Effective has no entry for your internal domain
Row written as an FQDN, not a suffix Only one exact name works; hosts under the domain fail The effective namespace has no leading dot
Row points at a DNS server the client cannot reach through the tunnel Queries time out Resolve-DnsName -Server to that address fails
Short names fail but fully qualified names work app01 fails, app01.corp.example.com works VPN interface has no suffix; suffix search list lacks your domain
Another interface has the lower metric Short names get the wrong suffix and DNS server Get-NetIPInterface shows the VPN interface is not preferred
Cached wrong or failed answer Fix applied, name still fails Clear-DnsClientCache, then retry
Persistent rule left in place Internal names time out after the VPN is disconnected A Persistent row exists and no VPN is up
The internal DNS server has no record, or answers differently Fully qualified name fails even when sent straight to that server Resolve-DnsName -Server <internal DNS>

Prove the cause before changing anything

These commands are taken from the Microsoft documentation and have not been run or lab-tested. All of them are read-only except Clear-DnsClientCache, which is harmless and listed last.

1. Read the effective NRPT.

Get-DnsClientNrptPolicy -Effective
Get-DnsClientNrptPolicy -Effective -Namespace ".corp.example.com"

The cmdlet page says -Effective returns the effective policy, and -Namespace limits the answer to one namespace. Each entry is a Namespace with its settings, including the DNS servers.

A good result has an entry for your internal suffix, with a leading dot if you expect subdomains to match, and the internal DNS servers in its server fields. A bad result has no such entry, an entry without the dot, or servers that are wrong. Microsoft's samples on that page are DirectAccess-style policies, so the field that holds the servers for a VPN rule may be shown differently from the page's sample. Read every populated server field.

2. Read the VPN interface's DNS servers and suffix.

Get-DnsClientServerAddress -InterfaceAlias "Contoso VPN"
Get-DnsClient -InterfaceAlias "Contoso VPN"
Get-DnsClientGlobalSetting

Get-DnsClientServerAddress returns the DNS servers on an interface. Get-DnsClient returns per-interface settings, including the connection-specific suffix. Get-DnsClientGlobalSetting returns SuffixSearchList and notes that the first entry is the primary DNS suffix, if configured. Use your VPN profile's real interface name. If your internal domain is not in the list, short names will not complete.

3. Check which interface is preferred.

Get-NetIPInterface

Get-NetIPInterface shows InterfaceMetric for each interface. The cmdlet page says that for route selection the overall metric is the sum of the route metric and the interface metric, and that the default is automatic. If the home or office adapter has a lower metric than the VPN adapter, step 2 of Microsoft's order sends short names to it.

4. Ask the internal DNS server directly.

Resolve-DnsName -Name app01.corp.example.com -Server 10.0.20.10 -DnsOnly

Resolve-DnsName accepts -Server to query a chosen DNS server and -DnsOnly to use only DNS (no LLMNR or NetBIOS). If this works and the plain command does not, the DNS server and the tunnel path are fine and the client's rules are wrong. If it fails, the server or the route to it is the problem.

5. Run the normal lookup and compare.

Resolve-DnsName -Name app01.corp.example.com -DnsOnly
Resolve-DnsName -Name app01 -DnsOnly

The first tests the rule match. The second tests the suffix path. Which of them fails tells you which part of Microsoft's order to look at. Use the same cmdlet for every comparison. Microsoft's pages do not say how other tools such as nslookup treat the NRPT, so do not mix them when you compare results.

6. Clear the client cache and retry. Clear-DnsClientCache clears the DNS client cache and, per its page, is equivalent to ipconfig /flushdns.

Fix

Fix the profile, not the individual laptop, so every device gets the same result.

Cause: missing or wrong NRPT row. In the VPN profile (an Intune DNS rule, or the DomainNameInformationList node in the profile XML), add a row with DomainName set to .corp.example.com and DnsServers set to the internal DNS servers, as in the CSP's suffix example. Include the leading dot if subdomains must match. Risk: the change goes to every device that receives the profile. Send it to a pilot device first, and keep the previous profile so you can restore it.

Cause: row sends everything to the VPN DNS. A row with DomainName . sends all queries through those servers, according to the CSP example. In a split tunnel that is rarely what you want, because public lookups then depend on the tunnel and the internal DNS. Risk: removing it changes where public names resolve for every user. Do this with a pilot group.

Cause: short names. Set DnsSuffix in the profile. The first suffix becomes the primary connection-specific suffix for the VPN interface, and the list joins the search list. Or ask users to use fully qualified names while the profile is corrected.

Cause: interface order. If the VPN adapter should win, change the interface metric. Disruptive step: changing a metric can move traffic between adapters on the device and may drop sessions in progress. Make the change at a quiet time, note the old value, and set it back to automatic to back out.

Cause: persistent rule. If internal names should not be sent to the VPN DNS when disconnected, remove Persistent or set it false in the profile. A rule that persists sends those names to the VPN servers even when no tunnel is up.

Cause: cache. Run Clear-DnsClientCache.

Cause: server-side. If step 4 failed, check that the record exists in the internal zone and that the DNS server listens on the tunnel address. That is a server and routing check, outside the client.

After a profile change, disconnect and reconnect the VPN on the pilot device and repeat the read-only checks.

Confirm the fix

  1. Get-DnsClientNrptPolicy -Effective -Namespace ".corp.example.com" shows your namespace and the internal DNS servers.
  2. Resolve-DnsName -Name app01.corp.example.com -DnsOnly returns the internal address.
  3. Resolve-DnsName -Name app01 -DnsOnly returns the same address if short names are expected to work.
  4. Open the application or share by name.
  5. Disconnect the VPN and confirm the behaviour is what you intended. Persistent rules should still apply; the others should not.

What to watch afterwards

  • Profile changes: re-run step 1 on a pilot device after every profile update.
  • New DNS servers or a changed domain name: update DnsServers and DnsSuffix in the profile with them.
  • Devices with a second adapter (docking stations, tethering), which can change which interface is preferred.
  • Reports of public names failing after you add a . rule.

Example (illustrative): lt-014 connects to a split-tunnel VPN. app01.corp.example.com fails, but Resolve-DnsName -Name app01.corp.example.com -Server 10.0.20.10 -DnsOnly works. Get-DnsClientNrptPolicy -Effective lists corp.example.com without a leading dot, so the rule matches only that exact name. The administrator changes the profile row to .corp.example.com, reconnects a pilot laptop, and the lookup resolves. The change is then rolled out to the group.

Risks and limits

  • Microsoft documents no event or error text for a missing NRPT rule; confirmation depends on reading the effective policy.
  • The Get-DnsClientNrptPolicy sample output on Microsoft Learn shows DirectAccess-style policies; field names for a VPN profile may differ.
  • Get-DnsClientNrptRule is documented for reading rules from a Group Policy object (-GpoName). Reading the effective policy instead shows what the client applies.
  • The name resolution page applies to Windows 10 and Windows 11 clients. Behaviour on other versions is not covered here.
  • Third-party VPN clients may manage DNS differently and are not covered.

Sources