VPN Kill Switch and DNS Leak Tests: How to Check Your VPN Is Really Protecting You
VPNkill switchDNS leaksIP leaksWebRTCprivacy testing

VPN Kill Switch and DNS Leak Tests: How to Check Your VPN Is Really Protecting You

AAnyConnect Editorial Team
2026-11-21
8 min read

Learn how to test VPN kill switches, DNS, IPv4, IPv6 and WebRTC leaks across Windows, macOS, Android, iPhone and routers.

A VPN can appear connected while a browser, operating system, or application still exposes identifying traffic. This guide gives you a repeatable VPN kill switch test, DNS leak test, IP leak check, and WebRTC check for Windows, macOS, Android, iPhone, and supported routers. It also explains how to interpret results when using IPv6, split tunnelling, different VPN protocols, or a provider’s application-specific kill switch.

Overview

A useful VPN privacy check tests more than the address shown on a web page. You should verify four separate controls:

  • IPv4 and IPv6 exposure: the public addresses websites can associate with your connection.
  • DNS leaks: whether domain lookups are being sent to your normal internet provider or another resolver instead of the VPN’s intended DNS path.
  • WebRTC exposure: whether a browser feature can reveal local or connectivity-related addresses to a website.
  • Kill-switch behaviour: whether traffic stops when the encrypted tunnel disappears, rather than silently switching to the ordinary connection.

A test result is configuration-specific. A clean result in one browser does not prove that every application is protected, and a clean result on Wi-Fi does not necessarily describe behaviour on mobile data. Treat the process as a baseline that you repeat after changing a setting or network.

CheckWhat to doExpected resultCommon complication
IPv4 addressRecord the address with the VPN disconnected, then connected.The connected result shows the VPN endpoint rather than your normal public address.Some services identify a VPN endpoint or show approximate location; that is not itself a leak.
IPv6 addressRepeat the address check on a network that supports IPv6.Your real IPv6 address is not visible while the VPN is active.A provider may disable IPv6, route it through the tunnel, or handle it differently by protocol.
DNSRun a DNS leak test with the VPN connected.Resolvers are consistent with the VPN configuration or your deliberately chosen secure DNS service.Resolver ownership and location can be displayed differently by different test sites.
WebRTCRun a WebRTC leak test in each browser you use for sensitive work.No unwanted local or public address is exposed.Browser settings, extensions, and operating-system networking can change the result.
Kill switchConnect the VPN, start a continuous test, then interrupt the tunnel.Traffic stops until the VPN reconnects or the block is manually disabled.Some modes protect only selected applications or activate only after a tunnel has been established.

How to estimate whether the VPN is working

Use a simple before-and-after record instead of relying on one green status indicator. Create a small test sheet with these columns: device, operating system, network, VPN protocol, IPv4 result, IPv6 result, DNS result, WebRTC result, kill-switch mode, and reconnect behaviour.

  1. Establish the baseline. Disconnect the VPN and record the public IPv4 address, any visible IPv6 address, and the DNS resolvers reported by your chosen test pages. Do not mistake the baseline for a failure: it is the comparison point.
  2. Connect the VPN. Wait until the application reports an established tunnel. Repeat the IP and DNS checks in the same browser and, where relevant, a second browser.
  3. Check WebRTC separately. WebRTC behaviour is browser-dependent. Test every browser used for work, banking, administration, or other privacy-sensitive activity.
  4. Test the kill switch. Start a long-running download or connection to a site you control, then use the VPN application’s disconnect control. If possible, also simulate a failure by changing network connection or temporarily blocking the VPN interface. Observe whether traffic stops immediately, pauses briefly, or continues over the ordinary connection.
  5. Test recovery. Reconnect the VPN and confirm that traffic resumes only after the tunnel is ready. Check whether applications reconnect automatically and whether any requests made during the interruption appear in logs or browser history.

For a repeatable estimate, score each control as pass, needs investigation, or not applicable. Do not average these into a misleading percentage. A DNS pass cannot compensate for a kill-switch failure during the type of connection you need to protect. Record the reason for every not-applicable result, such as an IPv4-only network or a device that does not support IPv6.

Device-by-device workflow

  • Windows: Connect the VPN and check IP, DNS, and WebRTC in the browsers you use. Review the VPN application for options named kill switch, network lock, block internet without VPN, or similar. Windows VPN clients may offer an application-level mode as well as a system-wide mode; test the mode you actually selected. If IPv6 is not supported by the provider, confirm how the client handles it rather than assuming it is protected.
  • macOS: Repeat the same browser checks, then test sleep and wake, Wi-Fi changes, and manual VPN disconnection. Look for a setting that blocks all traffic when the VPN is unavailable. Menu names differ between provider applications and macOS versions, so use the client’s current documentation when a setting is unclear.
  • Android: Check the VPN application’s always-on VPN and block-connections-without-VPN options in system settings where available. Test both Wi-Fi and mobile data, because the transition between them can expose reconnect behaviour. If an application is excluded through split tunnelling, test that application independently.
  • iPhone: Run the checks after connecting and again after switching between Wi-Fi and mobile data. iOS VPN clients vary in how they describe on-demand protection, kill switches, and reconnect handling. Do not infer system-wide coverage from a browser result alone if other applications are important to you.
  • Supported routers: Test from a device behind the router, and separately test devices or networks excluded from the VPN. Interrupt the router tunnel and observe whether clients lose internet access or fail over to the ordinary WAN connection. Check IPv6 routing, DNS settings, guest networks, and policy-based or split routing.

Inputs and assumptions

Your conclusion depends on the conditions under which you test. Record the following inputs:

  • VPN client and version: provider terminology and default settings can change after an update.
  • Protocol: switching between protocols can alter IPv6 handling, reconnect time, and kill-switch behaviour.
  • Network: test each important connection type, including home broadband, public Wi-Fi, office networks, and mobile data.
  • Browser and extensions: WebRTC controls and privacy extensions can affect what a test page reports.
  • DNS design: a provider’s resolver, a manually selected secure DNS service, and an organisation’s internal resolver produce different expected results.
  • Split tunnelling: excluded applications may use the ordinary connection by design. That is a routing choice, not automatically a leak, but it must match your risk model.

Test sites are useful indicators, not complete audits. A DNS page may identify resolvers but cannot prove that every application uses them. An IP page cannot inspect traffic from another device. WebRTC pages reflect a browser’s behaviour at that moment. For business or high-risk environments, combine these checks with firewall rules, endpoint logs, provider documentation, and controlled testing from a device you administer.

Also distinguish privacy from anonymity. A VPN can change the network path and reduce exposure on an untrusted network, but websites can still recognise accounts, cookies, browser characteristics, and other identifiers. For broader secure remote access decisions, compare the VPN design with your access requirements in this secure remote access checklist.

Worked examples

Example 1: Laptop on home broadband

The baseline shows one public IPv4 address and an IPv6 address. With the VPN connected, the IPv4 changes, DNS results match the configured VPN path, and WebRTC shows no unwanted public address. During a manual disconnect, the browser stops loading until the VPN reconnects. This is a strong result for the tested browser and network. The next check should confirm whether other applications are covered by the same system-wide kill-switch mode.

Example 2: Android phone using split tunnelling

The VPN-protected browser passes the IP and DNS checks, but a video application has been excluded to improve performance. During a tunnel interruption, the browser stops while the excluded application continues using mobile data. This is not necessarily a technical leak: it is the selected routing policy. It is a failure, however, if the intention was to protect every application. Remove the exclusion or use a system-wide mode, then repeat the test.

Example 3: Router with IPv6 enabled

Devices behind the router show a changed IPv4 address, but the original IPv6 address remains visible. The VPN tunnel is working for IPv4, yet IPv6 traffic is taking another path. The corrective action may be to configure IPv6 through the tunnel, disable IPv6 where appropriate, or choose a router and VPN combination that supports the required routing. Do not mark the setup as fully protected until the IPv6 decision is deliberate and verified.

For related encryption checks, see the guide to SHA-256 checksums and software signatures. If your concern is resolver privacy rather than VPN tunnelling, review the comparison of secure DNS options in the UK.

When to recalculate and retest

Repeat the full workflow after installing a VPN client update, changing provider or protocol, enabling split tunnelling, updating the operating system or browser, replacing a router, changing DNS settings, or moving between home, office, public Wi-Fi, and mobile networks. Retest after a security advisory involving the VPN client, browser networking, IPv6, WebRTC, or the router firmware.

A sensible recurring check is to keep a dated record and rerun it whenever the configuration changes, rather than treating a historic pass as permanent proof. For work devices, retest after endpoint-management policy changes and confirm that the organisation’s required applications are included in the protected route. If a test fails, disconnect from sensitive services, capture the configuration and timestamps, disable split tunnelling if appropriate, and contact the VPN provider or administrator before relying on the connection again.

The practical goal is not a perfect-looking test page. It is a documented configuration in which the addresses, DNS path, browser behaviour, and failure mode match what you intended to protect.

Related Topics

#VPN#kill switch#DNS leaks#IP leaks#WebRTC#privacy testing
A

AnyConnect Editorial Team

Cybersecurity Editor

Senior editor and content strategist. Writing about technology, design, and the future of digital media. Follow along for deep dives into the industry's moving parts.