A VPN connects, ping works, small web pages load — but large downloads hang, file shares freeze and some HTTPS sites never finish loading. This pattern is almost always an MTU problem: somewhere on the path, packets are larger than a link can carry, and the signal that should tell the sender to use smaller packets is being filtered.
The MTU Test finds the largest packet that crosses your connection without fragmentation, calculates the exact overhead of VPN and tunnel stacks, and generates MTU and TCP MSS clamping commands for common routers and operating systems.
What is MTU?
In simple terms
The MTU (Maximum Transmission Unit) is the largest packet a network link can carry in one piece. Standard Ethernet carries 1500 bytes. Every tunnel or extra header — PPPoE, a VPN, VXLAN — takes some of those bytes, so the packet that fits inside is smaller.
Technical background
Path MTU is the smallest MTU of all links between two hosts. Path MTU Discovery (RFC 1191 for IPv4, RFC 8201 for IPv6) sends packets with the Don't Fragment bit set; a router that cannot forward one returns ICMP "Fragmentation needed" (IPv4) or ICMPv6 "Packet Too Big", naming its MTU. When those ICMP messages are filtered, the sender never learns the right size and large packets disappear — a PMTU black hole. TCP avoids the problem when the MSS (maximum segment size) advertised in the SYN fits the path, which is why routers "clamp" the MSS of connections crossing a tunnel.
Why use an MTU test?
It tells you whether your connection carries full-size packets, which MTU to configure on a tunnel, and whether stalls are caused by a black hole rather than by bandwidth or packet loss. Checking before deploying a VPN avoids the classic "works for ping, breaks for everything else" outage.
Who should use it?
Network administrators setting up WireGuard, IPsec, OpenVPN, GRE or VXLAN; ISP and helpdesk staff troubleshooting PPPoE and fibre customers; home users whose VPN loads some sites but not others; and anyone who needs to configure MSS clamping on a router or firewall.
How the MTU test works
The NetTools server sends ICMP echo requests with Don't Fragment set to your own public IPv4 address — the address your request arrives from, usually your router or NAT. No other target can be entered. It first sends a small baseline packet; if that gets no reply, the result is BLOCKED. It then tries 1500 bytes and, if that fails, narrows the size with a binary search, using any "Fragmentation needed" reply to jump straight to the reported MTU. Each size is sent twice and retried once; a missing reply is never treated as proof that a packet was too big.
The HTTP transfer test downloads 32 KB to 2048 KB through your browser. The manual test gives you ping commands for Windows, Linux and macOS to measure from your own device, which is the most reliable result. The calculator and configuration generator run entirely in your browser. For loss and latency along the route, use MTR or Traceroute.
How to use the NetTools.tech MTU Test
- Open the MTU Test and click Run MTU test on the MTU Discovery tab.
- Read the result: the path MTU, its source (Measured, Observed, User supplied or Calculated) and the black-hole assessment.
- Optionally run the HTTP transfer test, and the manual test from your computer — enter the largest payload that worked to get the MTU.
- Switch to MTU & VPN Calculator, choose your tunnel technology and options, and read the effective MTU and MSS with a byte-by-byte breakdown.
- Open Configuration Generator, choose your platform and interface, and copy the commands. Export the session as PDF, CSV or JSON, or print it.
Understanding the results
Each method shows a status. SUCCESS means a value was determined. BLOCKED means the target does not answer ping. INCONCLUSIVE means larger packets vanished without any ICMP error, so the value is likely but not certain. UNSUPPORTED means the method cannot run on this deployment (TCP MSS observation and IPv6 probing). NOT TESTED means you have not run it. A value of "≥ 1500" means the path carries at least the largest packet the probe server can send. Black-hole status ranges from "No indication" to "PMTU black hole suspected"; suspected means small packets arrive, larger ones vanish and no ICMP error ever came back.
Example: WireGuard over PPPoE
A home connection uses PPPoE, so its MTU is 1500 − 8 = 1492. The test shows 1492 with a "Fragmentation needed" reply reporting 1492. WireGuard over IPv4 adds 60 bytes (20 IP, 8 UDP, 32 WireGuard), so the calculator gives a tunnel MTU of 1432 and an IPv4 MSS of 1392. The default of 1420 also works here; 1440, the right value on plain Ethernet, would be 8 bytes too large.
Common use cases
- Choosing the MTU for a new WireGuard, IPsec or OpenVPN tunnel.
- Explaining why large downloads or HTTPS pages hang over a VPN.
- Checking that a PPPoE line has the expected 1492 bytes.
- Calculating VXLAN or GENEVE overhead before raising the underlay MTU.
- Generating MSS clamping rules for a router or firewall.
Troubleshooting
BLOCKED
Your router drops ping from the internet. Use the manual test from a computer on your network, which measures the path in the other direction and needs no inbound ICMP.
INCONCLUSIVE and black hole suspected
ICMP "Fragmentation needed" messages are filtered somewhere. Allow ICMP type 3 code 4 (and ICMPv6 type 2) through firewalls, lower the tunnel MTU to the measured value, and clamp TCP MSS on the router at the tunnel edge.
Small transfers work, large ones stall
Lower the MTU of the tunnel interface or enable MSS clamping. Compare with a speed test: a black hole stalls completely, while a slow link just takes longer.
Manual test shows a lower value than the server test
The manual test measures from your device, including Wi-Fi, local VPN clients and the outbound path. Trust the lower value.
Common mistakes
- Lowering the MTU on every interface "to be safe" instead of only on the tunnel.
- Blocking all ICMP on a firewall, which breaks Path MTU Discovery.
- Setting MSS to the MTU value on platforms that expect MSS (pfSense and OPNsense are the exception: their MSS field takes the MTU).
- Forgetting IPv6, which needs at least 1280 bytes and MSS clamping of its own.
- Copying the 1420 WireGuard default onto a link that already loses bytes to PPPoE and another tunnel.
Limitations
The server probe ends at your public IPv4 address, usually your router, and does not include your local network. IPv6 probing and TCP MSS observation are not possible on this deployment and are reported as UNSUPPORTED. Many routers do not answer ping, so BLOCKED is common. The transfer test runs over HTTPS through Cloudflare, so it shows whether large transfers complete, not the exact MTU. Calculator results assume the standard header sizes of each protocol; vendor extensions can add bytes.
Security & privacy
The probe target is never user supplied: only the address your request comes from is tested, with at most 24 sizes of 2 packets each, one running test per visitor, a per-address budget and the normal rate limits. Generated commands are shown for you to review and are never executed. Results are kept in memory for 15 minutes and not stored.
Related tools
- Ping — check reachability and round-trip time of any host.
- Traceroute — see the routers on the path to a host.
- MTR — combine traceroute and ping to find loss on a path.
- Network Jitter Test — measure latency variation.
- Internet Speed Test — measure download and upload speed.
Summary
A wrong MTU breaks connections in confusing ways: ping works, large transfers do not. The NetTools.tech MTU Test measures the path MTU towards your connection, says clearly how each value was obtained, calculates tunnel overhead to the byte and generates the MTU and MSS settings to fix it.
Frequently asked questions
What MTU should I use?
The path MTU: 1500 on plain Ethernet, fibre and most cable links, 1492 on PPPoE (many DSL and some fibre lines), and lower inside tunnels — 1420 for WireGuard is the usual default. The calculator gives the exact value for your stack.
Why does the test say BLOCKED?
Your router does not answer ping from the internet, so the server cannot probe sizes. That is common and not a fault. Run the manual test from your own computer instead.
What is the difference between MTU and MSS?
The MTU is the largest IP packet a link carries. The TCP MSS is the largest TCP payload in one segment: the MTU minus 40 bytes for IPv4 or 60 bytes for IPv6.
Why can NetTools not see my TCP MSS?
NetTools.tech is reached through Cloudflare and a reverse proxy, which terminate the TCP connection. The probe server never receives your SYN, so the tool reports this as UNSUPPORTED instead of guessing.
Is a lower MTU always safer?
No. Too low an MTU wastes bandwidth on headers and can break IPv6, which needs at least 1280 bytes. Use the largest value that crosses the path without fragmentation.
Why is 1420 the WireGuard default when my calculation says 1440?
wg-quick assumes an IPv6 outer path, which costs 80 bytes. Over IPv4 only, WireGuard needs 60 bytes, so 1440 fits a 1500-byte path.