Finding the real path MTU across fiber and cloud VMs
Diagnosing silent TLS stalls over tunneled links with ping and tcpdump.
When small pings succeed over a tunnel but HTTPS handshakes or large JSON responses hang mid-flight, check path MTU before blaming DNS or TLS. An extra header on PPPoE, VLAN tags, or cloud overlay encapsulation often shaves 8 to 80 bytes off the nominal 1500-byte Ethernet frame.
Probe with Don't Fragment set
From the client side, run ping -M do -s 1472 -c 3 <peer> and step down in
increments of 16 bytes until packets pass cleanly. Remember that the payload argument excludes
the 28-byte IPv4 and ICMP headers. Once you find the largest passing payload on the underlay,
subtract 60 bytes for IPv4 WireGuard overhead (80 bytes for IPv6) and set that value explicitly
on the tunnel interface.
Verify with a real transfer
Confirm the fix by fetching a response larger than 16 KiB over TLS and watching
tcpdump -ni wg0 'tcp[tcpflags] & tcp-syn != 0' to check that MSS clamping
matches the tunnel MTU. Document the measured number in the interface comment so the next
kernel or router upgrade does not quietly revert it to 1420.
Did this work on your setup, or did you spot a mistake? Send the editor a private note or leave a public note for the next reader.