I've noticed some strange issues with PMTUD (Path MTU discovery) on my Plusnet connection. When pinging my public IP from outside of my network with packets larger than my PPPoE MTU (which is 1492), the result differs depending on which AC I am connected to.
With some ACs where PMTUD is working, I see the expected fragmentation needed response:
$ ping -c 10 -M do -s 1465 <my WAN IP>
PING <my WAN IP> (<my WAN IP>) 1465(1493) bytes of data.
From 84.93.253.82 icmp_seq=1 Frag needed and DF set (mtu = 1492)
From 84.93.253.82 icmp_seq=2 Frag needed and DF set (mtu = 1492)
From 84.93.253.82 icmp_seq=3 Frag needed and DF set (mtu = 1492)
From 84.93.253.82 icmp_seq=4 Frag needed and DF set (mtu = 1492)
From 84.93.253.82 icmp_seq=5 Frag needed and DF set (mtu = 1492)
From 84.93.253.82 icmp_seq=6 Frag needed and DF set (mtu = 1492)
From 84.93.253.82 icmp_seq=7 Frag needed and DF set (mtu = 1492)
From 84.93.253.82 icmp_seq=8 Frag needed and DF set (mtu = 1492)
From 84.93.253.82 icmp_seq=9 Frag needed and DF set (mtu = 1492)
From 84.93.253.82 icmp_seq=10 Frag needed and DF set (mtu = 1492)
--- <my WAN IP> ping statistics ---
10 packets transmitted, 0 received, +10 errors, 100% packet loss, time 9013m
But when connected to other ACs, I see the packets just being dropped:
$ ping -c 10 -M do -s 1465 <my WAN IP>
PING <my WAN IP> (<my WAN IP>) 1465(1493) bytes of data.
--- <my WAN IP> ping statistics ---
10 packets transmitted, 0 received, 100% packet loss, time 9213m
The only AC I can actually see this working on is 250.core.plus.net (195.166.130.250), the rest just result in packet loss.
I noticed this as I am trying to replace my MikroTik router with a Linux machine, but I'm seeing a lot of packet loss for larger packets when using my Linux router (e.g. HTTPS packets). I'm not seeing this on the MikroTik so I assume it's doing something clever (if anyone know what this might be I'd love to know).
Does anyone know what might be going on here? The ACs I've tried are 250, 254 and 255, with only 250 having working PMTUD.
