Moin > How does the route look like where the path MTU is saved? > netstat -rn has a Mtu column. Just noticed i sent route -n -T0 get instead of netstat -rn;
gw02.dus01.as59645.net ~ # route -T0 exec netstat -rn | grep 2a06:d1c0::b 2a06:d1c0::b/128 2a06:d1c0::dead:beef:a01 UG 0 5149063 - 48 vio0 (There is a fulltable in there; so with grep.) > Packet size 2078 seems large. Do use jumbo frames? > On which machine did you make the tcpdump? OpenBSD? > > You should disable TCP Segmentation Offload. Otherwise you never > know the packet sizes on the wire. > sysctl net.inet.tcp.tso=0 > Note that OpenBSD supports Large Receive Offload only on ix(4). > Other hardware interface don't do it. > ifconfig ix0 -tcplro > On Linux you can use ethtool to disable offloading. > > Does packet size change in tcpdump when you turm off TSO? pcaps with and without tcpdump (and using plain tcp with nc instead of bgpd) are here: https://rincewind.home.aperture-labs.org/~tfiebig/mtu_issue_obsd_tso_sm all.pcap https://rincewind.home.aperture-labs.org/~tfiebig/mtu_issue_obsd_notso_ small.pcap Packetsize does not change. I also just noticed that this only occurs when I use the loopback as the source address on gw02.dlft01, which is bound to lo. Weird though; Is there MTU information on NLRI exchanged with BGP? Lemme check... With best regards, Tobias
