On 1/15/19, 3:53 PM, "Geoff Huston" <gih@apnic.net> wrote: Does this have anything to do with the observation that A and J truncate large IPv6 UDP responses No, I don't think so. In the data that I've look at so far, it's been IPv4 sources exhibiting this rollover-and-die[1] behavior. We do observe an increase in truncation during ZSK rollovers. But at revocation, the overall truncation rate dropped a little, almost certainly due to the increase in volume from misbehaving sources. Admittedly that could be masking truncation from other sources so I'll take a closer look. DW [1] I wish they would.