Treehugger's problem looks like a network issue, but probably something outside of her control. I'll need to have a chat with rarified, but I don't really think it's something at his end either. There seems to be a "bad boy" in the route
.
Treehugger got a Wireshark capture for me of the data passing in/out of her PC while starting Minkata. What I can see is that during the updating phase she had 247 dropped packets, 116 suspected retransmitted packets and 8 out-of-order packets. All of these are things that can "just happen" with internet communication and the protocols deal with it, but usually they're in quite small numbers. For comparison, when I did the same data capture I had 7 dropped packets, 0 suspected retransmissions and 7 out-of-order packets.
Whenever a dropped packet occurs it gets noticed and there's a request to resend: In treehugger's case, after each dropped packet there are a LOT of duplicate ACKs being sent. Since the problem doesn't appear to manifest on MOULa then I'd conclude that the problem host doesn't get used on the route there. I think some further digging into what's happening in the respective routes is needed.