lucy24 - 6:40 am on Sep 13, 2013 (gmt 0)
If I know the requester, what can I tell him/her to do differently?
Are you on close speaking terms with the requester's browser? Because that's who you have to talk to.
:: detour here to pore over headers associated with 206 responses of my own-- unfortunately only available for page requests ::
Officially there are a couple of different ways to trigger the 206 response. In practice you see it when the request includes a "range" header. It doesn't always reflect a gigantic file: one robot I could name routinely logs a 206 in requests for robots.txt. (Really. Just how big do they think my robots.txt might be?!) Similarly, facebook requests for pages --not images, which you'd think would be vastly bigger-- almost always come through as 206.
Do you have any way of logging the request headers? At multiple MB I assume we're talking about something other than pages. The only question the server can really answer is whether it's sending out correctly sized chunks. For example if the Range header says 0-50000 and the actual file is much bigger, is each response about 50k? If it's significantly smaller the server is doing something wrong, but otherwise it's doing exactly what it has been told to do.
:: memo to self: check whether piwik's .svg file-- whose existence I never suspected-- still elicits a 206 if I connect with a different browser ::