slow distfiles mirror

[Daniel J. Luke]

On Aug 8, 2016, at 1:31 PM, Brandon Allbery <allbery.b at gmail.com> wrote:
> On Mon, Aug 8, 2016 at 1:24 PM, Daniel J. Luke <dluke at geeklair.net> wrote:
> I'm interested in what unique network situation causes the rtt test to be an unreliable indicator of which mirror to choose.
> 
> I have a feeling you misunderstood that.

Ryan said: "it's possible for a server to respond quickly to pings but slowly to actual file transfers"

I'm interested in what prompted that.

> Everyone has a unique situation which can include broken local stuff (I have to go through a local network not under my control that likes to have 80% loss on "ping" at various times of the day, all incurred locally),

Which (if you are experiencing poor mirror selection problems) is a good data point (ie, the rtt test maybe can't reliably use ICMP echo and needs a fallback or alternative strategy).

[snip speculation about local internet provider that may or may not do 'interesting' things to traffic]

-- 
Daniel J. Luke