On Wed, 2006-06-14 at 10:44 -0400, Tony Nelson wrote: > At 12:54 AM -0500 6/14/06, Les Mikesell wrote: > ... > >Yum, on the other hand, sees a current list of mirrors and > >should be able to do something intelligent, not psychic > >with them that would pick the same one for everyone at a > >location. Trying the one closest alphabetically to the > >local domain name would be better than nothing but there's > >probably a way to do a real geo-ip match for the mirror > >nearest the proxy. > > Alphabetically, hmm? That would be "closest to L", for localhost, and we'd > all be using the same mirror. Except you, of course. Your internet proxy probably has a problem talking to localhost as well... > You are asking yum and your web proxy to be psychic, and it won't happen. > Either do as I suggested, or set up your own repository that you control, > or quit bitching. I've got a reasonable amount of bandwidth here. I just don't see why the people running the mirrors enjoy sending many times the number of copies that would be needed if existing cache mechanisms were not thwarted by randomizing the URLs the way yum does it. I'm sorry you consider pointing out this waste of resources as bitching. Round-robin DNS would work if the repositories were in the same relative location. Any scheme that would more frequently choose the same mirror from the same location would work better than filling caches with randomly chosen URLs that won't be reused. -- Les Mikesell lesmikesell@xxxxxxxxx
