> Might be that your case would be handled just automatically. Are > 208.85.240.29 & 208.85.242.118 your two mirrors? (That's what > mirror.fast-serv.com resolves to.) > > Okay, let's see :-) > > The two mirrors are in AS29889, for which the following prefixes are > announced: > > select * from pfx2asn where asn = 29889; > pfx | asn > -----------------+------- > 74.115.208.0/22 | 29889 > 74.115.212.0/22 | 29889 > 208.85.240.0/23 | 29889 > 208.85.242.0/23 | 29889 > 209.9.238.0/24 | 29889 > > This is what GeoIP currently has to say about these network: > > 1) 74.115.208.0/22 AS TW Taipei > 2) 74.115.212.0/22 NA US Maryland Crownsville 39.030102,-76.606903 > 3) 208.85.240.0/23 NA US Maryland Crownsville 39.030102,-76.606903 > 4) 208.85.242.0/23 NA US California Escondido 33.134899,-117.041603 > 5) 209.9.238.0/24 NA US Virginia Herndon 38.984100,-77.382698 > > So we have: > > 1) I'm not sure if that is a GeoIP glitch regarding 74.115.208.0/22, > claiming that it is in Taipei? Might that be wrong? Anyway, clients from > that range will be sent to one of your two mirrors, because they are > in the same AS (unless another mirror in the same IP range exists). > > 2) 74.115.212.0/22 clients will be sent to your mirror in Maryland. None > of your mirrors is in the same IP range, both are in the same AS, and > the Maryland mirror is preferred because it is geographically closer. > Which should be what you want, right? > > 3) 208.85.240.0/23 clients will be sent to your mirror in Maryland, > because it is in the same IP range. > > 4) 208.85.242.0/23 clients will be sent to your mirror in California, > because it is in the same IP range. > > 5) 209.9.238.0/24 clients are sent to your mirror in Maryland. None > of the two mirrors is in that IP range, but both are in the same AS, > and the Maryland is geographically much closer, while the other one > is over 2000 miles away. MirrorBrain picks the one in Maryland therefore. > > Is that the situation, or did I guess it incorrectly? Are there further > networks that need consideration? Are the results which I described > above how you want it to be? > Other than the taiwan glitch (should be california) looks good. I was thinking there may have been some aggregation on the CIDRs which isn't the case (good!). Like I mentioned previously, as long as both our mirrors are returned for any IP in AS29889 fastestmirror will easily be able to choose the correct mirror for our clients regardless of GeoIP issues. -Randy