Comments on the fastestmirror plugin
Fred Williams
dukederf at googlemail.com
Fri Mar 12 16:47:25 UTC 2010
On 12 March 2010 16:42, Patrick O'Callaghan <pocallaghan at gmail.com> wrote:
> On Fri, 2010-03-12 at 14:21 +0000, Fred Williams wrote:
> > On 12 March 2010 14:08, Patrick O'Callaghan <pocallaghan at gmail.com>
> > wrote:
> > The yum fastestmirror plugin (yum-plugin-fastestmirror) claims
> > to
> > evaluate the speed of a bunch of repo mirrors and use the
> > fastest one
> > relative to the user's location.
> >
> > However AFAIK what it *actually* does is make a test
> > connection to the
> > to the candidate mirrors and order them according to response
> > time,
> > which in many cases is dominated by network latency, which can
> > distort
> > the results. For well-connected user machines in first-world
> > countries
> > it probably doesn't matter much, and may have the beneficial
> > effect of
> > spreading the load over a wider range of mirrors, but for
> > those of us in
> > a less privileged position it can matter a lot. Ironically,
> > these are
> > the cases where such an optimization could do the most good.
> >
> > A case in point: I live in Venezuela and on several recent
> > occasions yum
> > decided that my closest repo was in Puerto Rico, which as the
> > packet
> > flies is probably true. However the b/w I got as a result was
> > around 2
> > or 3kbps.
> >
> > I tried renewing the mirror cache. No difference (ping times
> > tend not to
> > vary much).
> >
> > I then manually edited the /var/cache/yum/timedhosts.txt file
> > to bias
> > the results against the mirror yum was choosing (I made it
> > worst rather
> > than best). Oddly, it again made no difference! It seems
> > there's a
> > cunning hidden cache of these results that I don't know about.
> > Finally I
> > disabled the plugin completely and got decent b/w without it.
> >
> > Perhaps we should be considering some kind of BitTorrent
> > version of the
> > repos in which the mirrors are seeds and the users are
> > leeches, though I
> > realize that this is harder than it looks, particularly when
> > taking into
> > account the synching of the mirrors themselves.
> >
> > poc
> >
> > --
> > users mailing list
> > users at lists.fedoraproject.org
> > To unsubscribe or change subscription options:
> > https://admin.fedoraproject.org/mailman/listinfo/users
> > Guidelines:
> > http://fedoraproject.org/wiki/Mailing_list_guidelines
> >
> > Perhaps not so difficult - though I've never used them myself, I
> > recall that in the Debian, if not Ubuntu (Sometimes hard to tell)
> > repositories are some packages that allow for bittorrent fetching of
> > deb packages - perhaps if they're still relevent and working, they
> > could be used as a base to create a means of implementing the same,
> > maybe as a plugin for Yum or similar.
> > Theoretically, I think the only main differences are the download
> > protocol. HTTP/FTP or BitTorrent. Once downloaded the package can
> > still be used in the same way, there's no difference there.
> > The main downside I see to it is that those users on an ISP which
> > throttles BitTorrent will suffer, and have to go back to standard
> > downloads, but if both are provided, then no issue. Or at least very
> > little.
> > Just my 2p. Or 2c, depending on your currency.
>
> Interesting. I'll see what I can can find on BT use in the Debian/Ubuntu
> world.
>
> poc
>
> --
> users mailing list
> users at lists.fedoraproject.org
> To unsubscribe or change subscription options:
> https://admin.fedoraproject.org/mailman/listinfo/users
> Guidelines: http://fedoraproject.org/wiki/Mailing_list_guidelines
>
A quick search for 'deb' on the Debian Package database returned a lot of
results but the specific one that matches would be this one, I believe:
http://packages.debian.org/lenny/debtorrent
At a glance, it's hard to say how useful it'll be, even as an example to
work from.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.fedoraproject.org/pipermail/users/attachments/20100312/23fa453d/attachment-0001.html
More information about the users
mailing list