[Rd] Set up new CRAN mirror; but have questions before finalizing.
hb at biostat.ucsf.edu
Sat Feb 19 19:00:29 CET 2011
On Fri, Feb 18, 2011 at 10:43 PM, Friedrich Leisch
<friedrich.leisch at stat.uni-muenchen.de> wrote:
>>>>>> On Fri, 18 Feb 2011 22:00:28 -0600,
>>>>>> Paul Johnson (PJ) wrote:
> > Hi, everybody
> > I have an account on Dreamhost.com and when I renewed it recently,
> > their message said my usage of storage and bandwidth had been
> > reasonably low. In an idle moment about 3 weeks ago, I followed your
> > instructions to set up a CRAN mirror on their server. Here it is:
> > http://www.freefaculty.org/cran
> > It is not hosted at my University, but it is a working, high
> > availability server. Is there any reason it could not be listed as a
> > CRAN mirror. (Although I really have no idea where these machines
> > exist. I'm pretty sure it is in the USA. I'll try to find out).
> We would need to know where it is because we lsit mirrors in the US
> sorted by state ...
> > Maybe you might try it and see? It has updated several times, no
> > trouble I can see in that.
> > I have a couple of small details to ask about. Maybe this first one
> > is a potential "bug report" for the CRAN mirror instructions.
> > 1. Permissions on "src" and "web" folders are 700,
> Then you must have made a mistake in setting up the mirror, because on
> the master we have 775 for both directories and all directories within
> these two. We also had never complaints about this before.
> > and so running "update.packages" or an apt-get update against the
> > debian stuff results in permission denied errors. I re-set the
> > permissions manually, but wonder if I'm actually supposed to mess
> > around with your archive. After doing that, it works. But I
> > worry a little bit about what else might not be readable "down
> > there" in the hierarchy. And I wonder why any body else's mirror
> > works without doing that.
> Well, they simply mirrored our permissions ...
> > 2. When I run "apt-get update" against my mirror, i get a lot of
> > harassment about the lack of a security key for my repository. Should
> > I be publishing an R Core team key to fix that, or my own key to do
> > what? I've never administered an apt repository before. I have
> > administered yum repositories and the security there is in the key on
> > the individual RPMS, I don't quite understand what the Debian thing is
> > asking me to do.
> It is quite common then non-official Debian mirrors have no security
> key, I would not worry too much about that.
> Best regards,
> Fritz Leisch
> R-devel at r-project.org mailing list
More information about the R-devel