A few notes regarding usage of wget for mirroring sites. 

In general I think this is a great tool.  It's well thought out, has
loads of options, and doesn't have the overhead that encrypted
transfers often have.  Plus, it's wonderful that http servers rarely, if ever,
refuse a connection attempt.  

So here was my wget to collect the Redhat Updates for v8.0 from
ibiblio.org:

wget -bvcm -o wgetlog -nH  --progress=dot:mega http://distro.ibiblio.org/pub/Linux/distributions/redhat/updates/8.0

This would have been a better link, as I'm getting some things I don't
really need = 3 architectures and 1 a set of source rpms.
/usr/mirror/redhat/pub/Linux/distributions/redhat/updates/8.0/en/os/i386

Oh well...

-b puts the process in the background.
-v makes the output verbose
-c isn't really required for a first download, but on subsequent
downloads it will pick up partial file transfers where they left off.
-m is for mirroring and specifies a bunch of things including
recursion.
-o specifies the log file
-nH disables generation of host directories, i.e. a directory named
with the hostname of the target at the top of the structure.