Having tired of babysitting the rsync cron job that was keeping my local Fedora mirror up-to-date, I tried the caching proxy approach suggested at http://fedoraproject.org/wiki/Infrastructure/Mirroring/SiteLocalMirrors for a few weeks. This, too, was unsatisfactory--I still want some control over the mirrored content and the ability to pre-populate the cache from a DVD ISO acquired via bittorrent when a new version of Fedora is released. ApacheMirror.py is a mod_python request handler that behaves like a caching proxy, except it maps the URL path of a cached document directly to a local directory rather than hashing the URL, this preserving the mirror directory structure. Just drop ApacheMirror.py into /usr/lib/python*/site-packages, set your preferred upstream server and point it at a local directory on a nice big disk, and forget it: <VirtualHost *:80> ServerName mirrors.sample.com ServerName mirrors DocumentRoot /mirrors SetHandler mod_python PythonHandler ApacheMirror PythonDebug on PythonOption ApacheMirror.upstream http://download.fedora.redhat.com </VirtualHost> The implementation is by no means bulletproof--consider this release 0.1--but it's worked well enough to serve local yum needs for the past few days. If there's interest, I could package up the script into an srpm (which seems overkill for 50 lines of Python) or submit it as a patch to some existing package. --Ed
Attachment:
ApacheMirror.py
Description: Binary data
-- fedora-devel-list mailing list fedora-devel-list@xxxxxxxxxx https://www.redhat.com/mailman/listinfo/fedora-devel-list
