> I guess I have to run emerge sync and emerge -uD world on each client > still, > but they use the http replicator instead of gentoo servers.... for both > updating the portage/rsync files and downloading the distfiles (via the > http > replicator)?
In the case of distfiles and sync's, you're still using primary gentoo servers for the first fetch; the cached files are used after that. Basically emerge is going to want a file foo.tar.bz2 to complete an ebuild; it builds a url in the form of http://gentoo.mirror.com/gentoo/ that comes from /etc/make.conf and adds the necessary components "/distfiles/foo.tar.bz2". It then calls wget to retrieve the components. Since the http-replicator proxy is in use, when wget calls through the proxy it will see if foo.tar.bz2 is in the cache; if so it returns the cached copy. If not it is retrieved from the url provided by wget so gentoo.mirror.com is used for the retrieval. For sync the primary server handles the emerge --sync against the official gentoo mirror; internal systems sync against the local server, but the data still originates from the official gentoo mirrors. > I still have to run 'emerge -uD world' on the server/workstation > manually each day? Each system is still administrated as if it is a standalone box; so you can choose to or not to "emerge -uD world" on a system by system basis. My primary rsync/http-replicator server is also my mail server, web server, firewall, etc. I try not to "emerge -uD world" on it until I really have to. It's your decision on your end when to do it on the server. -- gentoo-user@gentoo.org mailing list
