* Dan Price <[EMAIL PROTECTED]> [2008-07-21 22:26]: > On Mon 21 Jul 2008 at 02:55PM, [EMAIL PROTECTED] wrote: > > > > > Could you expand on how you see the depot UUID being used in a load > > > > > balancing situation? (For instance, pkg.opensolaris.org is load > > > > > balanced across a set of identical depots which use a common > > > > > catalog, > > > > > package manifests, and files, but keep separate statistics, and are > > > > > hosted on multiple systems. With what aspect of a depot's operation > > > > > is the UUID associated?) > > > > > > > > At the moment, it it is to provide a unique identifier for the RSS/Atom > > > > feed generate by the repository. > > > > > > > > All of the pkg.opensolaris.org servers would need to share the same id > > > > so that the feed generated by any of them would be considered the same > > > > one by feed readers. > > > > > > Why not just disable the feed on all but the primary? Since it's > > > 100% cacheable it doesn't seem like a big deal. > > > > In this example, how would you determine the webserver that is the > > primary? > > Umm, I don't know... are we having the notion of the authoritative > ultimate source of knowledge amongst a group of mirrors? If not, > then I guess never mind.
I think, to answer K's question, that the primary is known to the operator of the set of repositories. Note that these aren't independently accessible mirrors, like we've talked about for wider mirroring, but a set hidden behind a load balancer or a reverse proxy. - Stephen -- [EMAIL PROTECTED] http://blogs.sun.com/sch/ _______________________________________________ pkg-discuss mailing list [email protected] http://mail.opensolaris.org/mailman/listinfo/pkg-discuss
