Ove Kaaven wrote:
> But I seem to recall that the problems were not lack of volunteering
> sites, but something more technical...?
Indeed. The problem is that the app-database, CVS browsing and
source-browsing are on a fixed system (tiepmiep) linked through scripts.
Mirroring the apps-db can only be solved if we create a distributed
database (kinda difficult). Mirroring the CVS browsing (and
checkout/update/diff/etc...) means that the CVS-tree must be accessible
on the mirror-machine, without hardlinks (like now 'cvs.winehq.com').
Mirroring the source-browser can be done if we convert the lot from a
script to real pages (already suggested that for performance reasons);
we do not need to keep every release as source because it takes about
20M per snapshot release to store on the computer (maybe these should be
created automatically on each mirror instead of dragging this over the
net).
With these changes we can simply rsync the entire site from designated
masters or use CVS updates as done now (should mirror the CVS too to
keep traffic to the master-master site down).
Greetings Bertho
