On Sat, Aug 7, 2010 at 8:04 PM, seth vidal <skvidal@xxxxxxxxxxxxxxxxx> wrote: > I'll be blunt, though - you'll need some kind of client-authenticated > system to push this data to a common server. In order to keep any client > on the system from being able to fill your server with garbage data. > > A couple of options: > > 1. reverse it and have a client connect in and fetch the data at regular > intervals It's done like that. I have custom data fetching scripts and just tried to reformat the information with a yum-plugin (in a single json-stream which gets collected by some perl-scripts). I abandoned that and now query the history database directly. Perhaps auditing was the wrong word for that. I better call it an enhanced central package log. :) > 2. look at something like func to do the communication. Would be a better start but I already had some infrastructure for data collection. > I agree with as far as exceptions go - we can improve those and if you > see cases, I think that's a fair thing to do. But at some points a > plugin can just go horribly wrong and call sys.exit() or a library can > abort and then you're just hosed. For one, trying to develop a yum-plugin with my low python skills, knowing not to create inconsistencies in the database when doing something dumb, is a good thing, I think. :) Also, if plugins are doing network-IO and dependent servers are currently not available this should not abort the yum transaction, if the plugin does not handle that correctly. hannes _______________________________________________ Yum mailing list Yum@xxxxxxxxxxxxxxxxx http://lists.baseurl.org/mailman/listinfo/yum
