Gerd v. Egidy wrote: >>> In case of RHEL all this yum stuff doesn't really matter because you'll >>> use the RHEL update tools like satellite. >> >> I found satellite to be unusable for modest environments. The >> interface was very slow and unscriptable for our favorite upstream >> vendor's GUI, and investing in the Oracle database to run spacewalk >> locally was unconscionably expensive for even midsize environments. >> Coupled with the slow installation process and update downloading over >> thin network pipes for small clusters, it was unusable. > > Thanks for sharing your experience. I've tried it just once with a RHEL demo > and haven't checked all the features. > >>> So I propose to include it in rf that all users who like to use >>> yum-updatesd have an easy way to install it. >> >> Why? yum-cron, on a *nightly* basis, can do very similar work and not >> interfere in the middle of the day with other 'yum" based operations. > > That depends on the usecase. > > I want to have all updates installed within 1 hour after publishing them on my > local yum repo. But not all servers should install their update at the same > time and I want to keep their configuration as identical as possible. So no > crontabs with different yum-crons, the update time is just spread by different > boot times.
in centos 6 anacron is used for regular jobs. In fact cron.daily, weekly and monthly are run through anacrontab by default now. It allows you to add a random delay before starting jobs. You specify the max delay. Maybe check it out, seems like it could solve your problem. _______________________________________________ users mailing list [email protected] http://lists.repoforge.org/mailman/listinfo/users
