On Mon, 2005-04-25 at 12:15 -0700, Jesse Keating wrote: > On Mon, 2005-04-25 at 14:33 -0400, Jim Popovitch wrote: > > You sort of just described such a system. It makes good sense to have > > one or two production servers auto-update. > > No. Period. The autoupdating ones are the ones in the lab. Running as > close to a config as the production systems as possible. THESE can > auto-update for testing purposes. No, or yes. You start out saying "no", but end up saying "these can". > > Those sames systems can then > > be quickly tested and then used as the update server for the internal > > system. > > Bad idea. Don't autoupdate any production level system. If you can't > have it go down at any time, don't allow auto updates. Define "production level system" In my view, *any* system that is used in the course of business (be it for patch testing, or end-user facing) is a production system. I suppose that you can make the distinction between levels of systems (dev, test, acceptance, production?), but in my mind all those systems need to be up to vendor patch level else testing of a vendor patch is fruitless given the non-consistent environment. ...snip... > No changes happen to production servers here without this. Define "here". What is the environment there? -Jim P. -- fedora-legacy-list@xxxxxxxxxx http://www.redhat.com/mailman/listinfo/fedora-legacy-list
