[Yum] Upgrade in place succeeded, I think...

[Michael Stenner]

On Tue, Jun 18, 2002 at 01:34:19PM -0400, Robert G. Brown wrote:
> Any suggestions on how I might "cheat" the upgrade of the rest of my
> local hosts so they don't have to re-download rpms?  For example, I
> could make an actual scp of the rpm's on the target host's
> /var/cache/yum/server/packages.  Or I could export and make a rw,
> no_root_squash mount of /var/cache/yum/server/packages from lucifer
> (done) to archangel (waiting to be done) which might fool archangel into
> thinking that it has the rpm's long enough to complete the update.  Or I
> could move all the packages into a local repository and direct yum to
> use it as a source, first, if it could, if I had any idea how or if it
> could do such a thing.

I'm thinking seth might not be really excited about this, but how
about an option to yum that makes it update its cache of headers
and/or rpms without installing/upgrading/removing anything?

This might be nice to if you want to have it do all of the slow stuff
overnight, but don't quite trust it enough to let it do everything.
You can get SOME of this functionality by just waiting until morning
to hit 'y', but I think it still needs to get the rpms, then, right?

I think it would be reasonable to have yum automatically remove rpms
from the cache after they are successfully installed.  You'd also need
to provide a --clean-cache command though in case people never
followed through on the install.  These would be slightly more
advanced features, but I wouldn't think they'd be too hard to
implement.

Related, and this might address rgb's comments better... you could
provide multiple caches for yum to look in for rpms, perhaps giving
each of them an optional 'noremove' flag for shared caches.

					-Michael
-- 
  Michael Stenner                       Office Phone: 919-660-2513
  Duke University, Dept. of Physics       mstenner at phy.duke.edu
  Box 90305, Durham N.C. 27708-0305