Strangeness with recent FC5 update
fct-cornette at insight.rr.com
Wed Jul 12 11:14:28 UTC 2006
Michal Jaegermann wrote:
> On Tue, Jul 11, 2006 at 06:00:37PM -0400, Jim Cornette wrote:
>> Michal Jaegermann wrote:
>>> When problems occur this is because for reasons yet unknown, I
>>> think, yum decided to erase first an old version of a package (at
>>> this step all these .rpmsave renames happen) followed by an update.
>>> You should see instead an update step followed by a cleanup.
>> Thanks! I checked my system for any rpmsave files and none were found.
> There are situations when you can see a legitimate .rpmsave file (an
> old config file) _and_ a new replacement. You may want to carry
> over some customizations from an old file to a new one. The other
> possibility is that you deliberately removed some package and
> changes you did are not automagically deleted.
Periodically I clean up files manually such as *#prelink#* files and
rpmsave and rpmnew files. With the problem some are experiencing with
yum removing rpms and then somehow not at least completely installing
all the files that should be included if the package was initially
installed, I'll investigate the situation further before removing the files.
Usually I'll check the rpms for signs of missing files and
system-config-* rpms seem to show missing files from their rpm query.
I'll remove the db entry then reinstall the rpm again. The next
occurrence of missing files, I'll note any rpmsave files present.
Pasted is the last files missing episode and .pyc files seemed to be the
> In this particular situation we are talking about the trouble is
> that you may end up, say, with /etc/inittab.rpmsave but no new
> /etc/inittab at all. Such event may indeed cramp your style a bit.
:-) - quite a bit by some postings earlier.
>> There are a few rpmnew files for repositories, unrelated of course.
> Here you have new configuration files but your old ones should be
> still "good enough". Regardless you may want to check later how big
> those changes are and either adjust what have already (some new
> options or defaults may show up, for example) or simply remove
> those .rpmnew to keep things clean.
> And there is also a multilib situation where installed packages
> for different architectures may produce .rpmnew. A bit of a
> nuisance but easy to check.
Basically, I want people to know that when they use binary-only modules,
it's THEIR problem. I want people to know that in their bones, and I
want it shouted out from the rooftops. I want people to wake up in a
cold sweat every once in a while if they use binary-only modules.
-- Linus Torvalds
More information about the test