FC5 rpm database corrupt [Solved]
Paul Howarth
paul at city-fan.org
Tue Jun 13 07:58:18 UTC 2006
On Tue, 2006-06-13 at 02:42 +0100, Keith G. Robertson-Turner wrote:
> Claude Jones wrote:
> > On Sunday June 11 2006 22:24, Claude Jones wrote:
>
> >> Yum is still not working quite right. It just supposedly updated
> >> bind and the kernel and about four other packages, and when it was
> >> finished, I came up with the same exact packages needing to be
> >> updated on the next try...
>
> > I seem to have prevailed. Actually, I must give credit where it's
> > due - the final clean-up was carried out by PUP. I'd been using
> > Yumex
>
> Ahaaa!!!
>
> Funny (not really) that you should say that, because I've had two
> machines b0rked after a mass update using Yumex this week, on a
> network of seven machines all config'd to update from the same
> in-house "yam" server (yes yam with an "a"). The other 5 were fine,
> including the server itself, *because I updated them using just yum
> and not yumex*.
>
> The type of errors I was getting with yumex were:
>
> "scriptlet failed: error 250" or similar (sorry I don't remember the
> error number) over and over on every single updated package. The
> result was the RPM database became out of sync with the installed
> packages. This *only* happened using Yumex, never when just using
> yum. I had to mass-uninstall then reinstall huge portions of the
> distro to get it back in sync. Eventually I decided the installations
> were so b0rked, that the safest thing to do was wipe and start again.
I believe that the cause of this particular problem is that the FC5
out-of-the-box SELinux policy does not assign the correct context type
to yumex and it does not execute the rpm scriptlets in the correct
SELinux domain as a result, which causes the failures you've seen.
Updating the SELinux packages (and possibly yumex) first using yum
*should* result in a working yumex.
Paul.
More information about the fedora-list
mailing list