| Summary: | No clean upgrade from xorg-x11-drv-nvidia-304.64-3.fc17.i686 | ||
|---|---|---|---|
| Product: | Fedora | Reporter: | Lionel Le Folgoc <lionel> |
| Component: | xorg-x11-drv-nvidia-304xx | Assignee: | Nicolas Chauvet <kwizart> |
| Status: | RESOLVED FIXED | ||
| Severity: | normal | ||
| Priority: | P5 | ||
| Version: | 17 | ||
| Hardware: | All | ||
| OS: | GNU/Linux | ||
| namespace: | |||
| Attachments: |
Yum log (unsatisfied dependency)
Dep resolver ok with localinstall |
||
Created attachment 1130 [details]
Dep resolver ok with localinstall
I think this is an isolated (not commonly experienced) problem. But you are using rhnplugin to manage your channel, maybe something goes wrong because of this. Did the previous updates was made using channel and rhnplugin ? Thanks for the fast reply. No, I haven't changed the channel configuration between the previous upgrade tests and now. I'll check if I can find something obviously wrong with yum-rhn-plugin. I'm suspecting an empty epoch requirement, so the libs has a requires >= :304.88 instead of 1:304.88 or 304.88. That's probably a bug in spacewalk (whatever internal component is responsible for generating metadata), because I compared the primary.xml I got directly from rpmfusion to the one from spacewalk, and the Provides for these package are changed from epoch="0" to epoch=""... Can you retest with the updated xorg-x11-drv-nvidia-304xx from rpmfusion-nonfree-updates-testing ? The new packages upgrade fine here, I tested both fc17 and fc18. Thanks! (fyi, I filed a bug report against spacewalk last friday, so hopefully they'll fix the issue on their side...) |
Created attachment 1129 [details] Yum log (unsatisfied dependency) Hi, I've some systems still using fc17, and yum is unable to find a clean upgrade path from the current nvidia driver to the latest one in -updates. $ rpm -qa '*nvidia*' nvidia-xconfig-1.0-20.fc17.i686 kmod-nvidia-3.8.4-102.fc17.i686.PAE-304.64-7.fc17.9.i686 nvidia-settings-1.0-22.fc17.i686 xorg-x11-drv-nvidia-304.64-3.fc17.i686 kmod-nvidia-PAE-304.64-7.fc17.9.i686 xorg-x11-drv-nvidia-libs-304.64-3.fc17.i686 The error message is: Error: Package: kmod-nvidia-304xx-3.9.8-100.fc17.i686.PAE-304.88-1.fc17.5.i686 (fedora17-i386-rpmfusion-nonfree) Requires: nvidia-304xx-kmod-common >= 304.88 Installing: xorg-x11-drv-nvidia-304xx-304.88-8.fc17.i686 (fedora17-i386-rpmfusion-nonfree) nvidia-304xx-kmod-common = :304.88 A direct install fails with the same message, # yum install xorg-x11-drv-nvidia-304xx <failure> but yum manages to resolve the proper dependencies with a localinstall from the rpm file: # yum localinstall xorg-x11-drv-nvidia-304xx-304.88-8.fc17.i686.rpm <success> Uninstalling and then reinstalling all nvidia packages seem to work too. (See the attachements for more verbose logs, I can provide more if needed) I don't know yum enough to figure out if it is confused by the colon without epoch, but rpmdev-vercmp surely isn't: $ rpmdev-vercmp '304.88' ':304.88' 304.88 == 304.88 Note that I upgraded a few systems to 304.88 two weeks ago (before the 304xx split), and it went fine. Thanks!