Re: Can't get a GUI window to open up.

This is an OpenPGP/MIME signed message (RFC 2440 and 3156)
--------------enigB5501EC5CD32377603F30BFA
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: quoted-printable

Jim Beard grabbed a keyboard and wrote:
> On 10/24/2011 07:55 PM, David Guntner wrote:
>>
>> There's the current drivers.  I've run XFdrake in the console window a=
nd
>> select "test," and a nice graphic display comes up with the little "Is=

>> this correct?" window with the yes/no button to click on and all that
>> comes up just fine.  However, actually issuing a "telinit 5" results i=
n
>> a few messages, but then nothing - no GUI.  Hitting F12 to bring up th=
e
>> syslog reveals the following message:
>>
>>> Oct 24 16:52:05 janet init: Switching to runlevel: 5
>>> Oct 24 16:52:06 janet udevd[2121]: kernel-provided name 'capi' and
>>> NAME=3D 'capi/' disagree, please use SYMLINK+=3D or change the kernel=
 to
>>> provide the proper name
>>> Oct 24 16:52:06 janet udevd[2121]: mknod '/dev/capi/.udev-tmp' 68:0
>>> 020600 failed: Operation not permitted
>>> Oct 24 16:52:06 janet kdm[6991]: X server died during startup
>>> Oct 24 16:52:06 janet kdm[6991]: X server for display :0 cannot be
>>> started, session disabled
>>
>> So, any ideas on what's causing the current headache, and what can be
>> done to clear it up?
>=20
> You might try deleting old cruft.  I would do this from within run leve=
l 3.

Yes, I'm currently in run level 3.  I'm trying to switch to run level 5
in order to start up X, which is the part that's failing. :-)

> cd
> rm -f .*auth*
> rm -rf tmp/* /tmp/* /var/tmp/*

I will keep that in mind, but at the moment I don't believe it's at the
point where those files matter for anything (I did check for any stray
=2Exauth files, etc., though).  XFdrake (which configures your xorg.conf
file) can display a test screen to make sure things are configured
properly.  I suspect it uses X to do this. :-)  But when it comes time
to actually full-on start it up (via "telinit 5"), nothing happens and
the log shows the above messages.

I can't help but think those udevd messages might be some kind of clue,
but I'm not sure how to fix what it's complaining about.

> Then check the contents of /etc/X11/xorg.conf and see if all makes
> sense.  If you have some old versions lying around, use
> diff xorg.conf xorg.conf.something
> to see how they differ.

No old versions at this point.  What's in there seems to make sense.
And as I said, a test run from XFdrake *will* bring up what's clearly a
graphic window to make sure things are configured properly.  This is
part of what makes it so maddening.

> I know that there have been recent problems with some of the nvidia
> installers, but I currently use only ATI cards and cannot help beyond
> general guidance from years past.

Fair enough. :-)

> For proprietary drivers:
>=20
> 1.  Do not install both the Mandriva nVidia installer package and the
> nVidia installer package from the nVidia website.  Just having both on
> the system can results in problems, even if you only use one.

Oh yea, learned that one a long time ago. :-)

Again, my currently-installed driver packages are:
[root@janet ~ ]# rpm -qa|egrep nvidia
dkms-nvidia-current-280.13-4-mdv2011.0.x86_64
nvidia-current-doc-html-280.13-4-mdv2011.0.x86_64
x11-driver-video-nvidia-current-280.13-4-mdv2011.0.x86_64
[root@janet ~ ]#

That very last one is used as part of the dkms-nvidia-current package.

> 2.  You must have installed the kernel-devel package for the exact
> kernel you are running.  In times long past, I had to create the symlin=
k
> from /usr/src/version_of_kernel to
> /usr/src/linux before the installer would find the devel files. That
> should not be needed  (the installer should find the devel header files=

> even if the linux link is missing) but "should not" sometimes fails.

There were no errors reported when the install of the dkms package was
building its headers, etc., for the kernel in order to lodge the NVIDIA
drivers there.  During boot, I see the usual message about the driver
being installed in the kernel.

> 3.  You must check /etc/X11/xorg.conf and see if all is well there.
>=20
> Alternate drivers are the nv and nouveau drivers.  You may find that of=

> the three possibilities, proprietary, nv, and nouveau, only one will
> work with the current system.
>=20
> If you change drivers, remember to always remove the old cruft (stuff i=
n
> the tmp directories, $HOME/.*auth*).

I'll look into those, thanks, but I really don't think it's a driver
issue at the moment.  It seems to me that a *driver* issue would cause
XFdrake to not be able to open up a test graphic display, and it's able
to do that.  It's not until I try to start up X (not *login* to a
session as a user or anything, just start the server) that I see those
error messages in the log.  When I run the XFdrake test screen, the log
is quiet a a church mouse. :-)

                --Dave


--------------enigB5501EC5CD32377603F30BFA
Content-Type: application/pgp-signature; name="signature.asc"
Content-Description: OpenPGP digital signature
Content-Disposition: attachment; filename="signature.asc"

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.10 (MingW32)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/

iEYEARECAAYFAk6m6acACgkQvC3peSYrMN6wHgCg5x0r6dEfVujUpFZmqC3I7Rgm
l8wAnA1LC/2kkvreJVcVB5LsLAeK65xf
=J1K2
-----END PGP SIGNATURE-----

--------------enigB5501EC5CD32377603F30BFA--