I'm running FC5 on an i686 machine, and have yum nightly updates enabled.
I had to rebuild anaconda recently, which required (amongst other things)
libsepol-devel. I tried to install libsepol-devel, but it wouldn't
libsepol-1.12.6-1 (latest). I had to rpm --nodeps -e libsepol-1.12.6
reboot into rescue mode, wget libsepol-1.12.4, and rpm -i --root
install it back.
Is this a known issue?
What can prevent a user to log in a GNOME session? When entering my
username and password at the GDM log in screen, I immediately get a
dialog box telling me that the X session lasts less than 10 sec and that
some help can be provided by the ~/.xsession_error. Well, this file
doesn't even exist. I've tried the "Last session" and "Default GNOME
session" options in the GDM log in screen without a success. Same
problem for root. This is on an Itanium workstation (ia64) with a fresh
install of Fedora Core Rawhide on May 2nd, updated with the latest
packages. I don't know if this problem also appears on other
architectures. Roughly a month ago (before my filesystem problem), I
didn't have this problem. But a lot of things have changed since...
First of all I'd like to mention that I only need the cli
version of php, just so I can run some simple management
scripts and cron jobs on servers with no web interface.
Some are simple scripts called via snmpd.
Unfortunately we have no php-cli package, so we need to
use the php package. This results in the following
output from yum:
php i386 5.1.2-5 core 3.5 M
Installing for dependencies:
apr i386 1.2.2-7.2 core 121 k
apr-util i386 1.2.2-4.2 core 74 k
aspell i386 12:0.60.3-5 core 959 k
aspell-en i386 50:6.0-2 core 1.6 M
curl i386 7.15.1-3 updates 265 k
httpd i386 2.2.0-5.1.2 core 1.1 M
libidn i386 0.6.2-1.1 core 191 k
php-pear noarch 1:1.4.6-2.1 updates 351 k
postgresql-libs i386 8.1.3-1 core 193 k
This is way overkill for my situation, but what really
makes me wonder is why php depends on postgresql-libs.
I thought I had to install php-pgsql for that...?
And please consider splitting php-cli out in a separate
package for FC6, it'd make a minimal server install much
neater. Even adding a php-cli that would conflict
with the normal php package could work.
Also, what if I wanted to use lighttpd instead of httpd?
I have posted this to the general list, and got a difficult to believe
response, so I am reposting here in the hope of a more insider
response. (not to at all belittle the only person who has responded to
me os far)
I am trying to figure out a problem that I am having with my madwifi
supported wifi card not loading properly. As a result, i noticed that
all my boot.log files are empty. What could be causing this?
To be updated...
Should packages with source from outside of the xorg-x11 tree carry
this prefix (e.g. ivtv, nvidia, ati, etc)? E.g. is this a prefix like
often used "for <prefix>" or is it a cendor prefix, e.g. "by
How would a 3rd party driver package be best named?
xorg-x11-drv-<driver> or <3rd-party-vendor>-drv-<driver>?
Axel.Thimm at ATrpms.net
Here's a related issue with the PATHs (which is why I change
the /etc/profile on my systems): a simple 'su' won't munge the superuser
directories into the PATH, because /etc/profile isn't invoked. If the
superuser directories are in the default PATH, a plain 'su' becomes a
lot more useful (and yes, 'su -' works, but then you lose the current
directory). An alternative would be to put path munging code
in /etc/bashrc (or, somehow, the PAM config for 'su').
Horst von Brand wrote:
> But ifconfig(8) is not for luser consumption, and so are lots of
I have troubles comprehending that statement. What tool would you
recommend if a unprivileged user wanted to know the IP address of the
system? 'cat /sys/class/net/eth0/address' ?
Or is it useful for an unprivileged user to know the WCHAN of running
processes (/bin/ps), the current keymappings (/bin/dumpkeys), and the
date and time that the kernel was compiled (/bin/uname -a), but not the
Realistically (these days), there's a good chance that an inexperienced
user won't be on the command line at all (but that's a far cry from
saying that 'bash is not for luser consumption').
Leszek Matok wrote:
> The wrong thing IMHO would be having /sbin before /bin in the search
> path. As someone stated earlier, it can break consolehelper symlinks.
> Other than that, either let's make a symlinking-fest or add them to
> the path as many suggest, only after /bin and co.
Agreed - the code in /etc/profile should probably read:
# Path manipulation
if [ "$EUID" = 0 ]; then
pathmunge /sbin after
pathmunge /usr/sbin after
pathmunge /usr/local/sbin after
This is better than a symlink-fest because the sites that really don't
want the users to know their IP address can change it easily :-) The
downside is that the path order is potentially less secure after an
I've noticed the following error during the upgrade from hal-0.5.7-5 to
Cleanup : hal #####################
/var/tmp/rpm-tmp.50206: line 4: z#fi: command not found
error: %postun(hal-0.5.7-5.ia64) scriptlet failed, exit status 127
Cleanup : hal #####################
Unfortunately, /var/tmp/rpm-tmp.5020 already disappeared.
[emeric@zx6000 ~]$ rpm -q hal
Is there something I can try to solve the problem? Is this issue generic
or specific to the ia64 (Itanium) architecture?