Yum Update before Plesk 9 installation

[TerryM]

Before installing Plesk 9 I decided to update the system using Yum Update, however it had some errors/warnings, please can somebody tell me if these are good or bad and if bad how to fix them before I attempt to reboot my server.

Installing: hal [ 58/112]warning: /etc/rc.d/init.d/haldaemon saved as /etc/rc.d/init.d/haldaemon.rpmorig
Installing: kernel ##################### [ 69/112]
grubby: unable to open /boot/boot.b: No such file or directory
grubby fatal error: unable to find a suitable template
grubby fatal error: unable to find a suitable template
Installing: notification-daemon ##################### [ 43/112]
gconftool-2: symbol lookup error: /usr/lib64/libgconf-2.so.4: undefined symbol: g_slice_alloc0
Installing: gnome-vfs2 ##################### [ 76/112]
Resolved address "xml:merged:/etc/gconf/gconf.xml.defaults" to a writable configuration source at position 0
gconftool-2: symbol lookup error: /usr/lib64/libgconf-2.so.4: undefined symbol: g_slice_alloc0
Installing: libgnome ##################### [ 77/112]
gconftool-2: symbol lookup error: /usr/lib64/libgconf-2.so.4: undefined symbol: g_slice_alloc0
error: %post(libgnome-2.16.0-4.fc6.x86_64) scriptlet failed, exit status 127
gconftool-2: symbol lookup error: /usr/lib64/libgconf-2.so.4: undefined symbol: g_slice_alloc0
error: %post(gnome-mount-0.5-2.fc6.x86_64) scriptlet failed, exit status 127

Any ideas as how to fix these would be much appreciated.

Thanks
Terry

 

[atomicturtle]

thats what I would expect to see if you had a system where the default kernel was removed and replaced by a source install. You at 1&1 by any chance?

 

[TerryM]

thats what I would expect to see if you had a system where the default kernel was removed and replaced by a source install. You at 1&1 by any chance?

I have a dedicated root server from 1&1, any ideas how to get around this please?

 

[atomicturtle]

Its a lot of work, basically theres a system used by the centos/rhel/fedora installers that figure out what the hardware is on the system. This is used by a modular kernel, and stored in modprobe.conf, or /etc/modprobe.d/. Since they used a monolithic kernel, that data was never created, and so.. you'd have to do it manually. Theres no step by step or howto for doing that, since everyones hardware is different. basically you need to figure that out, add it to the modprobe configs, and then install the RPM based kernel.

 

[TerryM]

Its a lot of work, basically theres a system used by the centos/rhel/fedora installers that figure out what the hardware is on the system. This is used by a modular kernel, and stored in modprobe.conf, or /etc/modprobe.d/. Since they used a monolithic kernel, that data was never created, and so.. you'd have to do it manually. Theres no step by step or howto for doing that, since everyones hardware is different. basically you need to figure that out, add it to the modprobe configs, and then install the RPM based kernel.

Ah, couldn't recommend a great company for a dedicated root server could you lol for future use so we dont get these issues, however we have found 1and1 to be very reliable so may have to weigh up security over reliability

Thanks for your help
Terry

 

[atomicturtle]

I actually got fed up with that kind of "help" from hosting companies too. So I wrote this:

wget -q -O - http://www.atomicorp.com/installers/aooi |sh

That re-images a box with a normal CentOS image. I dont want to single out 1and1 (I really like their rescue system in fact) since other outfits are equally guilty of this kind of meddling (OVH for example).

 

[steogede]

Ah, couldn't recommend a great company for a dedicated root server could you lol for future use so we dont get these issues, however we have found 1and1 to be very reliable so may have to weigh up security over reliability

Hi Terry, just thought I would share my experience of 1and1 on this matter. I got really fed up with the default configuration my 1and1 Server - especially the default partitioning scheme, and lack of easy ways to change it (IMHO the re-image should allow you to specify the partition scheme). When the time came to upgrade my server, I ended up doing a manual reinstall of openSUSE over the serial connection. I used SUSE because I have long been a fan of the YAST as a text based (ncurses) configuration tool and I knew where to find instructions for installing over a serial connection. I am sure you could do the same thing with CentOS or whichever distro you prefer.

It certainly isn't an option for the faint of heart, but it is worth considering. Ofcourse it does mean that the 1&1 support team are less likely to be able to fix things should anything go wrong; however, the few times I have dealt with them, I haven't found them very helpful (a re-image is often the best they can offer). Personally I chose a 1&1 dedicated server because I didn't care about support and wanted full control so that I could support it myself, rather rely on (often) useless support techs.