New Kernel Update fails

[uluru]

is it good to enable the updates as well?

I also can't find the shell command launch the gui as root so I can enable update repos as well, can you please provide it?

Well, if you don't enable the update repos, you won't get any updates. Simple as that.
BTW, you also miss the tainted_updates repo, and you should disable the nonfree_backports.

For the command, for what do you need that? By the default all Mageia update repos are marked as update,
what do you want to change exactly? Normally you shouldn't fiddle with that.

I checked the tainted_updates and the nonfree_updates, disabling the nonfree_backports, updated the system, and now it boots fine.

after install I rebooted, the computer seemed hung on the fourth bubble for a couple minutes, then it rebooted by itself and the following boot was quite fast and smooth.

Thank you

Now all I miss is how to enable plugin (cpufreq, network and ram usage, etc.) in the top gnome-classic bar, but eventually I'll open a new thread for that.

[griffin]

the config and System.map symlinks are not a kernel update bug as such. The kernel and its install scripts does not touch those symlinks as that would break the systems for people switching kernels they boot.

The /boot/config and /boot/System.map symlinks must match the running kernel, not the latest one installed, and is managed by /etc/rc.sysinit at boot time.

so the real bug is to find out why it fails to update the symlinks for some users. A simple guess at this point is that we have a timing bug hiding somewhere...

The other question is why it did not show up during QA as I see people with atleast Intel, nVidia and Ati/AMD hardware testing the updates and they did not trigger it :/

We are investigating the issue to figure out what went wrong and will try to improve our testing so a problem like this wont slip through again...

Running cat /proc/cpuinfo on the system where this happened shows:

model name : AMD Athlon(tm) 64 X2 Dual Core Processor 5600+

The problem is persistent, that is, the boot process encounters a kernel panic and hangs before /etc/rc.d/rc.sysinit fixes the config and System.map links, so another reboot still doesn't work.

Since the kernel update and subsequent reboot work fine on other systems, which might have single-core (or quad-core) processors, maybe the problem affects only dual-core systems?

The easiest solution might be to single-thread running /etc/rc.d/rc.sysinit even though that might add a few seconds to boot time.

I can put the bad symlinks back and provide log listings, but I don't know how to trace the boot task dispatch process.

[wilcal]

A little bit of testing info here. I had a spare
(removable) drive with Mageia 2 on my 64-bit Nvidia
platform. So I booted that this morning, updated it
and all kind of bad things happened. So I kicked
that aside for now and popped in a blank drive
and booted with the 64-bit boot.iso file and did a
from the ground up install using my local repo.

Mageia 2, 64-bit installed using kernel
3.3.8-desktop-2.mga2. I then let the non-free
Nvidia driver install and that went well even
after a warm boot.

So for me a clean new install using an up-to-date
today repo and boot.iso ( 64-bit ) works just fine.

[DonaldFR]

Isadora, I tried your instructions (thank you!), and got this error on several tries.

[root@localhost donaldfr]# rpm --rebuilddb
error: rpmdb: Thread/process 4159/3074299584 failed: Thread died in Berkeley DB library
error: db4 error(-30974) from dbenv->failchk: DB_RUNRECOVERY: Fatal error, run database recovery
error: cannot open Packages index using db4 - (-30974)

BTW, thes is a dual core.

-- Donald

[doktor5000]

FWIW: http://www.oldrpm.org/hintskinks/repairdb/

[wilcal]

A little more testing here today

My primary OCZ VERTEX 4 SSD with M2 on a i7/Nvidia
platform I updated today without any problems. I was able
to warm reboot with no problems. It is now running the
latest kernel ( 3.3.8-2 ) and getting there with no problems.
I will continue to update here if I run into any problems.

Do note that the update process, even with the OCZ SSD,
took many minutes. Far more then I would have expected.
But eventually it completed and a warm boot was successful.
I am able to reboot back to the previous kernel with no problems.

[wilcal]

I've executed a from a blank drive up install
of Mageia 2 64-bit to my i7/Nvidia platform
using:

Mageia-2-x86_64-DVD.iso

Once that is installed from the DVD I enabled
all the repos then allowed the MCC to update
with the Nvidia proprietary driver. That
automatically installed kernel 3.3.8-2.
After the Nvidia driver install I did a
warm boot as requested and all went well.

After that a complete system update and
that went well as well.