Portal Home > Knowledgebase > Articles Database > can't upgrade kernel on a centos box, please help
can't upgrade kernel on a centos box, please help
|Posted by minadreapta, 03-19-2010, 05:16 PM|
|hi guys, i have a centos 5.4 machine with 2.6.18-164.11.1.el5PAE kernel on it.
i have run yum update, kernel was updated to 2.6.18-164.15.1.el5PAE
i double checked /etc/grub.conf so it shows default=0
rebooted the machine but when i run "uname -r" it still shows the old kernel: 2.6.18-164.11.1.el5PAE
here's the output of "cat /etc/grub.conf"
can someone please help me ? thanks.
|Posted by vapetrov, 03-19-2010, 07:02 PM|
|In centos by default /etc/grub.conf is symlink to /boot/grub/grub.conf
Is it so?
What is in your /boot/grub/grub.conf?
|Posted by minadreapta, 03-20-2010, 03:04 AM|
|yes, it is symlinked, it has the same content:
|Posted by andreyka, 03-20-2010, 05:05 AM|
|You have real server or virtual?
|Posted by assistanz247, 03-20-2010, 05:49 AM|
|What is the result of,
rpm -q kernel
it should show both the kernels if not the installation was not successful.
|Posted by minadreapta, 03-20-2010, 07:31 AM|
|it was showing nothing, then i ran yum install kernel
then rebooted and now it shows:
root@milo [~]# uname -r
root@milo [~]# rpm -q kernel
root@milo [~]# rpm -q kernel-PAE
root@milo [~]# rpm -q kernel-headers
and grub.conf looks like:
|Posted by assistanz247, 03-20-2010, 08:23 AM|
|It appears to be a kernel panic in the newly installed kernel, after the 5 seconds timeout it boots the old kernel again. The 5 seconds timeout has been set in your grub.conf,
Maybe you can try removing the following lines from grub.conf after backing up the grub.conf,
title CentOS (2.6.18-164.11.1.el5PAE)
kernel /vmlinuz-2.6.18-164.11.1.el5PAE ro root=/dev/md0
Please note this is at your own risk. If it fails to boot then you may need to contact the DC.
|Posted by Lightwave, 03-20-2010, 09:38 AM|
|You're doing this on a metadevice...
So you might want to redo the bootloader manually to make sure the right stuff got written to both? drives.
Afaik timeout has nothing to do with it's failback support. It just declares how much time to wait a the bootloader in case you want to make changes.
Last edited by Lightwave; 03-20-2010 at 09:43 AM.
|Posted by minadreapta, 03-20-2010, 11:17 AM|
|i don't know what's wrong, grub.conf now looks like this:
but after the reboot, "uname -r" shows an old kernel which is not even mentioned in the grub.conf:
any other ideeas?
|Posted by minadreapta, 03-20-2010, 11:23 AM|
|Lightwave: that didn't work either, after the reboot i have the same 2.6.18-164.11.1.el5PAE kernel
|Posted by minadreapta, 03-22-2010, 05:37 AM|
|anybody knows what should i do?
Last edited by minadreapta; 03-22-2010 at 05:43 AM.
|Posted by erectvps, 03-22-2010, 05:43 AM|
|Can you get console access to the server?
|Posted by VIPoint, 03-22-2010, 05:53 AM|
|What does the command dmesg show? Reboot the server, login as root and execute the command dmesg .
|Posted by minadreapta, 03-22-2010, 06:46 AM|
|erectvps: the server is online, it's working fine. i have full root acces to the server. the problem is that the new kernel is not on place.
I can get console acces if i want to, but why, i have SSH access already?
@VIPoint: i ran "dmesg |grep 'kernel'", i can't reboot the server now, there are about 50 live accounts on it in this moment.
the backupdriver thing is from the r1soft CDP agent i think
|Posted by minadreapta, 03-22-2010, 10:58 AM|
|any other ideeas?
|Posted by minadreapta, 03-23-2010, 03:24 AM|
|here is something from dmesg, i took out all the initialization texts...
|Posted by minadreapta, 03-23-2010, 05:38 PM|
|nobody can help me with this one?
|Posted by Lightwave, 03-25-2010, 02:23 AM|
|You should consider fixing your RAID 1 mirrors first.
Add to Favourites Print this Article