CentOS - Feb 2006 - After install centos does not find hardrives (SATA)

Hi we installed centos on a Intel server board with LSI rad (no raid configured 
for the lsi).  in Centos it sees both drives so we software raid them and the 
installation goes well and finishes properly.  The problem is when the server 
reboots it can no longer see the hard drives and no just boots to a low level 
diagnostic shell.  Any ideas?



Thanks

- Nathan
- http://www.netdigix.com

On 15/02/06, Nathan <lists at netdigix.com> wrote:
> Hi we installed centos on a Intel server board with LSI rad (no raid
configured
> for the lsi).  in Centos it sees both drives so we software raid them and
the
> installation goes well and finishes properly.  The problem is when the
server
> reboots it can no longer see the hard drives and no just boots to a low
level
> diagnostic shell.  Any ideas?
Is the diagnostic shell single user mode? That would lead me to think
the kernel's bootstrapping to some extent at least and seeing the
disks otherwise where that shell coming from? Unless it's the grub
shell?

What does  ...

# fdisk -l

... list if it runs? Can you boot into rescue mode from an install CD,
does it find the installed OS and allow you to chroot into it? If not
does fdisk see the disks at this level?

If it's not getting further than greb then it's likely there's no
support for the disk controller in the initrd installed in boot but
there is in the one used off the CD.

Will.

actually its dosnt even make it to Grub,  its the EMI menu on Intel servers... 
so I guess it cannot even find a proper boot device.. ill have to do more 
research on the issue.  if you have any ideas why that would also be great help.

thanks alot,



Quoting Aleksandar Milivojevic <alex at milivojevic.org>:
> Quoting Nathan <lists at netdigix.com>:
> 
> > no its the prompt centos goes to when it cannot find anything to boot
to.
> 
> So it's the grub prompt most likely.  If it is, then your boot 
> partition is either too large and part of it falls outside of BIOS 
> addressable disk space (first 1024 cylinders), or is not positioned 
> inside BIOS addressable disk space (basically, same thing).  Grub can 
> access disk drives only using BIOS calls.  It needs to be able to 
> access files in /boot/grub directory, as well as kernel and initrd 
> images in /boot directory.  All those files *must* be in BIOS 
> addressable range, or booting will fail.  Fdisk will issue a warning 
> describing this for disks with more than 1024 cylinders.  
> Unfortunately, you won't get such warning during graphical install.  
> Grub isn't really verbose about this either (it just drops to CLI, 
> without telling user where the problem is).
> 
> You can easily check if what I just described is the problem you are 
> having.  Boot from CD into rescue mode.  Invoke fdisk and print out 
> partition table.  The start and end cylinder of partition that holds 
> /boot directory must be bellow 1024.  If they are not, you are in 
> trouble.
> 
> If this is the problem you are having, the easiest way is to reinstall 
> with above limitations in mind.  Common way of dealing with this BIOS 
> limitation is to create separate partition for /boot, and to make sure 
> entire partition used for /boot is inside first 1024 cylinders.  For 
> example, by making /boot be first partition on the disk, and having it 
> relatively small (100MB is more than enough for this partition).  This 
> is limitation of the BIOS, nothing to do with CentOS (any other 
> operating system).
> 
> 
> ----------------------------------------------------------------
> This message was sent using IMP, the Internet Messaging Program.
> 
> 
> _______________________________________________
> CentOS mailing list
> CentOS at centos.org
> http://lists.centos.org/mailman/listinfo/centos
> 



Thanks

- Nathan
- http://www.netdigix.com