[CentOS] Re: Please help! boot problem

Les Mikesell lesmikesell at gmail.com
Mon May 1 20:33:12 UTC 2006


On Mon, 2006-05-01 at 15:27, Mace Eliason wrote:
> Okay to clarify
> 
> I have only hooked up the drive that is not bootable.  The drive that is 
> bootable has been disconnected.
> 
> If I run sfdisk on the nonbootable drive I get
> /dev/sda1  (id fd linux raid autodetect) marked * for boot
> /dev/sda2   "   "
> /dev/sda3   "   "
> 
> So yes it is software raid am I right.  The system has been running off 
> this drive for almost a month, until someone rebooted the server.

> Should I have both drives installed when I run the linux rescue?

It should autodetect the md devices with one missing.  However
if it doesn't find the partitions in the automatic scan you
can try it the hard way:
>From the command line in rescue mode:
 mount /dev/sda3 /mnt/sysinstall
 mount /dev/sda1 /mnt/sysinstall/boot
Assuming those succeed:
 chroot /mnt/sysinstall
Look around a bit to make sure you have the right thing
mounted, then proceed with the instructions for installing
grub.  If that doesn't work either, try:
grub <enter>
 device (hd0) /dev/sda
 root (hd0,0)
 setup (hd0)
 quit

then 'exit' twice to reboot and remove the rescue cd.


> Does linux automatically rebuild the raid when it boot?  If I do get 
> them to boot with the good drive I don't want it to over write the 
> current drive.

I'm not sure about the details but I think there is a counter
on each disk with the number of times the RAID has been stopped
cleanly.  If it does auto-sync, the one with the higher counter
will be the master.  If you can get grub installed and reboot
a few times with the disk you want to stay current it should
become the master when they are both seen - although I generally
don't trust this and do a scsi low-level when I have a chance
when swapping in a new drive, then make the new matching partitions
and add to the raid after booting.

-- 
  Les Mikesell
   lesmikesell at gmail.com





More information about the CentOS mailing list