[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

Re: [K12OSN] RAID1 and cloning K12LTSP



On Wed, 29 Jul 2009 21:37:55 -0400
Jesse McDonnell <jessemcdonnell verizon net> wrote:

> My existing setup is K12LTSP5.0.0EL-64bit (upgraded to Centos 5.3) installed on a 160G SATA drive with separate / /home & /swap partitions, 4G ram, Athlon X2 4600+ CPU. It works perfectly and I rsync the home directories to another pc to avoid data loss. Now, I want to replace the 160G drive with two 300G SATA drives in a raid1 setup and I'm looking for advice/modifications on the upgrade route of doing a virgin raid1 install and immediately replacing it like so:
> 
> 1. Physically disconnect 160G drive and connect two new drives.
> 2. Using K12LTSP5.0 discs (from 2007), do fresh minimal install setting up raid partitions / /home and /swap -reboot to make sure everything works.
> 3. Reconnect 160G drive, boot in rescue mode (maybe use knoppix disc). Copy /etc/fstab for new raid1 install somewhere safe.
> 4. Delete contents of /home and / partitions on each raid1 drive and copy over contents of / (except for /etc/fstab) and /home from 160G drive. 
> 5. Restore /etc/fstab to 300G drives.
> 6. Disconnect 160G drive and reboot in rescue mode, chroot to /mnt/sysinstall and run grub-install (on both drives?)
> 
> In theory this looks like it should work, but we all know the devil is in the details! Are there any major gotchas etc?

Reporting the outcome and following up on my own post. First, thanks to Warren T. and Paul S. for the links - helped me get a better sense of what was involved and lead to putting /boot on a separate partition. 

The devil was indeed in the details and after running into problems I changed direction in favor of a totally fresh install. Following my upgrade/cloning plan, the system wouldn't boot at first because of LVM configuration information copied over from the old /etc directory. After that was sorted out, it booted but with a bunch of exportfs errors. Since the clients depend heavily on the exported filesystems a fresh install was the logical/least troublesome option and I did a software raid install with home on a separate partition as planned. After transferring the old /home partition contents to the new /home raid partitions and replacing the /etc/group,gshadow,passwd,shadow files everything is working fine.

Jesse  


[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]