On 2013-04-09 02:40, James Harper wrote:
I have a server that had 4 x 1.5TB disks installed in a RAID5 configuration (except /boot is a 'RAID1' across all 4 disks). One of the disks failed recently and so was replaced with a 3TB disk, and the goal is to replace the remaining 3 disks with 3TB disks to get more storage (holds Bacula backups so more storage = more online backup history).
Being a 3TB disk I had to make it a GUID Partition Table, and aside from a clean install I did once where Debian installer just did everything for me, I've never used gpt before. I believe I'm supposed to make a partition called the BIOS boot partition... can I store data on here or is it separate to my existing /boot partition? I created such a partition and did grub-install /dev/sda but then on boot from that disk all I get is "GRUB _" (where _ is the flashing cursor).
Any hints or links to good online resources would be appreciated. I've googled and can find a bunch of information about it but it's not working for me so I'm missing something somewhere...
Hi James, I've long ago given up on GRUB, and now use Extlinux. I recently built a Debian Wheezy system with a single 250GB disk, and then replaced that with 4 3TB disks which had /boot in RAID1 and the rest of the space consumed by a single RAID10 partition containing everything else inside LVM. I successfully migrated the system from the 250GB disk to the new RAIDed disks with minimal pain. I have vaguely documented my setup, and supplimented it with information from Trent Buck: http://www.cyber.com.au/~twb/snarf/extlinux.page http://www.cyber.com.au/~twb/snarf/extlinux-gpt.page The steps I followed, mostly, were: | # For all 4 disks | for f in /dev/sd[abcd] | do | # Give them a GPT label, 256MiB /boot, <rest> LVM space | parted -a optimal -s "$f" mkl g mkpart boot 0% 256Mi set 1 r on mkpart lvm 256Mi 100% set 2 r on | # Set the GPT "boot" bit on each disk | sgdisk "$f" -A 1:set:2 | # Install the syslinux/extlinux GPT MBR on each disk | cat /usr/lib/syslinux/gptmbr.bin > "$f" | done | # Build the 2 RAID arrays. Note the metadata versions. Extlinux | # *requires* 0.90 (metadata at end) so it can find itself, and Linux | # should have 1.2 (metadata at start) for the last array to prevent it | # from seeing metadata at the end and thinking the whole *disk* is one | #big RAID array and getting very confused. | mdadm -C /dev/md0 -e 0.90 -n 4 -l 1 -b internal /dev/sd[abcd]1 | mdadm -C /dev/md1 -e 1.2 -n 4 -l 10 -b internal /dev/sd[abcd]2 * Copy over any data, generating VGs/LVs and formatting/dd-ing as necessary. * Ensure extlinux config is correct (see extlinux.page above) * Chroot into system on new disks, and make any necessary mdadm.conf changes, then re-generate the ramdisk so those changes are propagated. And... I think that's it. I realise these instructions don't exactly fit your use case, but hopefully they give you some idea of what's required, and in any case I can't be much more help if you want to stick with GRUB. Good luck! -- Regards, Matthew Cengia