I went through this recently with SLC4 (RHEL4 clone) on a Dell PowerEdge 2900. I reconfigure it to make one virtual disk out of 4x750GB disks. It did let me use this to do the install but as Joshua says grub wouldn't do anything when you reboot after it. I used a rescue disk to create a USB stick to boot grub with. That worked but then the large partition was actually less than reported during the install and while booting fsck complains the sizes disagree (filesystem vrs partition). I then tried changing it to use GPT with parted but then (as mentioned by Joshua also) grub couldn't read the partition table on the big disk. I ended up making two virtual disks, one of 100GB for the system and one with the rest as a data disk.

On Thu, 4 Oct 2007, Joshua Baker-LePain wrote:

On Thu, 4 Oct 2007 at 11:29am, David Miller wrote

 The first is a bug in disk druid at install time. If you have a Volume
 group with a logical volume larger than 2TB you can't edit the default
 setup parameters in disk druid. For example I have a Dell poweredge 2950
 with 6x750GB in RAID5 using the parc 5/i. By default these servers come
 configured with two virtual disks. Using Red Hat's default layout it will
 use LVM and combine the two disks into one volume group. this makes the
 root logical volume about 3.5Tb. The size listed is not correct. It shows
 3.57229e+06. If you try to edit the logical volume name or group and click
 ok it will say that the amount of space allocated is too large and gives a
 range of 0 to 2TB. Even manually typing in the amount shown as
 available(3574272) gives the same error. Of course if you leave it alone
 it will format the volumes and work just fine.

 The second bug is that if I use the perc 5/i and create one volume of
 3.5TB, RHEL5 will kernel panic during boot. I'm guessing that RHEL5 cannot
 use physical drives over 2TB for the /boot or / partition? Again I used
 Red Hat's default partition layout.

At this point, *no* distribution I know of can boot off of a device >2TiB. That's because grub (and LILO, for that matter) does not support devices with gpt disklabels, and you must use such a label on a device that large.

As for Disk Druid, I don't know what it uses on the back end, but, e.g., fdisk can handle >2TiB block devices either -- you have to use parted.



--
 /------------------------------------+-------------------------\
|Stephen J. Gowdy, SLAC               | CERN     Office: 32-2-A22|
|http://www.slac.stanford.edu/~gowdy/ | CH-1211 Geneva 23        |
|                                     | Switzerland              |
|EMail: [EMAIL PROTECTED]       | Tel: +41 22 767 5840     |
 \------------------------------------+-------------------------/

_______________________________________________
rhelv5-list mailing list
[email protected]
https://www.redhat.com/mailman/listinfo/rhelv5-list

Reply via email to