[CentOS] bugs with large partitions

Mon Apr 25 16:43:04 UTC 2005
Johnny Hughes <mailing-lists at hughesjr.com>

On Mon, April 25, 2005 10:22 am, Joshua Baker-LePain said:
> On Mon, 25 Apr 2005 at 4:12pm, John Moylan wrote
>> I installed centos4 on a raid system with 7 400GB Hitachi SATA disks in
>> RAID 5 on the 3Ware card. I set up a 2.2TB EXT3 Partition /opt.
> *snip*
>> Using /dev/sda
>> (parted) print
>> Disk geometry for /dev/sda: 0.000-2288754.000 megabytes
>> Disk label type: msdos
>> Minor    Start       End     Type      Filesystem  Flags
>> 1          0.031    101.975  primary   ext3        boot
>> 2        101.975   5098.754  primary   linux-swap
>> 3       5098.755   7091.191  primary   ext3
>> 4       7091.191 191600.624  extended
>> 5       7091.222   9083.627  logical   ext3
>> 6       9083.659  11076.064  logical   ext3
>> (parted) mkpart logical ext3 11076.065 2288754
>> Warning: You requested to create a partition at 11076.065-2288754.000Mb.
>> The closest Parted can manage is 11076.095-191595.520Mb.
>> OK/Cancel?
>> Any one have experience/ideas?
> Yes.  Oh yes.  And I was so annoyed by the experience that I submitted a
> documentation erratum to RH about it (see
> <https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=155728>).  I'll
> detail it a bit here for the archives.
> The short version is You Can't Do That.  Disk devices >2TB require gpt
> disk labels.  msdos disk labels (as you have above) can't handle devices
> of that size.  And neither grub nor lilo know how to boot from a gpt
> labelled device.  So, ATM, you cannot boot from a device >2TB.
> Your only option at this point is to add in a new boot device and install
> to that.  Once you've done that, go ahead and use parted on your big
> array, being sure to 'mklabel gpt' before you make any partitions on it.

Thanks for submitting the bug to RedHat concerning this issue.  This is an
upstream issue, as is anything like this.  CentOS-4 is a faithful rebuild
of the RHEL SRPMS and any limitations in RHEL4 will be carried forward.

If / when these issues are released in errata from RHEL, they will be
corrected in CentOS-4.

Johnny Hughes