[Lustre-discuss] 8TB Lun Limit

[Andreas Dilger]

On Nov 26, 2007  14:12 -0500, Aaron Knister wrote:
> How difficult would it be to migrate from the ext3 based ldiskfs to zfs?

It is planned that there will be an online object migrator.  This will
be useful for ext3->ZFS OST migration, and also space rebalancing or
performance tuning of objects locations if there are different types of
OSTs.

> On Nov 26, 2007, at 1:18 PM, Andreas Dilger wrote:
>> On Nov 26, 2007  11:08 -0500, Aaron Knister wrote:
>>> Is the 8TB lun size limit going to be lifted any time soon? Because of
>>> this limit I'm only getting 33TB out of 45TB raw using raid5 with
>>> terabyte drives. What is everybody else doing for large arrays over 8TB?
>>
>> There will be no 8TB limit for new ZFS-based filesystems, available when
>> Lustre 1.8 ships (planned for July '08).
>>
>> For ext3 filesystems, the 8TB limit will be removed once the upstream
>> kernel also removes this limit.  For RHEL5 I believe they are working
>> on up to 16TB per ext3 filesystem, but it isn't done yet, nor have we
>> tested it.  For ext4 based file system code (available from about Linux
>> 2.6.22 and on) the 8TB limit was also removed, but we need to port our
>> ldiskfs patches over to be ext4 based.
>>
>> Cheers, Andreas
>> --
>> Andreas Dilger
>> Sr. Staff Engineer, Lustre Group
>> Sun Microsystems of Canada, Inc.
>>
>
> Aaron Knister
> Associate Systems Administrator/Web Designer
> Center for Research on Environment and Water
>
> (301) 595-7001
> aaron at iges.org
>
>

Cheers, Andreas
--
Andreas Dilger
Sr. Staff Engineer, Lustre Group
Sun Microsystems of Canada, Inc.