[Lustre-discuss] lustre support on suse 11 with PPC64

[Andreas Dilger]

On 2009-11-04, at 14:05, Brian J. Murrell wrote:
> On Wed, 2009-11-04 at 11:06 -0600, Troy Benjegerdes wrote:
>>
>> Not that this is a big deal realistically, since it would be  
>> difficult
>> to find PPC64 hardware that works well for a server, unless you roll
>> your own embedded system...
>
> Sure.  But it counts for all machines of the same endianess, not just
> ppc64.
>
>> But why? The linux kernel has no problems providing an endian-clean
>> and wordsize-clean environment to operate in.
>
> Sure, but Lustre is quite a bit more than just a driver in any one  
> Linux
> kernel.  There is an entire networking stack written from the ground  
> up,
> not to mention the VFS client and server modules.  And Lustre is a lot
> bigger (I mean in concept) than the Linux kernel.  The Linux kernel  
> has
> the luxury of only having to deal with a single endianess at a time.
> Lustre has to deal with potentially different endianesses all at once.
>
> To be sure, these are not at all insurmountable difficulties.  They  
> are
> just things that are not being demanded enough to warrant attention
> (yet).  As always, we accept patches.
>
>> Are Sparc64 servers supported?
>
> I don't know.  Are they the same endianess as ppc64?  If so, then no.


Just as a bit of background here - the reason that we don't support  
big-endian
servers is due to problems in the ext3-patched ldiskfs code.  It  
wasn't written
originally to be endian clean, and that work wasn't done until later  
in the ext4
project when it was landed to the kernel.  In theory, the SLES11  
Lustre server
(with ext4-based ldiskfs) would be able to run on a big-endian system,  
if there
are no other issues with endian support in Lustre.  Since we don't  
test that
configuration I wouldn't bet my data on it without significant testing.

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