[Lustre-discuss] Lustre + qemu/kvm issue
andreas.dilger at intel.com
Tue Oct 16 02:42:49 PDT 2012
On 2012-10-15, at 2:15, Danny Sternkopf <danny.sternkopf at csc.fi> wrote:
> I wonder if there is anyone out there who has a working test or even
> production environment using Lustre for VM disk images?
> I run into the Lustre Direct I/O problematic which requires the
> application to read and write in 4K I/O chunks. qemu/kvm uses 512 byte
> as I/O size and refuses to start the VM.
Later this year, Advance Format hard drives 4TB and larger will no longer emulate 512-byte sectors as the 2TB and 3TB drives have been doing.
It probably makes sense that qemu/KVM get used to this, but I would guess that it already is...
Is it possible that the VM image that you are using has a 512-byte sector offset due to the partition table? Possibly that could be fixed by repartitioning the image and shifting the data blocks to be aligned on 4KB boundaries?
> In this case cache is "none".
> If I change the cache mode to writethrough (default) then the VM stored
> on Lustre can be started, but it has very poor write performance.
> (100-times worse) Any hints?
> Best regards,
> Lustre-discuss mailing list
> Lustre-discuss at lists.lustre.org
More information about the lustre-discuss