Re: [Xen-users] Support for file-based COW, present and future

[Grant McWilliams <grantmasterflash@xxxxxxxxx>]

On Mon, Jul 12, 2010 at 12:48 AM, Fajar A. Nugraha <fajar@xxxxxxxxx> wrote:

On Mon, Jul 12, 2010 at 2:11 PM, Grant McWilliams
<grantmasterflash@xxxxxxxxx> wrote:
> On Sun, Jul 11, 2010 at 8:28 PM, Garrett Robinson
> <Garrett.Robinson@xxxxxxxxxxx> wrote:
>>
>> I am currently working on a project that would be well-served by a
>> file-based COW filesystem such as is theoretically offered by QCOW,

>> What I'm really wondering is this: Has anybody has
>> success with implementing COW filesystems for their Xen guests using any
>> system, but in particular a file-based one? And does anybody know the
>> present status or future of COW/QCOW implementation in Xen?

> I fought with qcow and qcow2 with Xen for way too long. It seemed that no
> matter what version of Xen and the Linux kernel I was using there was always
> a bug that kept qcow from working. I eventually gave up. Rumor has it that
> Xen4 has working cow if you can get blktap2 to function.

If you want to use copy-on-write file backingstore, I'd actually
suggest using Xen 4. And even then it might be better to use vhd
instead of qcow.

http://wiki.xensource.com/xenwiki/blktap2
http://lxr.xensource.com/lxr/source/tools/blktap2/README

Personally, I gave up trying to use qcow a long time ago, settling
instead for storage-based COW functionality using zfs-fuse.

--
Fajar

_

Have you ran into any pitfalls with zfs-fuse? Are you using it in production? I'm waiting for everyone else to beta test Xen4. It's a little frustrating that qcow was never really finished because it isn't like the code is proprietary or anything, it's right there for the taking...


Grant McWilliams

Some people, when confronted 
with a problem, think "I know, I'll use Windows." 
Now they have two 
problems.
_______________________________________________
Xen-users mailing list
Xen-users@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-users