Re: [RFC] support more qdisk types

[Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>]

On Wed, Jan 27, 2016 at 07:42:49PM -0700, Jim Fehlig wrote:
> On 01/27/2016 02:09 PM, Konrad Rzeszutek Wilk wrote:
> > On Wed, Jan 27, 2016 at 02:25:51PM -0600, Doug Goldstein wrote:
> >> On 1/27/16 12:32 PM, Konrad Rzeszutek Wilk wrote:
> >>> On Mon, Jan 25, 2016 at 05:25:02PM -0700, Jim Fehlig wrote:
> >>>> I would like to hear the community's opinion on supporting more qdisk types in
> >>>> xl/libxl, e.g. nbd, rbd, iSCSI, etc. I prefer supporting additional qdisk types
> >>>> in libxl over apps like xl or libvirt doing all the setup, producing a block
> >>>> device, and then passing that to libxl. Each libxl app would have to
> >>>> re-implement functionality already provided by qdisk. libxl already supports
> >>>> IDE, AHCI, SCSI, and Xen PV qdisks. My suggestion is to extend that to initially
> >>>> include nbd, rbd, and iSCSI. Sheepdog, ssh, etc. could be added in the future.
> >>> ssh?
> >>>> I considered several approaches to supporting additional qdisk types, based
> >>>> primarily on changes to the disk cfg and interface. At one extreme is to change
> >>>> nothing and use the existing 'target=' to encode all required config for the
> >>>> additional qdisk types. libxl would need to be taught how to turn the blob into
> >>>> an appropriate qdisk. At the other extreme is extending xl-disk-configuration
> >>> Either way - new backends would require changes in both libxl and libvirt right?
> >>> The libxl would need to understand the new 'target=' blob to parse it out?
> >>>
> >> libvirt would probably just do what its doing now. Since it can setup
> >> the connection and pass the file descriptor into libxl. Honestly I don't
> >> see the advantage here because libvirt does a better job from a security
> >> standpoint and unless the goal is to have everything and the kitchen
> >> sink in libxl/xl. There's already a number of ways to skin the cat (xl,
> >> libvirt, xapi, openstack), why another one?
> > I believe what Jim is saying that there needs to be some parsing in libxl
> > so that it can pass the right information to QEMU.
> 
> Correct. The info is also needed when libxl creates the device in xenstore.

I think that would be awesome - especially with the iSCSI and Sheepdog.

The one thing that I am worried about is bitrotting. And I would think
if test-cases were added for this support - while it is bigger upfront
cost - would benefit us long term.

Granted I say that - but I hadn't done my share either (xSplice or some
other ones I have on mind) so feel free to ignore the recommendation.

> 
> Regards,
> Jim
>