Re: [Qemu-devel] Proposal for extensions of block job commands in QEMU 1.2

[Eric Blake]

On 05/21/2012 05:02 AM, Paolo Bonzini wrote:

> Eric, is it a problem for libvirt if a pause or target error during
> mirroring causes the job to exit steady state?  That means that after a
> target error the offset can go back from 100% to <100%.

Libvirt would really like to have events present.  If you emit an event
on every change from <100% to 100%, and a counterpart event on every
revert from 100% back to <100%, then that will help tremendously.

For RHEL 6.3, libvirt requires 100% before attempting a drive-reopen.
But since this proposal says qemu 1.2 won't even have a drive-reopen,
but instead have a new block-job-complete, then libvirt should be okay
if block-job-complete gracefully fails any time things are less than
100% (remember, drive-reopen has an attempted effect even if things are
less than 100%, which is why libvirt had to pre-filter for that
situation, but then again, RHEL never downgraded).  In other words, I
think downgrading out of steady state is not a problem, as long as
events exist to help management track that, and as long as the command
to pivot to the destination gracefully requires steady state.

> 
> On the other hand, in other cases it can be desirable (qemu -S, run
> streaming before the VM starts).

Yes, libvirt would really like a way to be able to start a VM with
mirroring active, in order to make it possible to support drive copy on
persistent domains.

-- 
Eric Blake   address@hidden    +1-919-301-3266
Libvirt virtualization library http://libvirt.org

signature.asc
Description: OpenPGP digital signature