Re: [Qemu-devel] [PATCH v3 resend 7/8] rdma: introduce MIG_STATE_NONE and change MIG_STATE_SETUP state transition

[Eric Blake <eblake@redhat.com>]

[-- Attachment #1: Type: text/plain, Size: 1360 bytes --]

On 10/08/2013 11:32 AM, Michael R. Hines wrote:
>>
>> In this case, in particular, I think a capability is a bit overkill.
>> Libvirt needs to be somewhat liberal in what it accepts; in this case it
>> can treat any unknown state as "active".
>>
>> Paolo
> 
> That makes sense to me too - the setup state is "effectively" the same
> as active and can be safely treated as such.
> 
> There's a similar issue MigrationInfo statistics - what's the
> backwards/forwards-compatible procedure for not breaking libvirt
> when new statics appear? - such as "setup-time", which timestamps
> the new state that was introduced?

Libvirt ignores fields it does not expect.  But for fields it expects,
it was not ignoring new enum values within those fields that it was not
expecting.  The 'setup-time' change is not a problem (new field); only
the new state ('setup') - and even then, Paolo is right that newer
libvirt will be taught to be tolerant of unknown states, and that it is
okay to require newer libvirt when using newer qemu (for example, when
qemu 1.0 first came out, you had to upgrade to a new-enough libvirt that
knew what to do with the different version number).

So sounds like nothing further to worry about in qemu.

-- 
Eric Blake   eblake redhat com    +1-919-301-3266
Libvirt virtualization library http://libvirt.org


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 621 bytes --]