Re: [Qemu-devel] [PULL 29/57] migrate_start_postcopy: Command to trigger transition to postcopy

[Dr. David Alan Gilbert]

* Eric Blake (address@hidden) wrote:
> [adding Markus for a qapi question]
> 
> On 11/09/2015 10:28 AM, Juan Quintela wrote:
> > From: "Dr. David Alan Gilbert" <address@hidden>
> > 
> > Once postcopy is enabled (with migrate_set_capability), the migration
> > will still start on precopy mode.  To cause a transition into postcopy
> > the:
> > 
> >   migrate_start_postcopy
> > 
> > command must be issued.  Postcopy will start sometime after this
> > (when it's next checked in the migration loop).
> > 
> > Issuing the command before migration has started will error,
> > and issuing after it has finished is ignored.
> > 
> > Signed-off-by: Dr. David Alan Gilbert <address@hidden>
> > Reviewed-by: Eric Blake <address@hidden>
> > Reviewed-by: Juan Quintela <address@hidden>
> > Reviewed-by: Amit Shah <address@hidden>
> > Signed-off-by: Juan Quintela <address@hidden>
> > ---
> 
> I know I reviewed an earlier version of this patch, but that was
> probably before 24/57 of this pull request spelled the migration
> capability bit as x-postcopy-ram.

It's been x-postcopy-ram since my first post, before we had
migrate_set_capability 
( https://lists.nongnu.org/archive/html/qemu-devel/2014-07/msg00869.html )

> > +++ b/qapi-schema.json
> > @@ -702,6 +702,14 @@
> >              '*tls-port': 'int', '*cert-subject': 'str' } }
> > 
> >  ##
> > +# @migrate-start-postcopy
> > +#
> > +# Switch migration to postcopy mode
> 
> No documentation on the relation to the [x-]postcopy-ram capability bit?

docs/migration.txt does have an explanation, but I'm happy to expand
this if you think it would be helpful.

>  Will this command always fail if that bit is not set?

Yes:
    if (!migrate_postcopy_ram()) {
        error_setg(errp, "Enable postcopy with migration_set_capability before"
                         " the start of migration");
        return;
    }


One alternative piece of text would be 
'Switch current migration to postcopy mode; the x-postcopy-ram capability must
be set before issuing this command.'

> > +#
> > +# Since: 2.5
> > +{ 'command': 'migrate-start-postcopy' }
> 
> Should we rename this command to 'x-migrate-start-postcopy' until we are
> ready to rename the entire feature to the stable namespace?

If you think it's best we could; however I took the 'x-' on the capability
just to be a flag to indicate it wasn't yet marked as stable;  I don't
think we're actually worrying about changes to naming.

> If so, I'm okay with that as a followup patch (so as not to delay the
> pull request), but we should really make up our minds what 2.5 will
> provide on this front.


Thanks,

Dave

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


--
Dr. David Alan Gilbert / address@hidden / Manchester, UK