Re: [libvirt] [PATCH v3 2/5] vz: add migration backbone code

On 25.08.2015 20:28, Dmitry Guryanov wrote: > On 08/25/2015 07:18 PM, Daniel P. Berrange wrote: >> On Tue, Aug 25, 2015 at 12:04:14PM +0300, nshirokovskiy(a)virtuozzo.com wrote: >>> From: Nikolay Shirokovskiy <nshirokovskiy(a)virtuozzo.com&gt; >>> >>> This patch makes basic vz migration possible. For example by virsh: >>> virsh -c vz:///system migrate --direct $NAME $STUB vz+ssh://$DST/system >>> >>> $STUB could be anything as it is required virsh argument but it is not >>> used in direct migration. >>> >>> Vz migration is implemented as direct migration. The reason is that vz sdk do >>> all the job. Prepare phase function is used to pass session uuid from >>> destination to source so we don't introduce new rpc call. >> I'm trying to understand why you need $STUB as dummy data. >> >> IIRC, when doing direct migration, you should be providing a valid >> URI for the first parameter, and not need the second uri. > > virsh uses virDomainMigrateToURI3 function, and for some reason it differs > destination URI for direct and peer-to-peer migrations. For p2p it uses > pconnuri argument and for direct -VIR_MIGRATE_PARAM_URI <http://libvirt.org/html/libvirt-libvirt-domain.html#VIR_MIGRATE_PARAM_URI... from typed params list. > AFAIU in virDomainMigrateToURI3 function, dconnuri argument is treated as URI to destination libvirt daemon. As direct migration doesn't need this kind of connection this argument is ignored. However as suggested vz migration implementation uses connection to destination libvirtd daemon to get destination hypervisor session uuid I should probably use dconnuri for for that purporse not hypervisor URI. This is related what Dmitry mentioned in previous letters where there was a confusion between hypervisor and remote access to daemon ports. This would break a invariant mentioned in the documentation that direct migration does not use dconnuri. Probably it is not too bad too get away from this restriction.