Re: [PATCH v3 1/1] vhost-user-fs: add migration type property

[Michael S. Tsirkin]

On Wed, Mar 01, 2023 at 06:29:50PM +0200, Anton Kuchin wrote:
> On 01/03/2023 17:52, Michael S. Tsirkin wrote:
> > On Wed, Mar 01, 2023 at 05:40:09PM +0200, Anton Kuchin wrote:
> > > So catching errors in not the only purpose of this property, but it
> > > definitely
> > > allows us to catch some obvious ones.
> > OK let's say this. If migration=external then migration just works.
> > If migration=internal it fails for now. We are agreed here right?
> > 
> > Our argument is whether to check on load or save?
> > 
> > I propose this compromize: two properties:
> >     migration-load and migration-save
> > 
> >     migration-load : how is incoming migration handled.
> >                      internal - through qemu
> >                      external - through the daemon
> > 
> >                       checked in pre-load
> > 
> >     migration-save : how is outgoing migration handled.
> >                      internal - through qemu
> >                      external - through the daemon
> > 
> >                       checked in post-save
> > 
> > This way whether the check is on source or destination or both
> > is up to the user.
> > 
> > Hmm?
> 
> Hmm, sounds interesting, this can be a really good compromise.
> 
> So migration-save will be "none" by default to protect unaware
> orchestrators.
> What do you think should migration-load be "none" too to force orchestrator
> set proper incoming migration type?

Yes.
I am also thinking whether we should block setting migration-load
if qemu is starting a VM as opposed to loading it.
Thoughts?


-- 
MST