[Libguestfs] [PATCH v7 6/6] v2v: Add -o rhv-upload output mode (RHBZ#1557273).

[Richard W.M. Jones]

On Sun, Mar 25, 2018 at 08:05:14PM +0000, Nir Soffer wrote:
> > > > +            sparse = params['output_sparse'],
> > > >
> > > The user cannot configure that. This must be based on the image
> > > format. The current coded may create images in unsupported
> > > combinations, e.g. raw/sparse on block storage, or fail validation
> > > in engine.
> >
> > In virt-v2v this can be configured using ‘-oa’.  What are the possible
> > combinations?
> >
> 
> I could not find documentation for this in the ovirt engine sdk.
> There is http://ovirt.github.io/ovirt-engine-sdk/master/  but it is
> not very useful for anything.
> 
> The ovirt-engine REST API has real documentation here:
> http://ovirt.github.io/ovirt-engine-api-model/master
> 
> But I could not find any documentation about creating disks there, so
> the above is based on current system behavior.
> 
> When creating disks from the UI, the user select the storage domain
> (implicitly selecting the file/block), and the allocation policy
> (thin/preallocated). The user cannot select the image format when
> creating disk, ovirt will choose the image format and sparse for you.
> 
> storage type  allocation   | image format  sparse  creating
> ---------------------------|-----------------------------------------------------------------
> file          thin         | raw           true    sparse file of
> provisioned_size bytes
> file          preallocated | raw           false   preallocated file of
> provisioned_size bytes
> block         thin         | qcow2         true    logical volume of
> initial_size bytes
> block         preallocated | raw           false   logical volume of
> provisioned_size bytes
> 
> When uploading disk, the user select a file, implicitly selecting the
> image format (raw/qcow2), and the storage domain, implicitly selecting
> the storage type (file/block).  oVirt selects the allocation and sparse
> value for you.
> 
> storage type  image format | allocation    sparse   creating
> ---------------------------|---------------------------------------------------------------
> file          qcow2        | thin          true     sparse file of image
> size bytes
> file          raw          | thin          true     sparse file of
> provisioned_size bytes
> block         qcow2        | thin          true     logical volume of
> initial_size bytes
> block         raw          | preallocated  false    logical volume of
> provisioned_size bytes
> 
> Notes:
> - file,qcow2 cannot be created from the UI.
> - no way to create preallocated uploading from the UI (looks like a bug to
> me)
> 
> When using the sdk, you can select both the image format and sparse, so you
> can create invalid combinations. oVirt selects the allocation for you.
> 
> storage type  image format  sparse | allocation      creating
> -----------------------------------|--------------------------------------------------------
> file          qcow2         true   | thin           sparse file of image
> size bytes
> file          raw           false  | preallocated   preallocated file of
> provisioned_size bytes
> file          raw           true   | thin           sparse file of
> provisioned_size
> bytes
> file          qcow2         false  | -              unsupported
> block         qcow2         true   | thin           logical volume of
> initial_size bytes
> block         raw           false  | preallocated   logical volume of
> provisioned_size bytes
> block         qcow2         false  | -              unsupported
> block         raw           true   | -              unsupported
> 
> The only case when selecting the sparse value is useful is raw file
> on file based storage domain, allowing creation of sparse or preallocated
> disk.

I don't think this is true.  Both LVs (lvmthin) and SAN LUNs can
support sparse allocation meaningfully.

In any case it sounds like we need to know if the storage domain
corresponds to file or block storage, or else just force non-sparse
everywhere.  I'm going to force non-sparse for now, and add an 'XXX'
comment about it.

> I guess you can check if a storage domain is file based or block basedu
> using the SDK, but I'm not familiar with it.

OK.

> Daniel, do we expose this info?
> 
> > [snipped]
> > >
> > > > +# Can we issue zero, trim or flush requests?
> > > >
> > > +def get_options(h):
> > > > +    if h['got_options']:
> > > > +        return
> > > > +    h['got_options'] = True
> > > > +
> > > > +    http = h['http']
> > > > +    transfer=h['transfer']
> > > > +
> > > > +    http.putrequest("OPTIONS", h['path'])
> > > > +    http.putheader("Authorization", transfer.signed_ticket)
> > > > +    http.endheaders()
> > > > +
> > > > +    r = http.getresponse()
> > > > +    if r.status == 200:
> > > > +        j = json.loads(r.read())
> 
> > > +        h['can_zero'] = "zero" in j['features']
> > > > +        h['can_trim'] = "trim" in j['features']
> > > > +        h['can_flush'] = "flush" in j['features']
> > > > +
> > > > +        # If can_zero is true then it means we're using the new
> > imageio
> > > > +        # which never needs the Authorization header.
> > > > +        if h['can_zero']:
> > > > +            h['needs_auth'] = False
> >
> 
> In older proxy, we do support OPTIONS, used for allowing access
> from enigne UI, but it does not return any content. Fortunately, we
> return httplib.NO_CONTENT so it is easy to detect.
> 
> So we need to handle gracefully r.status = 204

OK done.

> > > >
> > >
> > > If we got here, we are working with new daemon or proxy, and both
> > > of them do not need auth, so we can set 'needs_auth' to False
> > > if OPTIONS returned 200 OK.
> >
> > Which is what this code does, unless I'm misunderstanding things.
> >
> 
> Accessing readonly ticket will return empty features list, and
> h['needs_auth']
> will be True, while this server does not need auth. Yes, it cannot happen
> with this code but it is not the right check.

Right, I reread what you said originally and it makes sense after a
good nights sleep, thanks!

Rich.

-- 
Richard Jones, Virtualization Group, Red Hat http://people.redhat.com/~rjones
Read my programming and virtualization blog: http://rwmj.wordpress.com
virt-df lists disk usage of guests without needing to install any
software inside the virtual machine.  Supports Linux and Windows.
http://people.redhat.com/~rjones/virt-df/