[Libguestfs] Pseudo code for v2v

[Matthew Booth <mbooth redhat com>]

I've attached my initial thoughts on the design for the v2v tool.
--
Matthew Booth, RHCA, RHCSS
Red Hat Engineering, Virtualisation Team

M:       +44 (0)7977 267231
GPG ID:  D33C3490
GPG FPR: 3733 612D 2D05 5458 8A8A 1600 3441 EA19 D33C 3490

GuestOS e.g. RHEL/Debian
  enable driver
  disable driver
  add application
  remove application
  remap device names
  prepare bootable

HVTarget == KVM
  get command line options
  handle command line option
  is ready

  specialize (virt-inspector)

  configure (guestos, virt-inspector) # Passes through to specialized below

HVTarget::KVM == Linux
  can_configure (virt-inspector)

  configure-guest (guestos, virt-inspector)
  configure-metadata (libvirt DOM)

HVSource e.g. XenFV::Linux XenPV::Linux VMWare::Linux
  unconfigure-guest (guestos, virt-inspector)
  unconfigure-metadata (libvirt DOM)

MetadataReader e.g. libvirt/libvirtxml/ovf/command line
  get command line options
  handle command line option
  is ready
  handle arguments

  get libvirt DOM

NewGuest
  new (metadatareader, storage)

  init storage
  get storage locations
  init guest

  get libvirt dom
  create

Storage e.g. in place/snapshot/copy
  get command line options
  handle command line option
  is_ready

  get mapped location (location)
  create mapped location (location)

virt-v2v \
  --target kvm --no-virtio \  -- HVTarget::KVM
  --with rhel5.i386.kernel=/path/to/rhel5-kernel.rpm \ -- HVTarget::KVM
  --storage snapshot --snapshot-dir /tmp \ -- Storage::Snapshot
  --source libvirtxml --network default \ -- MetadataSource::LibvirtXML
  rhel53xen.xml rhel53kvm

# Think we'll need 2 passes on the command line options to implement this
{
    hvtarget = HVTarget::KVM
    storage = Storage::Snapshot
    mdr = MetadataReader::LibvirtXML

    hvtarget->handle_option(--no-virtio) # Just an example. Ignored below.
    hvtarget->handle_option(--with rhel5.i386.kernel=/path/to/rhel5-kernel.rpm)
    storage->handle_option(--snapshot-dir /tmp)
    mdr->handle_option(--network default)
}

mdr->handle_arguments(qw(rhel53xen.xml rhel53kvm))

Check: hvtarget/storage/mdr->is_ready()

newguest = NewGuest->new(mdr, storage)

# Create the underlying storage for the target guest
newguest->init_storage

%desc = virt-inspector (md->get_images())

# Infer guest OS type from %desc
# Actually, don't do this, make modules do their own discovery on %desc
guestos = GuestOS::RHEL

# Infer HVsource type from %desc
# Actually, don't do this, make modules do their own discovery on %desc
hvsource = HVSource::XenFV::Linux
hvsource->unconfigure-guest(guestos, %desc)
    # This looks for and removes xen specific options in the guest
    # Remove pv kernels if installed
    # For VMware it would also remove vmware-tools

hvsource->unconfigure-metadata(newguest->get_libvirt_dom)
    # Remove anything HV specific and replace with FV equivalent
    # e.g. /domain/devices/input/@bus = xen -> ps2
    #      Remove /domain/bootloader
    #
    # Don't touch:
    #      /domain/@type
    #      /domain/os
    # These must be rewritten by hvtarget

# hvtarget is now a HVTarget::KVM::Linux
hvtarget->configure-guest(guestos, %desc)
    # Calls self->specialize(%desc) to find a guestos specific implementation
    #
    # HVTarget::KVM::Linux->configure:
    # Detect rhel5.i386 guest, install rhel5.i386.kernel
    # Configure virtio drivers
    # Remap drives in /etc/fstab
    # guestos->prepare_boot

hvtarget->configure-metadata(newguest->get_libvirt_dom)
    # Set /domain/@type and /domain/os
    # Configure device types (virtio)

newguest->create