[pve-devel] [PATCH-SERIES qemu-server] migration: nbd alloc: improve format fallback

[Fiona Ebner]

Currently, remote migration behaves a bit differently, because an
explicitly passed-in format that is not supported by the target
storage is not overwritten with the storage's default format.

This meant remote live migration with qcow2 to e.g. LVM-thin would not
work, because the code here wrongly tried to allocate a qcow2 disk.

There also is a qemu_img_format() call which uses the target storage's
$scfg and the source storage's volume name, which is not very nice
either. Change that first and explain how it only affects certain edge
cases and then clean up the whole mess by moving, getting the format
hint to the call side for local migration too.

Fiona Ebner (2):
  migration: alloc nbd disks: base format hint off source storage
  migration: alloc nbd disks: fix fall-back for remote live migration

 PVE/API2/Qemu.pm  |  1 -
 PVE/QemuServer.pm | 32 +++++++++-----------------------
 2 files changed, 9 insertions(+), 24 deletions(-)

-- 
2.39.2