I have a KVM host with a software RAID 10 array. The software RAID 10 array is a PV on the host and there's a single VG.
I have a couple of LVs used at the host level (a datastore for KVM guest
Normally when adding a new disk to a guest I create the LV on the host then add that to the guest's config. Then
pvcreate directly on the block device (no partition),
mkfs on the new block device within the guest.
I don't bother partitioning the block device before
pvcreate because it's one less step to do when resizing the LV on the host. I realise the downside is in theory some systems might decide to automatically partition the unpartitioned block device because they don't recognise the LVM metadata, though I have never ever seen this happen in practice.
When resizing the LV on the host the free space in the PV is picked up and I have to
resize2fs in the guest.
I don't suppose there's a way I can skip having LVM running in the guest and just
lvextend on the host then
resize2fs in the guest?