Hi,
as outlined above the behavior is still wrong.
The patch suggested in [1] will fix the behavior of libvirt to what it was meant to be when preallocation=falloc was added.
But now virt-manager comes into play, the XML sent by virt-manager is exactly matching the definition of the use case to use falloc.
So virt-manager triggered storage request will continue to use falloc despite libvirt being "fixed to how it was intended".
The question now is should we consider changing the XML virt-manager submits so that we are back at the much faster preallocation=metadata mode?
Hi, falloc was added.
as outlined above the behavior is still wrong.
The patch suggested in [1] will fix the behavior of libvirt to what it was meant to be when preallocation=
But now virt-manager comes into play, the XML sent by virt-manager is exactly matching the definition of the use case to use falloc.
So virt-manager triggered storage request will continue to use falloc despite libvirt being "fixed to how it was intended".
The question now is should we consider changing the XML virt-manager submits so that we are back at the much faster preallocation= metadata mode?
[1]: https:/ /www.redhat. com/archives/ libvir- list/2020- September/ msg00105. html