Comment 3 for bug 2035225

well,
I try to describe in depth my setup.
I have a 4-node nautilus ceph on ubuntu 18.04-LTS and 2 other ubuntu server for VMs (op1 and op3).
op1 is an ubuntu 18.04-lts using virsh/rbdmap for vm disks while op3 is an ubuntu 20.04-lts and uses virsh/libceph (by libvirt ceph pool).
The ceph pool is the same for the 2 servers op1 and op3.
The problem I noted is always on op3 where for example It takes quite 5 minutes to have the vol list from the pool (virsh vol-list POOLNAME).
If I use virt-manager the ssh connection (qemu+ssh) takes always a lot of minutes to complete while the logs are full of

Sep 12 15:15:10 op3 libvirtd[6987]: failed to open the RBD image 'A2_ST_MIUR': No such file or directory
Sep 12 15:15:44 op3 libvirtd[6987]: failed to open the RBD image 'ndc01-1': No such file or directory
Sep 12 15:16:16 op3 libvirtd[6987]: failed to open the RBD image 'arss-sys': No such file or directory

I suppose that virt-manager try to obtain the volumes list of the pool, but it takes a lot of time and sometimes doesn't find a lot of volumes that exist but are in use by the op1 server using rbdmap.