This is correct, however while watching the content of the snap list filtering by the volume ID it is clear that the snapshot does get created for this empty volume, having a name of "rbd_mirror.<volume_id>", and it stays there for about a second. Maybe this is how the images of the volume gets created on the other side of the mirror (list a metadata sync), it is more a question to the behaviour of ceph-rbd-mirror daemon itself.
This is correct, however while watching the content of the snap list filtering by the volume ID it is clear that the snapshot does get created for this empty volume, having a name of "rbd_mirror. <volume_ id>", and it stays there for about a second. Maybe this is how the images of the volume gets created on the other side of the mirror (list a metadata sync), it is more a question to the behaviour of ceph-rbd-mirror daemon itself.