Ceph » rbd

Jason Dillaman wrote:

... if the cluster minimum client release is set to luminous or later (i.e. trash is guaranteed to be supported). This will help resolve some of the race conditions where you can start deleting an image but another client can open the image and start writing to the image while it's being deleted.

and hold them there until 'rbd trash purge' is called? Equivalent to making this config option "rbd_move_to_trash_on_remove" default to 'true'?

@Mahati: it would first move the image to the trash unconditionally and then if "rbd_move_to_trash_on_remove == false" it would proceed to delete it from the trash.