For SSD/NVMe and other non-rotating storage … overwriting the “same” location doesn’t really work. So, if one releases with “erase disks+secure erase” … and it fails, the current behavior does the traditional overwrite … which is very slow, not legitimately effective, and shortens the life of the device. It would be helpful to allow the user option of “release failed” (better, erase failed) rather than going ahead with an inadequate fallback.
As a related request, provide a hook to allow a user provided script to execute from the rack controller (to exploit OOB erase facilities on appropriate systems (e.g. Dell via ssh/racadm, supermicro via sum, etc.). The ideal (for service providers) would be to sanitize the system and then inform a NaaS system to swing the system back to the "build/MAAS network).