Feature #18978
openceph-disk: Implement easy replace of OSD
0%
Description
As discussed here: http://marc.info/?l=ceph-devel&m=148717797822608&w=2
Maybe something like:
ceph-disk prepare --replace --bluestore /dev/sdX
This will then fetch the current OSD's UUID, wipe the whole disk, reformat it with BlueStore (if supplied) and start the OSD again.
This needs some work in the Monitors to allow the bootstrap-osd key to do this, but this ticket is there to track this work.
Updated by Harry Coin over 6 years ago
I urge raising the priority of this feature request because among the most common ceph task is replacing a failed drive. The present list of steps in the approved documented method number 16.
As someone evaluating alternatives, the knowledge required to do this mostly physical task is 10x more complicated than any other.
Accomplishing the task of dramatically simplifying failed drive replacement would reduce a significant barrier to ceph adoption. That it takes so very many steps to do a most common task is something of a red flag regarding usability confidence.