Downing a node for replacement....
referencing:
http://ceph.com/docs/master/rados/operations/add-or-rm-osds/

take out the osd's

reweight to drive the data off

#for i in {0..23}; do ceph osd reweight $i .1; done

Take the OSD's out of service

#for i in {0..23}; do ceph osd out $i; done

Remove them from the crush map.

#for i in {0..23}; do ceph osd crush remove osd.$i; done

Remove them from mon auth

#for i in {0..23}; do ceph auth del osd.$i; done

Stop them

/etc/init.d/ceph stop

Delete them. This TOTALLY didn't work.... WTF?  First run complained that they were not down.

for i in {0..23}; do ceph auth del osd.$i; done

And they show up in the tree. Eventually most went down....

...

profit