Comment 2 for bug 1374160

Reproduced.

It is possible to manage only one OSD per node under Ubuntu.

Dima, if you tried to stop osd.4 instead of osd.1 on the same node, you would get this error.
Only single OSD is registered in Ubuntu upstart. CentOS has no such issue.

root@node-3:~# ceph osd tree
# id weight type name up/down reweight
-1 0.24 root default
-2 0.12 host node-3
0 0.06 osd.0 up 1
1 0.06 osd.1 up 1
-3 0.12 host node-1
2 0.06 osd.2 up 1
3 0.06 osd.3 up 1

root@node-3:~# ps -ef | grep ceph
root 18414 18295 0 19:15 pts/3 00:00:00 grep --color=auto ceph
root 19989 1 0 Oct14 ? 00:03:14 /usr/bin/ceph-osd --cluster=ceph -i 0 -f
root 20830 1 0 Oct14 ? 00:02:51 /usr/bin/ceph-osd -i 1 --pid-file /var/run/ceph/osd.1.pid -c /etc/ceph/ceph.conf --cluster ceph

root@node-3:~# initctl list | grep ceph
ceph-osd-all start/running
ceph-mds-all start/running
ceph-mds-all-starter stop/waiting
ceph-osd-all-starter stop/waiting
ceph-mon-all start/running
ceph-all start/running
ceph-mon-all-starter stop/waiting
ceph-mon stop/waiting
ceph-create-keys stop/waiting
ceph-osd (ceph/0) start/running, process 19989
ceph-mds stop/waiting

root@node-3:~# stop ceph-osd id=1
stop: Unknown instance: ceph/1