ceph-deploy osd create / prepare logic

gistfile1.txt

1) if ceph-deploy osd create/prepare is passed a whole device (ie sda) then it will format the device and create two partitions, one for the journal, and one of the osd its self

2) if ceph-deploy osd create/prepare is passed a partitioned device (ie sda1) it formats the partition and creates a journal FILE inside the partition 

3) if ceph-deploy osd create/prepare is passed a mounted path, ie (/mnt/osd) it will do create a journal FILE in side the mount path

4) in the case of ceph-deploy being passed a device/sub-device/path and a journal file-path (ie sda:/mounts/journal/journal-sda) it will do whichever is relevant (a whole device will get a partition and mkfs) (a subdevice will get a mkfs) and a journal FILE will be created at the journal file-path. from my experience, if the journal file-path is invalid it will default to its default location for the journal.

examples:

ceph-deploy osd create node-1:sda

result:
/dev/sda1 journal: part, mkfs, mount
/dev/sda2 osd:     part, mkfs, mount

ceph-deploy osd create node-1:sda1
/dev/sda1 osd:     part, mkfs, mount
sda1/journal:      zeroed

ceph-deploy osd create node-1:/mnt/sda1
/mnt/sda1/ osd:    noop
/mnt/sda1/journal: zeroed

ceph-deploy osd create node-1:[sda|sda1|/mnt/sda1]:/mnt/ssd1/journal-sda
osd: as in other examples
/mnt/ssd/journal-sda: zeroed