gpt partition or not when giving ceph-nano a device?

[icepic]

Lchanged ownership of '/var/lib/ceph/osd/ceph-0' from root:root to ceph:ceph
J++/opt/ceph-container/bin/demo.sh:112: bootstrap_osd(): [[ -n /dev/sda ]]
`++/opt/ceph-container/bin/demo.sh:113: bootstrap_osd(): ceph-volume lvm prepare --data /dev/sda
Husage: ceph-volume lvm prepare [-h] --data DATA [--data-size DATA_SIZE]
G                               [--data-slots DATA_SLOTS] [--filestore]
3                               [--journal JOURNAL]
K                               [--journal-size JOURNAL_SIZE] [--bluestore]
5                               [--block.db BLOCK_DB]
?                               [--block.db-size BLOCK_DB_SIZE]
A                               [--block.db-slots BLOCK_DB_SLOTS]
7                               [--block.wal BLOCK_WAL]
A                               [--block.wal-size BLOCK_WAL_SIZE]
C                               [--block.wal-slots BLOCK_WAL_SLOTS]
G                               [--osd-id OSD_ID] [--osd-fsid OSD_FSID]
=                               [--cluster-fsid CLUSTER_FSID]
I                               [--crush-device-class CRUSH_DEVICE_CLASS]
:                               [--dmcrypt] [--no-systemd]
Uceph-volume lvm prepare: error: GPT headers found, they must be removed on: /dev/sda

2020/05/05 11:32:46 Please open an issue at: https://github.com/ceph/cn with the logs above.

ok, "GPT must be removed", doing that.

# sgdisk -Z /dev/sda
GPT data structures destroyed! You may now partition the disk using fdisk or
other utilities.
# dd if=/dev/zero of=/dev/sda bs=1M count=4096

$ cn cluster start ... -b /dev/sda 

If you gave me a whole device, make sure it has a partition table (e.g: gpt). 
If you gave me a partition, I don't support partitions yet, give me a whole device.
As an alternative, you can create a filesystem on this partition and give the mountpoint to me.

Also if the disk was an OSD you need to zap it (e.g: with 'ceph-disk zap').