Ceph » Orchestrator

How to manually (re-)deploy OSDs¶

In order to manually deploy cephadm OSDs, first run ceph-volume (skip this step if needed)

(OSD host)$ cephadm ceph-volume lvm batch --no-auto /dev/mydatadevice1 /dev/mydatadevice2 --db-devices /dev/mydbdevice --yes --no-systemd

Then print a list of all OSDs:

(OSD host)$ cephadm ceph-volume lvm list

Save the osd-fsid of each osd from this command

For each osd:

Then get or create the auth entry

(Admin host)$ ceph auth get osd.XYZ

Get the ceph.conf:

(Admin host)$ ceph config generate-minimal-conf

Get the container image:

(Admin host)$ ceph config get "osd.xyz" container_image

Create the config json:

(OSD host)$ cat config-json.json
{
"config": "# minimal ceph.conf for 8255263a-a97e-4934-822c-00bfe029b28f\n[global]\n\tfsid = 8255263a-a97e-4934-822c-00bfe029b28f\n\tmon_host = [v2:192.168.178.28:40483/0,v1:192.168.178.28:40484/0]\n",
"keyring": "[osd.XYZ]\n\tkey = AQDSb0ZfJqfDNxAAhgAVuH8Wg0yaRUAHwXdTQA==\n" 
}

Deploy the daemon:

(OSD host)$ cephadm --image <container-image> deploy --fsid <fsid> --name osd.xyz --config-json config-json.json --osd-fsid <osd-fsid>