I think because ceph-disk or ceph-deploy doesn't support --osd-uuid.
On Wed, Dec 31, 2014 at 10:30 AM, Andrey Korolyov <andrey@xxxxxxx> wrote:
On Wed, Dec 31, 2014 at 8:20 PM, Wido den Hollander <wido@xxxxxxxx> wrote:
> On 12/31/2014 05:54 PM, Andrey Korolyov wrote:
>> On Wed, Dec 31, 2014 at 7:34 PM, Wido den Hollander <wido@xxxxxxxx> wrote:
>>> Hi,
>>>
>>> Is there a way to set a OSD to exists,new in the OSDMap? I want to
>>> re-install a OSD and re-use the ID and it's key.
>>>
>>> I know I can remove the OSD and re-add it, but that triggers balancing
>>> and I want to prevent that by simply marking the OSD a new and booting
>>> it with a freshly formatted XFS filesystem.
>>>
>>
>> Yes, you can call mkfs with existing UUID: --osd-uuid xxx.
>>
>
> Ah, but will that also mark the OSD as new in the OSDMap? Will it
> receive all old maps?
>
Yes, technically I do not understand why most replacement guides are
going through out-in plus optional rm-add procedure, it works
perfectly with freshly formatted filestore using existing uuid and
key.
_______________________________________________
ceph-users mailing list
ceph-users@xxxxxxxxxxxxxx
http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
_______________________________________________ ceph-users mailing list ceph-users@xxxxxxxxxxxxxx http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
