zombie pgs might occured when i remove some data pools.
but, with pgs in stale state, i can't delete it?

I found this guide, but I don't understand it.
http://ceph.com/docs/next/dev/osd_internals/pg_removal/

Thanks!
--tuantaba

On 07/18/2013 09:22 AM, Ta Ba Tuan wrote:
> I'm using Ceph-0.61.4,
> I removed each osds (2TB) on data hosts and re-create with disks (4TB).
> When converting finish, Ceph warns that have 4 pgs in stale state and 
> warning: i don't have pgid <pgid>
> after, I created 4 pgs by command: ceph pg force_create_pg <pgid>
>
> Now (after the long time), Ceph still warning: "pgmap v57451: 22944 
> pgs: *4 creating*, 22940 active+clean;"
>
> I don't know how to remove those pgs?.
>  Please guiding this error help me!
>
> Thank you!
> --tuantaba
> TA BA TUAN
>
>
> On 07/18/2013 01:16 AM, Samuel Just wrote:
>> What version are you running?  How did you move the osds from 2TB to 4TB?
>> -Sam
>>
>> On Wed, Jul 17, 2013 at 12:59 AM, Ta Ba Tuan<tuantb-QlevPasa8l681eZEIcUDRw@public.gmane.org>  wrote:
>>> Hi everyone,
>>>
>>> I converted every osds from 2TB to 4TB, and when moving complete, show log
>>> Ceph realtime"ceph -w":
>>> displays error: "I don't have pgid 0.2c8"
>>>
>>> after then, I run:  "ceph pg force_create_pg 0.2c8"
>>> Ceph warning: pgmap v55175: 22944 pgs: 1 creating, 22940 active+clean, 3
>>> stale+active+degraded
>>>
>>> then, I can't read/write data to mounted CephFS on Client-side ==> notify on
>>> client side: "Operation not permitted"
>>>
>>> now, "ceph -w" still notify "22944 pgs: 1 creating, 22940 active+clean, 3
>>> stale+active+degraded"
>>> and I don't understand something occuring?
>>> Please, help me!!!
>>>
>>> Thanks to everyone.
>>>
>>> --tuantaba
>>>
>>>
>>>
>>>
>>> _______________________________________________
>>> ceph-users mailing list
>>> ceph-users-idqoXFIVOFJgJs9I8MT0rw@public.gmane.org
>>> http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
>>>
>
>
>
> _______________________________________________
> ceph-users mailing list
> ceph-users-idqoXFIVOFJgJs9I8MT0rw@public.gmane.org
> http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com