Eugen, thanks! That looks like a nice improvement over what I'm doing now. I'll try that.
On Tue, Jul 3, 2018 at 3:15 AM Eugen Block <eblock@xxxxxx> wrote:
Hi,
we had to recreate some block.db's for some OSDs just a couple of
weeks ago because our existing journal SSD had failed. This way we
avoided to rebalance the whole cluster, just the OSD had to be filled
up. Maybe this will help you too.
http://heiterbiswolkig.blogs.nde.ag/2018/04/08/resetting-an-existing-bluestore-osd/
Regards,
Eugen
Zitat von Caspar Smit <casparsmit@xxxxxxxxxxx>:
> 2018-07-03 4:27 GMT+02:00 Brad Fitzpatrick <brad@xxxxxxxxx>:
>
>> Hello,
>>
>> I was wondering if it's possible or how best to add a DB & WAL to an OSD
>> retroactively? (Still using Luminous)
>>
>> I hurriedly created some HDD-backed bluestore OSDs without their WAL & DB
>> on SSDs, and then loaded them up with data.
>>
>> When I realized I forgot to use the SSD partitions for its WAL & DB, I
>> marked the first OSD out, let it drain, nuked it, and recreated it, and am
>> now waiting for it to re-populate.
>>
>> In the meantime, I've been searching for how to best add them
>> retroactively on the remaining OSDs.
>>
>> I found https://swamireddy.wordpress.com/2016/02/19/ceph-
>> how-to-add-the-ssd-journal/ which seems to describe how to add the DB,
>> but not the WAL.
>>
>>
> Hi Brad, these steps are for Filestore backed OSD's only. Afaik there
> currently isn't a procedure to seperate the DB/WAL after OSD creation, so
> your current method of marking out, drain and re-create is your only option.
>
> Kind regards,
> Caspar
>
>
>
>> Any tips?
>>
>> Thanks!
>>
>> - Brad
>>
>>
>>
>> _______________________________________________
>> 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
_______________________________________________ ceph-users mailing list ceph-users@xxxxxxxxxxxxxx http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
