Ok it seems that OSD pools are finally to correctly restored (quite same
amount of objects/datas than before restoration) so that should explain the
situation
I will have to dig why

Vincent

Le ven. 7 juin 2019 à 08:41, Vincent Pharabot <[email protected]>
a écrit :

> Hello Cephers,
>
>
>
> I’m trying to understand Cephfs design and specially how file system view
> reflect OSD pools in order to perform a backup / restore operation (VM
> context).
>
>
>
> I’m able to backup and restore sucessfully OSDs but I have some issues
> with filesystem layer.
>
> When I create files after backup time, after OSD restoration I’m still
> seeing these files even if they are not usable (since osd do not have
> knowledge of these).
>
>
>
> I tried to shutdown/start fs before/after restoration but no change
>
> Same no luck with journal export/import
>
>
>
> I had a look at http://docs.ceph.com/docs/mimic/cephfs/disaster-recovery/
> but since I do not really know what’s beneath all the tools/procedures, I
> don’t know how to start with this
>
>
>
> I know there’s a really usefull feature called cephfs snapshot (.snap) but
> for some reasons I could not use it here.
>
>
>
> Could someone kind enough to guide me ?
>
>
>
> My goal is to restore datas and cephfs layer.
>
> OSD restoration seems ok but how to make cephfs consistent with the OSD
> pool restoration ?
>
> Is there some cache files somewhere on MDS to refresh ? (did not find any)
>
> Or maybe something wrong with the OSD metadata pool linked to my cephfs fs
> but I don’t really know how to check
>
>
>
> I’m using Ceph Mimic (ceph version 13.2.5
> (cbff874f9007f1869bfd3821b7e33b2a6ffd4988) mimic (stable)) with 3 MDS / 1
> active and bluestore
>
>
>
> Thank you for your precious help !
>
>
>
> Vincent
>
_______________________________________________
ceph-users mailing list
[email protected]
http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com

Reply via email to