You can find the inconsistence pg via " ceph pg dump" and then run "ceph pg repair <pg_name>"
On Wed, Apr 30, 2014 at 5:00 PM, vernon1...@126.com <vernon1...@126.com> wrote: > Hi, > I have some problem now. A large number of osds have down before. When some > of them become up, I found a pg was "incomplete". Now this pg's map is > [35,29,42]. > the pg's folders in osd.35 and osd.29 are empty. But there are 9.2G capacity > in osd.42. Like this: > > ----here is osd.35 > [root@ceph952 49.6_head]# ls > [root@ceph952 49.6_head]# > > ----here is osd.42 > [root@ceph960 49.6_head]# ls > DIR_6 DIR_E > [root@ceph960 49.6_head]# > > I want to know how to repair this pg? > And I found, when i stop osd.35, the map change like [0,29,42]. I run "ceph > pg 49.6 query", and it show me: > > [root@ceph960 ~]# ceph pg 49.6 query > ... ... > "probing_osds": [ > "(0,255)", > "(7,255)", > "(20,255)", > "(21,255)", > "(25,255)", > "(26,255)", > "(29,255)", > "(33,255)", > "(34,255)", > "(35,255)", > "(39,255)", > "(41,255)", > "(42,255)"], > "down_osds_we_would_probe": [ > 38], > "peering_blocked_by": []}, > { "name": "Started", > "enter_time": "2014-04-30 16:52:24.181956"}]} > > Can I delete all this "probing_osds" but 42, and set the osd.42 as the > up_primary ? > > Thanks. > > ________________________________ > vernon1...@126.com > > _______________________________________________ > ceph-users mailing list > ceph-users@lists.ceph.com > http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com > -- Best Regards, Wheat _______________________________________________ ceph-users mailing list ceph-users@lists.ceph.com http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com