ceph 0.72.2 on SL6.5 from offical repo. After down one of OSDs (for further the sever out) one of PGs become incomplte: $ ceph health detail HEALTH_WARN 1 pgs incomplete; 1 pgs stuck inactive; 1 pgs stuck unclean; 2 requests are blocked > 32 sec; 1 osds have slow requests pg 4.77 is stuck inactive since forever, current state incomplete, last acting [7,2] pg 4.77 is stuck unclean since forever, current state incomplete, last acting [7,2] pg 4.77 is incomplete, acting [7,2] 2 ops are blocked > 2097.15 sec 2 ops are blocked > 2097.15 sec on osd.7 1 osds have slow requests
$ ceph pg map 4.77 osdmap e21759 pg 4.77 (4.77) -> up [7,2] acting [7,2] $ ceph osd tree # id weight type name up/down reweight -1 17.26 root default -2 2.64 host s01 4 0 osd.4 up 1 5 0.88 osd.5 up 1 0 0.88 osd.0 up 1 7 0.88 osd.7 up 1 -3 2.06 host s02 3 0.3 osd.3 up 1 6 0.88 osd.6 up 1 2 0.88 osd.2 up 1 1 0 osd.1 up 1 -4 5.38 host s03 9 1.79 osd.9 up 1 10 1.79 osd.10 up 1 14 1.8 osd.14 up 1 15 0 osd.15 up 1 -5 7.18 host s04 8 1.79 osd.8 up 1 11 1.79 osd.11 up 1 12 1.8 osd.12 up 1 13 1.8 osd.13 up 1 Ceph used as a big ext4 storage image over rbd. The problem is then the Visrtual Machine (SL6) do request to the PG it's freezing (as some ext4 metadata hits PG, storage become absolutly unusable). How can I "unincompleted" that? I tried "pg force_create_pg", the PG lost all OSDs, but not created anywhere (until restart osd.7, then it is incomplte again).
pgpApk0MBvupE.pgp
Description: PGP signature
_______________________________________________ ceph-users mailing list ceph-users@lists.ceph.com http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com