（6）ceph集群osd down 故障處理

（1）查看集群狀態，發現2個osd 狀態為down

[root@node140 /]# ceph -s 
  cluster:
    id:     58a12719-a5ed-4f95-b312-6efd6e34e558
    health: HEALTH_ERR
            noout flag(s) set
            2 osds down
            1 scrub errors
            Possible data damage: 1 pg inconsistent
            Degraded data redundancy: 1633/10191 objects degraded (16.024%), 84 pgs degraded, 122 pgs undersized

  services:
    mon: 2 daemons, quorum node140,node142 (age 3d)
    mgr: admin(active, since 3d), standbys: node140
    osd: 18 osds: 16 up (since 3d), 18 in (since 5d)
         flags noout

  data:
    pools:   2 pools, 384 pgs
    objects: 3.40k objects, 9.8 GiB
    usage:   43 GiB used, 8.7 TiB / 8.7 TiB avail
    pgs:     1633/10191 objects degraded (16.024%)
             261 active+clean
             84  active+undersized+degraded
             38  active+undersized
             1   active+clean+inconsistent

（2）查看osd狀態

[root@node140 /]# ceph  osd tree
ID CLASS WEIGHT  TYPE NAME        STATUS REWEIGHT PRI-AFF 
-1       9.80804 root default                             
-2       3.26935     host node140                         
 0   hdd 0.54489         osd.0        up  1.00000 1.00000 
 1   hdd 0.54489         osd.1        up  1.00000 1.00000 
 2   hdd 0.54489         osd.2        up  1.00000 1.00000 
 3   hdd 0.54489         osd.3        up  1.00000 1.00000 
 4   hdd 0.54489         osd.4        up  1.00000 1.00000 
 5   hdd 0.54489         osd.5        up  1.00000 1.00000 
-3       3.26935     host node141                         
12   hdd 0.54489         osd.12       up  1.00000 1.00000 
13   hdd 0.54489         osd.13       up  1.00000 1.00000 
14   hdd 0.54489         osd.14       up  1.00000 1.00000 
15   hdd 0.54489         osd.15       up  1.00000 1.00000 
16   hdd 0.54489         osd.16       up  1.00000 1.00000 
17   hdd 0.54489         osd.17       up  1.00000 1.00000 
-4       3.26935     host node142                         
 6   hdd 0.54489         osd.6        up  1.00000 1.00000 
 7   hdd 0.54489         osd.7      down  1.00000 1.00000 
 8   hdd 0.54489         osd.8      down  1.00000 1.00000 
 9   hdd 0.54489         osd.9        up  1.00000 1.00000 
10   hdd 0.54489         osd.10       up  1.00000 1.00000 
11   hdd 0.54489         osd.11       up  1.00000 1.00000 

（3）osd 7 osd 8狀態查看，已經failed了，重啟也無法啟動

[root@node140 /]# systemctl status ceph-osd@8.service
● ceph-osd@8.service - Ceph object storage daemon osd.8
   Loaded: loaded (/usr/lib/systemd/system/ceph-osd@.service; enabled; vendor preset: disabled)
   Active: failed (Result: start-limit) since Fri 2019-08-30 17:36:50 CST; 1min 20s ago
  Process: 433642 ExecStartPre=/usr/lib/ceph/ceph-osd-prestart.sh --cluster ${CLUSTER} --id %i (code=exited, status=1/FAILURE)

Aug 30 17:36:50 node140 systemd[1]: Failed to start Ceph object storage daemon osd.8.
Aug 30 17:36:50 node140 systemd[1]: Unit ceph-osd@8.service entered failed state.
Aug 30 17:36:50 node140 systemd[1]: ceph-osd@8.service failed.
Aug 30 17:36:50 node140 systemd[1]: ceph-osd@8.service holdoff time over, scheduling restart.
Aug 30 17:36:50 node140 systemd[1]: Stopped Ceph object storage daemon osd.8.
Aug 30 17:36:50 node140 systemd[1]: start request repeated too quickly for ceph-osd@8.service
Aug 30 17:36:50 node140 systemd[1]: Failed to start Ceph object storage daemon osd.8.
Aug 30 17:36:50 node140 systemd[1]: Unit ceph-osd@8.service entered failed state.
Aug 30 17:36:50 node140 systemd[1]: ceph-osd@8.service failed.

（4）osd硬盤故障，狀態變化
osd硬盤故障，狀態變為down。在經過mod osd down out interval 設定的時間間隔后，ceph將其標記為out，并開始進行數據遷移恢復。 為了降低影響可以先關閉，待硬盤更換完成后再開啟
[root@node140 /]# cat /etc/ceph/ceph.conf
[global]
mon osd down out interval = 900

（5）停止數據均衡
[root@node140 /]# for i in noout nobackfill norecover noscrub nodeep-scrub;do ceph osd set $i;done

（6）定位i故障盤
[root@node140 /]# ceph osd tree | grep -i down
7 hdd 0.54489 osd.7 down 0 1.00000
8 hdd 0.54489 osd.8 down 0 1.00000

（7）卸載故障的節點
[root@node142 ~]# umount /var/lib/ceph/osd/ceph-7
[root@node142 ~]# umount /var/lib/ceph/osd/ceph-8

（8）從crush map 中移除osd
[root@node142 ~]# ceph osd crush remove osd.7
removed item id 7 name 'osd.7' from crush map
[root@node142 ~]# ceph osd crush remove osd.8
removed item id 8 name 'osd.8' from crush map

（9）刪除故障osd的密鑰
[root@node142 ~]# ceph auth del osd.7
updated
[root@node142 ~]# ceph auth del osd.8
updated

（10）刪除故障osd

[root@node142 ~]# ceph osd rm 7 
removed osd.7
[root@node142 ~]# ceph osd rm 8
removed osd.8
[root@node142 ~]# ceph osd tree 
ID CLASS WEIGHT  TYPE NAME        STATUS REWEIGHT PRI-AFF 
-1       8.71826 root default                             
-2       3.26935     host node140                         
 0   hdd 0.54489         osd.0        up  1.00000 1.00000 
 1   hdd 0.54489         osd.1        up  1.00000 1.00000 
 2   hdd 0.54489         osd.2        up  1.00000 1.00000 
 3   hdd 0.54489         osd.3        up  1.00000 1.00000 
 4   hdd 0.54489         osd.4        up  1.00000 1.00000 
 5   hdd 0.54489         osd.5        up  1.00000 1.00000 
-3       3.26935     host node141                         
12   hdd 0.54489         osd.12       up  1.00000 1.00000 
13   hdd 0.54489         osd.13       up  1.00000 1.00000 
14   hdd 0.54489         osd.14       up  1.00000 1.00000 
15   hdd 0.54489         osd.15       up  1.00000 1.00000 
16   hdd 0.54489         osd.16       up  1.00000 1.00000 
17   hdd 0.54489         osd.17       up  1.00000 1.00000 
-4       2.17957     host node142                         
 6   hdd 0.54489         osd.6        up  1.00000 1.00000 
 9   hdd 0.54489         osd.9        up  1.00000 1.00000 
10   hdd 0.54489         osd.10       up  1.00000 1.00000 
11   hdd 0.54489         osd.11       up  1.00000 1.00000 
[root@node142 ~]# 

（11）更換故障硬盤查看盤符，然后重建
[root@node142 ~]# ceph-volume lvm create --data /dev/sdd
[root@node142 ~]# ceph-volume lvm create --data /dev/sdc

（12）
[root@node142 ~]# ceph-volume lvm list
（13）待新osd添加crush map后，重新開啟集群禁用標志
for i in noout nobackfill norecover noscrub nodeep-scrub;do ceph osd unset $i;done