Ceph osd异常退出故障处理
如果一个硬盘故障导致osd节点出现如下的down状态，且一直无法恢复（ reweight列等于0，表示osd已经out此集群）# ceph osd tree
# id weighttype name up/down reweight
-1 4 root default
-2 1 host os-node5
0 1 osd.24 down 0
通过命令获得down状态的osd的ID
osd_id =` ceph osd tree | grep down | grep osd | awk '{print $3}' | awk -F . '{print $2}`
1）、在集群中删除一个osd硬盘
# ceph osd rm 24
removed osd.24
2）、在集群中删除一个osd 硬盘 crush map
# ceph osd crush rm osd.24
removed item id 24 name 'osd.24' from crush map
3）、删除此osd在ceph集群中的认证
# ceph auth del osd.24
updated
4）、卸载osd所挂载的硬盘
# umount /var/lib/ceph/osd/ceph-24
摘掉osd的脚本如下
osd_id=`ceph osd tree | grep down | grep osd | awk '{print $3}' | awk -F . '{print $2}'`
ceph osd rm ${osd_id}
ceph osd crush rm osd.${osd_id}
ceph auth del osd.${osd_id}
umount /var/lib/ceph/osd/ceph-${osd_id}
更换完硬盘后再把此硬盘重新加入集群osd
osd_id=`ceph osd create`
mkfs.xfs -f /dev/sdf
mount /dev/sdf /var/lib/ceph/osd/ceph-${osd_id}
mount -o remount,user_xattr /var/lib/ceph/osd/ceph-${osd_id}
ceph-osd -i ${osd_id} --mkfs --mkkey
ceph auth add osd.${osd_id} osd 'allow *' mon 'allow profile osd' -i /var/lib/ceph/osd/ceph-${osd_id}/keyring
touch /var/lib/ceph/osd/ceph-${osd_id}/sysvinit
/etc/init.d/ceph start osd.${osd_id}
页:
[1]