CEPH換盤

Filestore類型OSD的換盤操做：

一、umount壞盤目錄

umount /var/lib/ceph/osd/ceph-176

二、經過MegaCli建立Riad0硬盤

MegaCli64 -CfgLdAdd -r0 [32:3] WT RA Direct -a0

三、對磁盤分區

parted /dev/sdj "mklabel gpt"
parted /dev/sdj "mkpart primary 1 -1"

四、格式化硬盤並掛載到OSD目錄

mkfs.xfs -m uuid=「$(ceph osd dump |grep osd.176 |awk ‘{print $NF}')」 /dev/sdj1
mount /dev/sdj1 /var/lib/ceph/osd/ceph-176

五、恢復fsid和journal

cd /var/lib/ceph/osd/ceph-176
echo 「$(blkid -o export /dev/sdj1 |grep -i ^uuid |awk -F= ‘{print $2}’)」 > fsid
ln -sf /dev/disk/ceph/by-uuid/[journal_id] journal

六、初始化OSD數據目錄

dd if=/dev/zero of=/dev/sdo3 bs=100M count=1
ceph-osd -i 176 —mkfs
chown -R ceph:ceph /var/lib/ceph/osd/ceph-176

七、啓動OSD並加入集羣

systemctl start ceph-osd@176
ceph osd in 176

Bluestore類型OSD的換盤操做：

一、刪除OSD目錄下ready和mkfs_done文件

cd /var/lib/ceph/osd/ceph-1/
rm -f mkfs_done ready

二、經過MegaCli建立Riad0硬盤

MegaCli64 -CfgLdAdd -r0 [32:3] WT RA Direct -a0

三、生成OSD block連接

cd /dev/disk/ceph/by-uuid/
ln -sf /dev/sde d93d1c35-ee59-4b64-a339-8d00915a08df

四、初始化OSD目錄

ceph-osd -i 1 —mkfs
chown ceph:ceph -R /var/lib/ceph/osd/ceph-1
systemctl start ceph-osd\@1