[問題] 刪除 ceph-node1 上的 osd.4 節點的方法？
[解答] 一般步驟是將 osd.4 標記為out， 停止 osd.4 對應的 OSD 服務
， 將 osd.4 從 cursh map中刪除，將 osd.4 對應 osd data 和 osd journal銷燬
[email protected]:~/my-cluster$ ceph osd tree
ID WEIGHT TYPE NAME UP/DOWN REWEIGHT PRIMARY-AFFINITY
-1 0.07149 root default
-2 0.02859 host ceph-node2
0 0.01430 osd.0 down 0 1.00000
2 0.01430 osd.2 up 1.00000 1.00000
-3 0.01430 host ceph-node3
1 0.01430 osd.1 down 0 1.00000
-4 0.01430 host ceph-node1
4 0.01430 osd.4 up 1.00000 1.00000
-5 0.01430 host ceph-client-node
3 0.01430 osd.3 up 1.00000 1.00000

[email protected]:~/my-cluster$ ssh ceph-node1
Welcome to Ubuntu 14.04.5 LTS (GNU/Linux 4.4.0-31-generic x86_64)

• System information as of Sun Feb 26 11:28:25 CST 2017

  System load: 0.43 Processes: 115
  Usage of /: 47.6% of 14.64GB Users logged in: 0
  Memory usage: 26% IP address for eth0: 192.168.1.249
  Swap usage: 0%

New release ‘16.04.2 LTS’ available.
Run ‘do-release-upgrade’ to upgrade to it.

Last login: Sun Feb 26 11:28:25 2017

將 osd.4 設定為 out 狀態

ceph osd out osd.4

停止 ceph-node1 上的 osd.4 的服務程序

sudostopceph−osdid=4ceph−osdstop/waiting ceph osd tree 檢視 osd.4 進入到 down 狀態。
[email protected]:~cep

hosdtreeIDWEIGHTTYPENAMEUP/DOWNREWEIGHTPRIMARY−AFFINITY−10.04289rootdefault−20.01430hostceph−node220.01430osd.2up1.000001.00000−30.01430hostceph−node330.01430osd.3up1.000001.00000−40.01430hostceph−node140.01430osd.4down01.00000將osd.4從crushmap中刪除掉harvis@ceph−node1: ceph osd tree
ID WEIGHT TYPE NAME UP/DOWN REWEIGHT PRIMARY-AFFINITY
-1 0.02859 root default
-2 0.01430 host ceph-node2
2 0.01430 osd.2 up 1.00000 1.00000
-3 0.01430 host ceph-node3
3 0.01430 osd.3 up 1.00000 1.00000
-4 0 host ceph-node1
-5 0 host ceph-client-node
0 0 osd.0 down 0 1.00000
1 0 osd.1 down 0 1.00000
4 0 osd.4 down 0 1.00000

執行 auth 刪除命令

[email protected]:~cephauthdelosd.4updatedharvis@ceph−node1: ceph osd rm 4
removed osd.4
刪除 osd.4 對應的儲存目錄(這通常是個軟連結)
sudo rm -fr /var/lib/ceph/osd/ceph-4
這裡對應的實際目錄是 /var/local/osd3/ 將它刪除掉 sudo rm -fr /var/local/osd3

如果 osd.4 對應的裝置，可以使用ceph-disk zap命令將裝置恢復到建立分割槽前的狀態(因為ceph-disk zap會刪除所有分割槽包括分割槽表)

[問題] heartbeat_map reset_timeout ‘OSD::osd_op_tp thread 0x7f104a89a700’ had timed out after 4
[解答]