Ubuntu 20.04
Sponsored Link

Ceph Octopus : Add or Remove OSDs
2020/08/31
 
This is how to add or remove OSDs from exisiting Cluster.
                                         |
        +--------------------+           |           +----------------------+
        |   [dlp.srv.world]  |10.0.0.30  |  10.0.0.31|    [www.srv.world]   |
        |     Ceph Client    +-----------+-----------+        RADOSGW       |
        |                    |           |           |                      |
        +--------------------+           |           +----------------------+
            +----------------------------+----------------------------+
            |                            |                            |
            |10.0.0.51                   |10.0.0.52                   |10.0.0.53 
+-----------+-----------+    +-----------+-----------+    +-----------+-----------+
|   [node01.srv.world]  |    |   [node02.srv.world]  |    |   [node03.srv.world]  |
|     Object Storage    +----+     Object Storage    +----+     Object Storage    |
|     Monitor Daemon    |    |                       |    |                       |
|     Manager Daemon    |    |                       |    |                       |
+-----------------------+    +-----------------------+    +-----------------------+

[1] For example, Add a [node04] node to OSDs on Admin Node.
For Block device on new [node04] Node, use [/dev/sdb] on this example.
# transfer public key

root@node01:~#
ssh-copy-id node04

# install required packages

root@node01:~#
ssh node04 "apt update; apt -y install ceph"
# transfer required files

root@node01:~#
scp /etc/ceph/ceph.conf node04:/etc/ceph/ceph.conf

root@node01:~#
scp /etc/ceph/ceph.client.admin.keyring node04:/etc/ceph

root@node01:~#
scp /var/lib/ceph/bootstrap-osd/ceph.keyring node04:/var/lib/ceph/bootstrap-osd
# configure OSD

root@node01:~# ssh node04 \
"chown ceph. /etc/ceph/ceph.* /var/lib/ceph/bootstrap-osd/*; \
parted --script /dev/sdb 'mklabel gpt'; \
parted --script /dev/sdb "mkpart primary 0% 100%"; \
ceph-volume lvm create --data /dev/sdb1" 
Running command: /usr/bin/ceph-authtool --gen-print-key
Running command: /usr/bin/ceph --cluster ceph --name client.bootstrap-osd --keyring /var/lib/ceph/bootstrap-osd/ceph.keyring -i - osd new 2c941f07-96f7-4b51-91fd-5296a102eaa9
Running command: /usr/sbin/vgcreate --force --yes ceph-d299b0ff-e127-4de7-a0f4-bf8dffd95308 /dev/vdb1
 stdout: Physical volume "/dev/vdb1" successfully created.
 stdout: Volume group "ceph-d299b0ff-e127-4de7-a0f4-bf8dffd95308" successfully created
Running command: /usr/sbin/lvcreate --yes -l 100%FREE -n osd-block-2c941f07-96f7-4b51-91fd-5296a102eaa9 ceph-d299b0ff-e127-4de7-a0f4-bf8dffd95308
 stdout: Logical volume "osd-block-2c941f07-96f7-4b51-91fd-5296a102eaa9" created.
.....
.....
Running command: /usr/bin/systemctl start ceph-osd@3
--> ceph-volume lvm activate successful for osd ID: 3
--> ceph-volume lvm create successful for: /dev/vdb1

root@node01:~# ceph -s 
  cluster:
    id:     72840c24-3a82-4e28-be87-cf9f905918fb
    health: HEALTH_OK

  services:
    mon: 1 daemons, quorum node01 (age 3h)
    mgr: node01(active, since 31m)
    mds: cephfs:1 {0=node01=up:active}
    osd: 4 osds: 4 up (since 4m), 4 in (since 4m)
    rgw: 1 daemon active (www)

  task status:
    scrub status:
        mds.node01: idle

  data:
    pools:   8 pools, 193 pgs
    objects: 215 objects, 35 KiB
    usage:   4.4 GiB used, 316 GiB / 320 GiB avail
    pgs:     193 active+clean
[2] To remove an OSD Node from existing Cluster, run commands like follows.
For example, Remove [node04] node.
root@node01:~#
ceph -s

  cluster:
    id:     72840c24-3a82-4e28-be87-cf9f905918fb
    health: HEALTH_OK

  services:
    mon: 1 daemons, quorum node01 (age 3h)
    mgr: node01(active, since 32m)
    mds: cephfs:1 {0=node01=up:active}
    osd: 4 osds: 4 up (since 4m), 4 in (since 4m)
    rgw: 1 daemon active (www)

  task status:
    scrub status:
        mds.node01: idle

  data:
    pools:   8 pools, 193 pgs
    objects: 215 objects, 35 KiB
    usage:   4.4 GiB used, 316 GiB / 320 GiB avail
    pgs:     193 active+clean

root@node01:~#
ceph osd tree

ID  CLASS  WEIGHT   TYPE NAME        STATUS  REWEIGHT  PRI-AFF
-1         0.31238  root default
-3         0.07809      host node01
 0    hdd  0.07809          osd.0        up   1.00000  1.00000
-5         0.07809      host node02
 1    hdd  0.07809          osd.1        up   1.00000  1.00000
-7         0.07809      host node03
 2    hdd  0.07809          osd.2        up   1.00000  1.00000
-9         0.07809      host node04
 3    hdd  0.07809          osd.3        up   1.00000  1.00000

# specify OSD ID of a node you'd like to remove

root@node01:~#
ceph osd out 3

marked out osd.3.
# live watch cluster status

# after running [ceph osd out ***], rebalancing is executed automatically

# to quit live watch, push [Ctrl + c]

root@node01:~#
ceph -w

  cluster:
    id:     72840c24-3a82-4e28-be87-cf9f905918fb
    health: HEALTH_WARN
            Reduced data availability: 33 pgs inactive, 15 pgs peering
            Degraded data redundancy: 15/645 objects degraded (2.326%), 4 pgs degraded

  services:
    mon: 1 daemons, quorum node01 (age 3h)
    mgr: node01(active, since 33m)
    mds: cephfs:1 {0=node01=up:active}
    osd: 4 osds: 4 up (since 5m), 3 in (since 8s); 46 remapped pgs
    rgw: 1 daemon active (www)

  task status:
    scrub status:
        mds.node01: idle

  data:
    pools:   8 pools, 193 pgs
    objects: 215 objects, 35 KiB
    usage:   3.3 GiB used, 237 GiB / 240 GiB avail
    pgs:     17.617% pgs unknown
             59.067% pgs not active
             15/645 objects degraded (2.326%)
             95 activating
             45 active+clean
             34 unknown
             9  remapped+peering
             6  peering
             4  activating+degraded

  progress:
    Rebalancing after osd.3 marked out (5s)
      [==..........................] (remaining: 45s)


2020-08-31T19:38:47.069524+0900 mon.node01 [WRN] Health check update: Degraded data redundancy: 160/645 objects degraded (24.806%), 32 pgs degraded (PG_DEGRADED)
2020-08-31T19:38:47.069550+0900 mon.node01 [INF] Health check cleared: PG_AVAILABILITY (was: Reduced data availability: 33 pgs inactive, 15 pgs peering)
.....
.....
2020-08-31T19:40:10.658884+0900 mon.node01 [INF] Health check cleared: PG_DEGRADED (was: Degraded data redundancy: 1 pg undersized)
2020-08-31T19:40:10.658994+0900 mon.node01 [INF] Cluster is now healthy

# after status turns to [HEALTH_OK], disable OSD service on the target node

root@node01:~#
ssh node04 "systemctl disable --now ceph-osd@3.service"

# remove the node to specify target OSD ID

root@node01:~#
ceph osd purge 3 --yes-i-really-mean-it

purged osd.3
root@node01:~#
ceph -s

  cluster:
    id:     72840c24-3a82-4e28-be87-cf9f905918fb
    health: HEALTH_OK

  services:
    mon: 1 daemons, quorum node01 (age 3h)
    mgr: node01(active, since 36m)
    mds: cephfs:1 {0=node01=up:active}
    osd: 3 osds: 3 up (since 28s), 3 in (since 3m)
    rgw: 1 daemon active (www)

  task status:
    scrub status:
        mds.node01: idle

  data:
    pools:   8 pools, 193 pgs
    objects: 215 objects, 35 KiB
    usage:   3.3 GiB used, 237 GiB / 240 GiB avail
    pgs:     193 active+clean
Matched Content