008 Ceph集群数据同步

介绍，目前已经创建一个名为ceph的Ceph集群，和一个backup（单节点）Ceph集群，是的这两个集群的数据可以同步，做备份恢复功能

一、配置集群的相互访问

1.1 安装rbd mirror

rbd-mirror是一个新的守护进程，负责将一个镜像从一个集群同步到另一个集群

如果是单向同步，则只需要在备份集群上安装

如果是双向同步，则需要在两个集群上都安装

rbd-mirror需要连接本地和远程集群

每个集群只需要运行一个rbd-mirror进程，必须手动按章

[root@ceph5 ceph]# yum install rbd-mirror

1.2 互传配置文件

Ceph集群----->backu盘集群

[root@ceph2 ceph]# scp /etc/ceph/ceph.conf  /etc/ceph/ceph.client.admin.keyring ceph5:/etc/ceph/

[root@ceph5 ceph]# ll -h

[root@ceph5 ceph]# chown ceph.ceph -R ./
[root@ceph5 ceph]# ll

[root@ceph5 ceph]# ceph -s --cluster ceph

cluster:
    id:     35a91e48--4e96-a7ee-980ab989d20d
    health: HEALTH_OK
  services:
    mon:  daemons, quorum ceph2,ceph3,ceph4
    mgr: ceph4(active), standbys: ceph2, ceph3
    osd:  osds:  up,  in
  data:
    pools:    pools,  pgs
    objects:  objects,  MB
    usage:    MB used,  GB /  GB avail
    pgs:      active+clean

1.3 把配置文件从从端传到主端

[root@ceph5 ceph]# scp /etc/ceph/backup.conf  /etc/ceph/backup.client.admin.keyring ceph2:/etc/ceph/

[root@ceph2 ceph]# ll

[root@ceph2 ceph]# chown -R ceph.ceph /etc/ceph/

[root@ceph2 ceph]# ll

[root@ceph2 ceph]#  ceph -s --cluster backup

cluster:
    id:     51dda18c--4edb-8ba9-27330ead81a7
    health: HEALTH_OK

  services:
    mon:  daemons, quorum ceph5
    mgr: ceph5(active)
    osd:  osds:  up,  in

  data:
    pools:    pools,  pgs
    objects:  objects,  bytes
    usage:    MB used,  MB /  MB avail
    pgs:      active+clean

二、 配置镜像为池模式

如果是单向同步，则只需要在主集群修改，如果是双向同步，需要在两个集群上都修改，(如果只启用单向备份，则不需要在备份集群上开启镜像模式）

rbd_default_features = 125

也可不修改配置文件，而在创建镜像时指定需要启用的功能

2.1 创建镜像池

[root@ceph2 ceph]# ceph osd pool create rbdmirror 32 32

[root@ceph5 ceph]# ceph osd pool create rbdmirror 32 32

[root@ceph2 ceph]# yum -y install rbd-mirror

[root@ceph2 ceph]# rbd mirror pool enable rbdmirror pool --cluster ceph

[root@ceph2 ceph]# rbd mirror pool enable rbdmirror pool --cluster backup

2.2 增加同伴集群

需要将ceph和backup两个集群设置为同伴，这是为了让rbd-mirror进程找到它peer的集群的存储池

[root@ceph2 ceph]# rbd mirror pool peer add rbdmirror client.admin@ceph --cluster backup

[root@ceph2 ceph]# rbd mirror pool peer add rbdmirror client.admin@backup --cluster ceph

[root@ceph2 ceph]#  rbd mirror pool info --pool=rbdmirror --cluster backup

[root@ceph2 ceph]#  rbd mirror pool info --pool=rbdmirror --cluster ceph

[root@ceph2 ceph]#  rbd mirror pool status rbdmirror

[root@ceph2 ceph]#  rbd mirror pool status rbdmirror --cluster backup

2.3 创建RBD并开启journaling

在rbdmirror池中创建一个test镜像

[root@ceph2 ceph]# rbd create --size 1G rbdmirror/test

[root@ceph2 ceph]# rbd ls rbdmirror

在从端查看，为空并没有同步，是因为没有开启journing功能，从端没有开启mirror

[root@ceph2 ceph]# rbd ls rbdmirror --cluster backup

[root@ceph2 ceph]# rbd info rbdmirror/test

rbd image 'test':
    size  MB in  objects
    order  ( kB objects)
    block_name_prefix: rbd_data.fbcd643c9869
    format:
    features: layering, exclusive-lock, object-map, fast-diff, deep-flatten
    flags:
    create_timestamp: Mon Mar  :: 

[root@ceph2 ceph]# rbd feature enable rbdmirror/test journaling

[root@ceph2 ceph]# rbd info rbdmirror/test

rbd image 'test':
    size  MB in  objects
    order  ( kB objects)
    block_name_prefix: rbd_data.fbcd643c9869
    format:
    features: layering, exclusive-lock, object-map, fast-diff, deep-flatten, journaling
    flags:
    create_timestamp: Mon Mar  ::
    journal: fbcd643c9869
    mirroring state: enabled
    mirroring global id: 0f6993f2-8c34-4c49-a546-61fcb452ff40
    mirroring primary: true                #主是true

2.4 从端开启mirror进程

[root@ceph2 ceph]#  rbd mirror pool status rbdmirror

2.5 前台手动启动调试

[root@ceph5 ceph]# rbd-mirror -d --setuser ceph --setgroup ceph --cluster backup -i admin

-- ::39.521126 7f3a7f7c5340   set uid:gid to : (ceph:ceph)
-- ::39.521150 7f3a7f7c5340   ceph version 12.2.-.el7cp (c6d85fd953226c9e8168c9abe81f499d66cc2716) luminous (stable), process (unknown), pid
-- ::39.544922 7f3a7f7c5340   mgrc service_daemon_register rbd-mirror.admin metadata {arch=x86_64,ceph_version=ceph version 12.2.-.el7cp (c6d85fd953226c9e8168c9abe81f499d66cc2716) luminous (stable),cpu=QEMU Virtual CPU version 1.5.,distro=rhel,distro_description=Red Hat Enterprise Linux Server 7.4 (Maipo),distro_version=7.4,hostname=ceph5,instance_id=,kernel_description=# SMP Thu Dec  :: EST ,kernel_version=3.10.-693.11..el7.x86_64,mem_swap_kb=,mem_total_kb=,os=Linux}
-- ::42.743067 7f3a6bfff700 - rbd::mirror::ImageReplayer: 0x7f3a50016c70 [/0f6993f2-8c34-4c49-a546-61fcb452ff40] handle_init_remote_journaler: image_id=1037140e0f76, m_client_meta.image_id=1037140e0f76, client.state=connected

[root@ceph5 ceph]# systemctl start ceph-rbd-mirror@admin

[root@ceph5 ceph]# ps aux|grep rbd

ceph         1.6  1.5   ?       Ssl  :   : /usr/bin/rbd-mirror -f --cluster backup --id admin --setuser ceph --setgroup ceph

2.6 从端验证

查看已经同步

[root@ceph5 ceph]# rbd ls rbdmirror

[root@ceph5 ceph]# rbd info rbdmirror/test

rbd image 'test':
    size  MB in  objects
    order  ( kB objects)
    block_name_prefix: rbd_data.1037140e0f76
    format:
    features: layering, exclusive-lock, object-map, fast-diff, deep-flatten, journaling
    flags:
    create_timestamp: Mon Mar  ::
    journal: 1037140e0f76
    mirroring state: enabled
    mirroring global id: 0f6993f2-8c34-4c49-a546-61fcb452ff40
    mirroring primary: false        #从为false

现在相当于单向同步，

2.7 配置双向同步

主端手动启一下mirror

[root@ceph2 ceph]#  /usr/bin/rbd-mirror -f --cluster ceph --id admin --setuser ceph --setgroup ceph -d

[root@ceph2 ceph]# systemctl restart ceph-rbd-mirror@admin

[root@ceph2 ceph]# ps aux|grep rbd

ceph        1.0  0.4   ?       Ssl  :   : /usr/bin/rbd-mirror -f --cluster ceph --id admin --setuser ceph --setgroup ceph

从端做主端，创建一个镜像：

[root@ceph5 ceph]# rbd create --size 1G rbdmirror/ceph5-rbd --image-feature journaling --image-feature exclusive-lock

[root@ceph5 ceph]# rbd info rbdmirror/ceph5-rbd

rbd image 'ceph5-rbd':
    size  MB in  objects
    order  ( kB objects)
    block_name_prefix: rbd_data.103e74b0dc51
    format:
    features: exclusive-lock, journaling
    flags:
    create_timestamp: Mon Mar  ::
    journal: 103e74b0dc51
    mirroring state: enabled
    mirroring global id: c41dc1d8-43e5-44eb-ba3f-87dbb17f0d15
    mirroring primary: true

[root@ceph5 ceph]# rbd ls rbdmirror

主端变从端

[root@ceph2 ceph]# rbd info rbdmirror/ceph5-rbd

rbd image 'ceph5-rbd':
    size  MB in  objects
    order  ( kB objects)
    block_name_prefix: rbd_data.fbe541a7c4c9
    format:
    features: exclusive-lock, journaling
    flags:
    create_timestamp: Mon Mar  ::
    journal: fbe541a7c4c9
    mirroring state: enabled
    mirroring global id: c41dc1d8-43e5-44eb-ba3f-87dbb17f0d15
    mirroring primary: false

查看基于池的同步信息

[root@ceph2 ceph]# rbd mirror pool status --pool=rbdmirror

查看image状态

[root@ceph2 ceph]# rbd mirror image status rbdmirror/test

test:
  global_id:   0f6993f2-8c34-4c49-a546-61fcb452ff40
  state:       up+stopped
  description: local image is primary
  last_update: -- ::

三、基于镜像级别的双向同步

把池中的cephrbd1，作为镜像同步

[root@ceph2 ceph]# rbd ls

3.1 先为rbd池开启镜像模式

[root@ceph2 ceph]# rbd mirror pool enable rbd image --cluster ceph

[root@ceph2 ceph]# rbd mirror pool enable rbd image --cluster backup

3.2 增加同伴

[root@ceph2 ceph]# rbd mirror pool peer add rbd client.admin@ceph --cluster backup

[root@ceph2 ceph]# rbd mirror pool peer add rbd client.admin@backup --cluster ceph

[root@ceph2 ceph]# ceph osd pool ls

[root@ceph2 ceph]# rbd info cephrbd1

rbd image 'cephrbd1':
    size  MB in  objects
    order  ( kB objects)
    block_name_prefix: rbd_data.fb2074b0dc51
    format:
    features: layering
    flags:
    create_timestamp: Sun Mar  :: 

3.3 打开journaling

[root@ceph2 ceph]# rbd feature enable cephrbd1 exclusive-lock

[root@ceph2 ceph]# rbd feature enable cephrbd1 journaling

[root@ceph2 ceph]# rbd info cephrbd1

rbd image 'cephrbd1':
    size  MB in  objects
    order  ( kB objects)
    block_name_prefix: rbd_data.fb2074b0dc51
    format:
    features: layering, exclusive-lock, journaling
    flags:
    create_timestamp: Sun Mar  ::
    journal: fb2074b0dc51
    mirroring state: disabled

3.4 开启镜像的mirror功能

[root@ceph2 ceph]# rbd mirror image enable  rbd/cephrbd1

[root@ceph2 ceph]# rbd info rbd/cephrbd1

rbd image 'cephrbd1':
size  MB in  objects
order  ( kB objects)
block_name_prefix: rbd_data.fb2074b0dc51
format:
features: layering, exclusive-lock, journaling
flags:
create_timestamp: Sun Mar  ::
journal: fb2074b0dc51
mirroring state: enabled
mirroring global id: cfa19d96-1c41-47e3-bae0-0b9acfdcb72d
mirroring primary: true

[root@ceph2 ceph]# rbd ls

3.5 从端验证

[root@ceph5 ceph]# rbd ls

同步成功

如果主挂掉，可以把从设为主，但是实验状态下，可以把主降级处理，然后把从设为主

3.6 降级处理

[root@ceph2 ceph]#  rbd mirror  image  demote rbd/cephrbd1
Image demoted to non-primary
[root@ceph2 ceph]# rbd info cephrbd1
rbd image 'cephrbd1':
    size  MB in  objects
    order  ( kB objects)
    block_name_prefix: rbd_data.fb2074b0dc51
    format:
    features: layering, exclusive-lock, journaling
    flags:
    create_timestamp: Sun Mar  ::
    journal: fb2074b0dc51
    mirroring state: enabled
    mirroring global id: cfa19d96-1c41-47e3-bae0-0b9acfdcb72d
    mirroring primary: false
[root@ceph5 ceph]#  rbd mirror image promote cephrbd1
Image promoted to primary
[root@ceph5 ceph]# rbd info cephrbd1
rbd image 'cephrbd1':
    size  MB in  objects
    order  ( kB objects)
    block_name_prefix: rbd_data.1043836c40e
    format:
    features: layering, exclusive-lock, journaling
    flags:
    create_timestamp: Mon Mar  ::
    journal: 1043836c40e
    mirroring state: enabled
    mirroring global id: cfa19d96-1c41-47e3-bae0-0b9acfdcb72d
    mirroring primary: true                      #成为主，客户端就可以挂载到这里

3.7 池的升降级

[root@ceph2 ceph]#  rbd mirror pool demote rbdmirror
Demoted  mirrored images
[root@ceph2 ceph]# rbd mirror pool demote rbdmirror
Demoted  mirrored images
[root@ceph2 ceph]# rbd info rbdmirror/ceph5-rbd
rbd image 'ceph5-rbd':
size  MB in  objects
order  ( kB objects)
block_name_prefix: rbd_data.fbe541a7c4c9
format:
features: exclusive-lock, journaling
flags:
create_timestamp: Mon Mar  ::
journal: fbe541a7c4c9
mirroring state: enabled
mirroring global id: c41dc1d8-43e5-44eb-ba3f-87dbb17f0d15
mirroring primary: false

[root@ceph5 ceph]# rbd mirror pool promote rbdmirror
Promoted  mirrored images
[root@ceph5 ceph]# rbd ls rbdmirror
ceph5-rbd
test
[root@ceph5 ceph]#  rbd info rbdmirror/ceph5-rbd
rbd image 'ceph5-rbd':
    size  MB in  objects
    order  ( kB objects)
    block_name_prefix: rbd_data.103e74b0dc51
    format:
    features: exclusive-lock, journaling
    flags:
    create_timestamp: Mon Mar  ::
    journal: 103e74b0dc51
    mirroring state: enabled
    mirroring global id: c41dc1d8-43e5-44eb-ba3f-87dbb17f0d15
    mirroring primary: true
[root@ceph5 ceph]#  rbd info rbdmirror/test
rbd image 'test':
    size  MB in  objects
    order  ( kB objects)
    block_name_prefix: rbd_data.1037140e0f76
    format:
    features: layering, exclusive-lock, object-map, fast-diff, deep-flatten, journaling
    flags:
    create_timestamp: Mon Mar  ::
    journal: 1037140e0f76
    mirroring state: enabled
    mirroring global id: 0f6993f2-8c34-4c49-a546-61fcb452ff40
    mirroring primary: true

3.8 摘掉mirror

[root@ceph2 ceph]# rbd mirror pool info --pool=rbdmirror
Mode: pool
Peers:
  UUID                                 NAME   CLIENT
  40e4fd11-5f9d-44b2-9f11-b57968be62cf backup client.admin
[root@ceph2 ceph]# rbd mirror pool info --pool=rbdmirror --cluster backup
Mode: pool
Peers:
  UUID                                 NAME CLIENT
  ccb2faba--454b-9c87-1a1bf292a76b ceph client.admin

[root@ceph5 ceph]#  rbd mirror pool peer remove --pool=rbdmirror 761872b4-03d8-4f0a-972e-dc61575a785f  --cluster backup

[root@ceph5 ceph]# rbd mirror image disable rbd/cephrbd1 --force

[root@ceph5 ceph]# rbd info cephrbd1

rbd image 'cephrbd1':
    size  MB in  objects
    order  ( kB objects)
    block_name_prefix: rbd_data.1043836c40e
    format:
    features: layering, exclusive-lock, journaling
    flags:
    create_timestamp: Mon Mar  ::
    journal: 1043836c40e
    mirroring state: disabled  #已经关闭了镜像的mirror

Ceph的集群间相互访问备份配置完成！！！

---

博主声明：本文的内容来源主要来自誉天教育晏威老师，由本人实验完成操作验证，需要的博友请联系誉天教育（http://www.yutianedu.com/），获得官方同意或者晏老师（https://www.cnblogs.com/breezey/）本人同意即可转载，谢谢！