Ceph cluster installation

Cluster size: 1 monitor, 2 osd machines

Monitor- | _ 172.16.x.x (ip2,10.1.x.x/24) ceph01

| |-osd1 _ 172.16.x.x (ip2,10.1.x.x/24) ceph02 |

| |-osd2 _ 172.16.x.x (ip2,10.1.x.x/24) ceph03 |

1. Preparatory work

All nodes update the system and install ceph-deploy

Sudo yum update & & sudo yum install ceph-deploy

All nodes install ntp,ssh

Sudo yum install ntp ntpdate ntp-doc

Sudo yum install openssh-server

The ceph management machine must log in to all permissions as an ordinary user, so ceph users should have sudo permissions without passwords

1 > create a normal ceph user

Ansible all-m shell-a 'useradd ceph & & echo "ceph" | passwd-- stdin ceph'

2 > make sure the node has sudo permissions

Echo "ceph ALL = (root) NOPASSWD:ALL" | sudo tee / etc/sudoers.d/ceph

Sudo chmod 0440 / etc/sudoers.d/ceph

3 > generate the key of Ceph users and go to each machine to be secret-free.

Ssh-copy-id ceph@ceph01

Ssh-copy-id ceph@ceph02

Ssh-copy-id ceph@ceph03

4 > config file of ceph to ensure that config permission is 600 and ssh ceph01 ceph02 ceph03 is allowed.

/ home/ceph/.ssh & & touch config & & chmod 600. / config

5 > all firewalls are turned off

Ansible all-m shell-a "systemctl disabled firewalld & & systemctl stop firewalld & & setenforce 0"

6 > install yum-plugin on centos

Yum install yum-plugin-priorities

2. Ceph-cluster installation

1 > create a directory and enter the implementation directory my-cluster

Su ceph & & cd & & mkdir my-cluster & & cd my-cluster

2 > create a cluster, execute the following command to generate three files (ceph configuration file, monitor key, log file)

Ceph-deploy new ceph01

3 > modify the default number of copies\ Nic

Osd pool default size = 2

Public network = 172.16.x.0/24 provides access to the outside world

Cluster network = 10.1.x.0/24 internal heartbeat

4 > install ceph

Ceph-deploy install ceph01 ceph02 ceph03

5 > configure the initial monitor (s), collect all keys, and collect

Ceph-deploy mon create-initial

Ceph-deploy gatherkeys ceph01

6 > View node disks and initialize all node disks

Ceph-deploy disk list ceph02

Ceph-deploy disk zap cehp02:sdb

7. Activate osd

Ceph-disk activate-all

Or it can be executed on the deployment machine, and all osd nodes must be executed to ensure that they can join the cluster normally.

Ceph-deploy osd prepare ceph02:/dev/sdb:/dev/sdc

Ceph-deploy osd activate ceph02:/dev/sdb1:/dev/sdc1

8. Cluster status