Newer
Older
### Installation
## ssh keys exchange
Choose an admin node to use for the installation process.
Distribute the ssh key to all the hosts in the cluster.
## Install the deploy script
Add the ceph repository
```
yum install ceph-deploy
yum install ntp ntpdate ntp-doc
```
## Purge cluster
```
ceph-deploy purge qn-cnfslhc ds-001 ds-002 ds-303 ds-304 ds-507
ceph-deploy purgedata qn-cnfslhc ds-001 ds-002 ds-303 ds-304 ds-507
ceph-deploy forgetkeys
```
Create first monitor nodes:
```
ceph-deploy new qn-cnfslhc
```
This will create the following files:
```
ceph.conf
ceph.mon.keyring
```
Add the public network to the configuration file:
```
public_network = 131.154.128.0/22
```
Install the nodes:
```
ceph-deploy install node1 node2 node3
```
Deploy the initial monitoring node:
```
ceph-deploy mon create-initial
```
and the admin keys to the nodes of your cluster
```
```
Then deploy the manager node
```
ceph-deploy -v mgr create qn-cnfslhc
```
If you have a dirty installation you may receive errors like:
```
[qn-cnfslhc][ERROR ] [errno 1] error connecting to the cluster
[qn-cnfslhc][ERROR ] exit code from command was: 1
[ceph_deploy.mgr][ERROR ] could not create mgr
[ceph_deploy][ERROR ] GenericError: Failed to create 1 MGRs
```
This means that you must remove the old keys from `/var/lib/ceph`
```
rm -rf /var/lib/ceph/bootstrap-mgr/
```
### Enable dashboard
```
yum install ceph-mgr-dashboard # for nautilus
ceph mgr module enable dashboard
ceph config set mgr mgr/dashboard/$name/server_addr 131.154.130.69
ceph config set mgr mgr/dashboard/$name/server_port 5000
ceph dashboard set-login-credentials admin <password>
The only way to do it on an already initialized monitor map is to manipulate the monitor map. Starting with a single monitor ` qn-cnfslhc` add the other monitors
```
monmaptool --create --add mon.ds-001 131.154.128.145:6789 --add ds-303 131.154.130.77:6789 --add mon.ds-507 131.154.128.131:6789 --add mon.cs-001 131.154.130.166:6789 --clobber monmap
```
```
```