OpenStack Cinder Deploy

README.md

Installing OpenStack Cinder in 10 Minutes

The Cinder project provides volume storage management for OpenStack deployments. Cinder volumes appear as block storage devices to instances started in an OpenStack environment. Cinder is also used by OpenStack for storage of instance snapshots and can be used to boot cloned images of existing running instances.

You can familiarize yourself with the Cinder install process by watching the video below.

Before we drop into the guide, I'd like to thank Dell Computers for donating the test hardware, and the awesome folks at Rackspace for writing and supporting the Chef scripts which are used for the bulk of the setup process.

Prerequisites for Install

Before installing Cinder, you should ensure you've installed your OpenStack cluster using the Installing OpenStack in 10 Minutes guide. The primary requirement is you have a local instance of the Vagrant Chef server running which will be used in this guide to deploy Cinder to your nodes.

Cinder does storage management, so you'll also need a storage device configured to store data. Typically this is done with hard drives or SSDs. This guide currently uses the method of creating a 1TB loopback device attached to a file stored in the root partition on each node in your cluster. If your root partition doesn't have 1TB of storage available on it, you can either shrink the number to accomidate or install and provision a physical volume on another storage device.

Build the Loopback File and Persist It

Start by checking on the amount of storage you have available on your node. This guide assume the loopback file will be created in the root partition. Become root and then execute the following command:

df -ah /

Take a gander at the result:

root@flop:/home/kord# df -ah /
Filesystem      Size  Used Avail Use% Mounted on
/dev/sda2       2.8T  2.5G  2.6T   1% /

We're good to create a 1TB file on that partition. Now execute the command to create the file we'll loop:

dd if=/dev/zero of=cinder-volumes bs=1 count=0 seek=1024G

Results come back instantly:

root@flop:/home/kord# dd if=/dev/zero of=/cinder-volumes bs=1 count=0 seek=1024G
0+0 records in
0+0 records out
0 bytes (0 B) copied, 1.035e-05 s, 0.0 kB/s

Now loop up the file to the loopback device (we'll use loop2 here):

losetup /dev/loop2 /cinder-volumes

Finally, create a rc file to mount the loopback file after a reboot:

echo "losetup /dev/loop2 /cinder-volumes; exit 0;" > /etc/init.d/cinder-setup-backing-file
chmod 755 /etc/init.d/cinder-setup-backing-file
ln -s /etc/init.d/cinder-setup-backing-file /etc/rc2.d/S10cinder-setup-backing-file

Physical & Volume Group Creation

The next step is to initialize a LVM volume named cinder-volumes. That name is used by default for all Cinder installs. It can be changed in the configuration files in /etc/cinder/ after you install it, but we'll assume here we're leaving it as the default. Run the commands in one cut and paste frenzy:

sudo pvcreate /dev/loop2
sudo vgcreate cinder-volumes /dev/loop2

Revel in the vast awesomeness of your successful paste job:

root@flop:/home/kord# sudo pvcreate /dev/loop2
Physical volume "/dev/loop2" successfully created
root@flop:/home/kord# sudo vgcreate cinder-volumes /dev/loop2
Volume group "cinder-volumes" successfully created

Double check your work with pvdisplay and vgdisplay:

root@flop:/home/kord# pvdisplay; vgdisplay
  --- Physical volume ---
  PV Name               /dev/loop2
  VG Name               cinder-volumes
  PV Size               1.00 TiB / not usable 4.00 MiB
  Allocatable           yes
  PE Size               4.00 MiB
  Total PE              262143
  Free PE               262143
  Allocated PE          0
  PV UUID               XnvA5r-wBw0-WIhd-9FHd-n4lT-71em-uYCLJt

        --- Volume group ---
  VG Name               cinder-volumes
  System ID
  Format                lvm2
  Metadata Areas        1
  Metadata Sequence No  1
  VG Access             read/write
  VG Status             resizable
  MAX LV                0
  Cur LV                0
  Open LV               0
  Max PV                0
  Cur PV                1
  Act PV                1
  VG Size               1024.00 GiB
  PE Size               4.00 MiB
  Total PE              262143
  Alloc PE / Size       0 / 0
  Free  PE / Size       262143 / 1024.00 GiB
  VG UUID               vabQ1q-mxZ2-ofrn-6640-rriR-jln1-WHm4jm

Install Cinder via Vagrant Chef

Move on over to the computer that was used to provision the nodes in your install. If you've halted the Vagrant server, restart it first and then ssh into it and become root:

Beast:texasholdem kord$ cd bluechipstack/
Beast:bluechipstack kord$ vagrant up
Bringing machine 'default' up with 'virtualbox' provider...
...
[default] -- /vagrant
Beast:bluechipstack kord$ vagrant ssh
Welcome to Ubuntu 12.04 LTS (GNU/Linux 3.2.0-23-generic x86_64)
...
Welcome to your Vagrant-built virtual machine.
Last login: Wed Sep  4 18:54:12 2013 from 10.0.2.2
vagrant@chef-server:~$ sudo su
root@chef-server:/home/vagrant#

Now test Chef is still aware of the servers:

knife node list

Time to install Cinder on the nodes. Type the following to add the Cinder storage role to the nodes (making sure to change the node name(s) in the process):

knife node run_list add flop 'role[cinder-volume]'
knife node run_list add turn 'role[cinder-volume]'

Volume Clear

Thank you for these informations.
Q: I don't really know OpenStack. Is it possible to deploy a Cinder node without a complete OpenStack infrastructure ? The goal is to use Cinder in oVirt, which is the only way (at this time) to use a Ceph RBD backend..

Thanks !