Getting Started with Nova

This code base is continually changing, so dependencies also change. If you encounter any problems, see the ../community page. The contrib/nova.sh script should be kept up to date, and may be a good resource to review when debugging.

The purpose of this document is to get a system installed that you can use to test your setup assumptions. Working from this base installtion you can tweak configurations and work with different flags to monitor interaction with your hardware, network, and other factors that will allow you to determine suitability for your deployment. After following this setup method, you should be able to experiment with different managers, drivers, and flags to get the best performance.

Dependencies

Related servers we rely on

RabbitMQ: messaging queue, used for all communication between components

Optional servers

OpenLDAP: By default, the auth server uses the RDBMS-backed datastore by setting FLAGS.auth_driver to nova.auth.dbdriver.DbDriver. But OpenLDAP (or LDAP) could be configured by specifying nova.auth.ldapdriver.LdapDriver. There is a script in the sources (nova/auth/slap.sh) to install a very basic openldap server on ubuntu.
ReDIS: There is a fake ldap auth driver nova.auth.ldapdriver.FakeLdapDriver that backends to redis. This was created for testing ldap implementation on systems that don't have an easy means to install ldap.
MySQL: Either MySQL or another database supported by sqlalchemy needs to be avilable. Currently, only sqlite3 an mysql have been tested.

Python libraries that we use (from pip-requires):

../../../tools/pip-requires

Other libraries:

XenAPI: Needed only for Xen Cloud Platform or XenServer support. Available from http://wiki.xensource.com/xenwiki/XCP_SDK or http://community.citrix.com/cdn/xs/sdks.

External unix tools that are required:

iptables
ebtables
gawk
curl
kvm
libvirt
dnsmasq
vlan
open-iscsi and iscsitarget (if you use iscsi volumes)
aoetools and vblade-persist (if you use aoe-volumes)

Nova uses cutting-edge versions of many packages. There are ubuntu packages in the nova-core ppa. You can use add this ppa to your sources list on an ubuntu machine with the following commands:

sudo apt-get install -y python-software-properties
sudo add-apt-repository ppa:nova-core/ppa

Installation

You can install from packages for your particular Linux distribution if they are available. Otherwise you can install from source by checking out the source files from the Nova Source Code Repository and running:

python setup.py install

Configuration

Configuring the host system

As you read through the Administration Guide you will notice configuration hints inline with documentation on the subsystem you are configuring. Presented in this "Getting Started with Nova" document, we only provide what you need to get started as quickly as possible. For a more detailed description of system configuration, start reading through multi.node.install.

Create a volume group (you can use an actual disk for the volume group as well):

# This creates a 1GB file to create volumes out of
dd if=/dev/zero of=MY_FILE_PATH bs=100M count=10
losetup --show -f MY_FILE_PATH
# replace /dev/loop0 below with whatever losetup returns
# nova-volumes is the default for the --volume_group flag
vgcreate nova-volumes /dev/loop0

Configuring Nova

Configuration of the entire system is performed through python-gflags. The best way to track configuration is through the use of a flagfile.

A flagfile is specified with the --flagfile=FILEPATH argument to the binary when you launch it. Flagfiles for nova are typically stored in /etc/nova/nova.conf, and flags specific to a certain program are stored in /etc/nova/nova-COMMAND.conf. Each configuration file can include another flagfile, so typically a file like nova-manage.conf would have as its first line --flagfile=/etc/nova/nova.conf to load the common flags before specifying overrides or additional options.

A sample configuration to test the system follows:

--verbose
--nodaemon
--auth_driver=nova.auth.dbdriver.DbDriver

Running

There are many parts to the nova system, each with a specific function. They are built to be highly-available, so there are may configurations they can be run in (ie: on many machines, many listeners per machine, etc). This part of the guide only gets you started quickly, to learn about HA options, see multi.node.install.

Launch supporting services

rabbitmq
redis (optional)
mysql (optional)
openldap (optional)

Launch nova components, each should have --flagfile=/etc/nova/nova.conf

nova-api
nova-compute
nova-objectstore
nova-volume
nova-scheduler

5.8 KiB Raw Blame History