Nova Installation Script
This script will install and run nova on ubuntu.
Unless you want to spend a lot of time fiddling with permissions and sudoers, you should probably run nova as root.
If you are concerned about security, nova runs just fine inside a virtual machine.
Use the script to install and run the current trunk. You can also specify a specific branch by putting lp:~someone/nova/some-branch after the branch command
./nova.sh branch ./nova.sh install ./nova.sh run
The run command will drop you into a screen session with all of the workers running in different windows You can use eucatools to run commands against the cloud.
euca-add-keypair test > test.pem euca-run-instances -k test -t m1.tiny ami-tiny euca-describe-instances
To see output from the various workers, switch screen windows
will give you a list of running windows.
When the instance is running, you should be able to ssh to it.
chmod 600 test.pem ssh -i test.pem firstname.lastname@example.org
When you exit screen
nova will terminate. It may take a while for nova to finish cleaning up. If you exit the process before it is done because there were some problems in your build, you may have to clean up the nova processes manually. If you had any instances running, you can attempt to kill them through the api:
Then you can destroy the screen:
If things get particularly messed up, you might need to do some more intense cleanup. Be careful, the following command will manually destroy all runnning virsh instances and attempt to delete all vlans and bridges.
You can edit files in the install directory or do a bzr pull to pick up new versions. You only need to do
to run nova after the first install. The database should be cleaned up on each run.
A sample image should be downloaded by the script, but if necessary you can download it by hand:
untar the file to create a usable images directory
tar -zxf /path/to/images.tgz
If you want to be able to contact the metadata server and route to the outside world from instances, you will need to make sure $HOST_IP is set properly. The script attemps to grab it from ifconfig, but if you have multiple adapters set up, it may fail. Fix it with export HOST_IP="":
If you want to use volumes, you will need to create a volume group called nova-volumes before starting the workers. If you have a spare hard disk attached to your dev box at /dev/sdb (virtual or otherwise), you can create a logical volume with one command:
sudo vgcreate nova-volumes /dev/sdb
If you don't have a spare drive, you can create a flat file to back the volume group. Here is an example for a 100G volume store:
dd if=/dev/zero of=volumes bs=1G skip=99 count=1 mkfs.ext4 -F volumes DEV=`sudo losetup -f --show volumes` sudo vgcreate nova-volumes $DEV
You can make nova use mysql instead of sqlite with USE_MYSQL, it will attempt to install mysql with the specified root password and create a database called nova.
If you are running nova on bare metal that supports hardware virtualization, you should probably edit the libvirt line near the top
If you are running in a virtual machine and software emulation is too slow for you, you can use user mode linux.
If you have any issues, there is usually someone in #openstack on irc.freenode.net that can help you out.