When I first started working on Space I knew I would need to use something like libvirt to make all of the calls to the hypervisor its self. I had heard of libvirt several times during my tenure in the IT industry so I figured it would be realtively easy to get things running fairly quickly but when I started reading the libvirt documentation I was very surprised at how poor it was.
I'm not hoping to rewrite all of their documentation, but I would like to give a few examples for doing basic things using the libvirt API in Python.
Connecting to the Hypervisor
You'll need to establish a connection with your hypervisor before you can interact with it. A simple function to build that connection looks like this:
import libvirt def connect(): connection = libvirt.open("SYSTEM_TYPE:///system") return conn
You'll end up returning a connection object that you can then use to do other things.
SYSTEM_TYPE will be something like
Getting a List of Domains (Virtual Machines)
There are actually two calls required to do this,
listDefinedDOmains() which returns a list of all defined domains (running and not running), and
listDomainsID() which returns a list of only running domains.
import libvirt connection = libvirt.open("SYSTEM_TYPE:///system") # This lists all domains on the host connection.listDefinedDomains() # Returns ['vm1','vm2','vm3'] # This lists running domains on the host connection.listDomainsID() # Returns  # These are the ID's as used by the hypervisor - you'll need to store this somewhere to tie this to a virtual machine as it will change each time the virtual machine is started.
Find a Domain
You can find a domain two different ways, you can use the ID that was mentioned above (this changes, so it's tough to use this unless you maintain this information somewhere) or you can use the name of the virtual machine.
I generally use the name because I set it and it stays the same regardless of what happens to the virtual machine. The name is what you'll see if you run
virsh list --all:
virsh list --all Id Name State ---------------------------------------------------- 2 vm54d42cd2a5ee223e6be95bc2 running
So you can do that using the following functions:
# Assume the connection function is included here virtual_machine = connection.lookupByName("vm54d42cd2a5ee223e6be95bc2") # Returns the object representing this virtual machine virtual_machine = connection.lookupByID(2) # Returns the same object, but uses the ID as shown in the virsh output. Probably not the one you want to use, but it is there if you need it.
Once you have the domain object you can start/stop/redfine it fairly easily. Let's look at starting it:
virtual_machine.create() # Will create (start) this virtual machine if it is defined
virtual_machine.destroy() # Will destroy (stop) this virtual machine if it is defined
Now lets make a brand-spankin'-new domain. This wil take a bit of work, so lets dig in.
First you'll need to make a domain configuration file - this is an XML file that defines a bunch of things about the domain. I've provided an example below:
<domain type="kvm"> <name>vm11</name> <cpu> <topology cores="4" sockets="1" threads="4" /> </cpu> <uuid>90e7bca4-97b2-11e4-86bf-001e682ee78a</uuid> <memory unit="MB">512</memory> <currentMemory unit="MB">512</currentMemory> <vcpu placement="static">2</vcpu> <os> <type>hvm</type> <boot dev="hd" /> </os> <features> <acpi /> <apic /> <pae /> </features> <clock offset="utc" /> <on_poweroff>destroy</on_poweroff> <on_reboot>restart</on_reboot> <on_crash>restart</on_crash> <devices> <emulator>/usr/libexec/qemu-kvm</emulator> <disk device="disk" type="file"> <driver cache="none" name="qemu" type="raw" /> <source file="/var/disks/vm11.img" /> <target dev="hda" /> <address bus="0" controller="0" target="0" type="drive" unit="0" /> </disk> <disk device="cdrom" type="file"> <source file="/var/images/ubuntu-14.04.1-server-amd64.iso" /> <driver name="qemu" type="raw" /> <target bus="ide" dev="hdc" /> <readyonly /> <address bus="1" controller="0" target="0" type="drive" unit="0" /> </disk> <interface type="network"> <source network="default" /> </interface> <graphics port="-1" type="vnc" /> </devices> </domain>
I won't dig into how this is made (perhaps I'll do that in another post), but long story short you'll need to make this file and save it somewhere on your filesystem. You can take a look here to see how I did this in Space.
Next, we'll need to use this configuration to define a new virtual machine. Defining a virtual machine essentially means you are creating it, once it is defined you can undefine it (delete it), start it, and shut it down. To define the domain, you can use
# First you'll want to get your XML file into memory xml = "" with open(path_to_xml_config, "r") as file: xml = file.read() connection.defineXML(xml) # If you don't get an error, the domain has been created, woot!
Deleting a domain is significantly easier than making one. You'll just need to use the
virtual_machine = connection.lookupByName("your_domain_name") virtual_machine.undefine() # Bye bye virtual machine, its dead!
I'm sure some of you have put two and two together here and have pieced together how you would update a domain from that - if not, I'll detail that now.
You'll first need to overwrite the old configuration file (or make a new one). The new configuration file should include the new options you want to update the domain to. Next, you'll need to undefine the current domain using the method described above.
Once the domain has been undefined, simply define it again as described in the create a domain section. Start it up and you'll have a domain with updated settings.
I hope this saves somebody time in the future because it took me a while to figure out how to get all this stuff working my first time around. If you have any questions about anything I didn't cover, please feel free to comment or hit me up on Twitter @joseph_pettit.