New Nebula

OpenNebula Configuration File

You can specify the configuration for your OpenNebula setup in the oned.conf file. This file is critical to your OpenNebula deployment. The folks at OpenNebula provide all possible configurations for existing drivers that ship with the standard OpenNebula distribution. You can comment and uncomment as required for your environment. If, for example, you are interested in using Xen, just uncomment the relevant section in the oned.conf file (Listing 2).

01 #IM_MAD = [
02 #    name       = "im_xen",
03 #    executable = "one_im_ssh",
04 #    arguments  = "im_xen/im_xen.conf",
05 #    default    = "im_xen/im_xen.conf" ]
06
07 #VM_MAD = [
08 #    name       = "vmm_xen",
09 #    executable = "one_vmm_xen",
10 #    default    = "vmm_xen/vmm_xen.conf",
11 #    type       = "xen" ]

The IM_MAD section relates to the information monitoring driver for the status monitoring. The VM_MAD section deals with the hypervisor driver.

Similarly, if you were using a global shared NFS storage between all your nodes, you would uncomment the relevant configuration variables. Listing 3 shows the portion you need to uncomment.

01 #---------------------------------------------------------------
02 # NFS Transfer Manager Driver sample configuration
03 #---------------------------------------------------------------
04 #  TM_MAD = [
05 #      name       = "tm_nfs",
06 #      executable = "one_tm",
07 #      arguments  = "tm_nfs/tm_nfs.conf",
08 #      default    = "tm_nfs/tm_nfs.conf" ]

After you have added the worker nodes, your private three-node cloud is ready. However, before you can start launching virtual machines, you need to define a virtual network and then create and launch a virtual machine.

Virtual Networks

You can define custom private networks in OpenNebula, and virtual machines can be associated with several of these virtual networks. Different networks can link a specific set of virtual machines together. You could also permit some machines access to the Internet by connecting them to a specific virtual network.

To define your own virtual network, you need to write a template file. Fortunately, the syntax of a virtual network template file is pretty simple. In OpenNebula, you can define a " fixed" virtual network or a " ranged" network. A fixed network specifies certain IP addresses against specific virtual machines identified by their MAC addresses. On the other hand, a ranged network is more like a DHCP configuration, where a base network address is specified and all addresses follow that same pattern. A sample fixed network configuration template is shown in Listing 4, and Listing 5 shows a sample ranged network configuration.

01 NAME = "Private Cloud"
02 TYPE = FIXED
03 BRIDGE = vbr0
04 LEASES = [IP=192.168.0.1, MAC=50:20:20:20:20:20]

01 NAME = "wanNetwork"
02 TYPE = RANGED
03 BRIDGE = vbr0
04 NETWORK_SIZE = C
05 NETWORK_ADDRESS = 192.168.0.0

In creating a virtual network, I will use the fixed network template file specified in Listing 4, save the contents of Listing 4 to a file called network.template, and enter the commands shown in Listing 6.

01 nebula-user@nebula-cloud-server:~$ onevnet create network.template
02 nebula-user@nebula-cloud-server:~$ onevnet list
03  NID NAME              TYPE BRIDGE
04    0 Private Cloud    Fixed   eth0

Creating a KVM Virtual Machine

The VM used to test the three-node cloud infrastructure is available for download [5]; however, you can create your own. To create your own KVM virtual machine, you would start by creating the disk file:

dd if=/dev/zero of= ubuntu_server.img bs=1M count=4096

Use the kvm command to start the installation from the Ubuntu ISO file:

kvm -m 512 -cdrom /isos/ubuntu.iso -boot d ubuntu_server.img

After you run this command, a QEMU window should pop up and launch the Ubuntu CD boot screen. Select Install Ubuntu and follow the instructions. After the virtual machine is created, the next step is to submit the VM to OpenNebula.

Before you can submit your VM to OpenNebula, you need to carry out a few tasks. To begin, you must copy the ubuntu_server.img created in the last section to the virtual machine repository location. This location is specified in the oned.conf file. The variable is called VM_DIR, and it is commented by default. You'll need to store all images in /opt/vmImages/ on the front end and update the configuration file accordingly, keeping in mind that you must restart the daemon after every configuration change.

After you've saved the image to the appropriate directory, you can write a VM description for OpenNebula. VM descriptions are mostly hypervisor specific; however, some general fields are required (Listing 7).

01   NAME   = kvmVM  #specify the name
02     CPU    = 1 # How many CPUs required?
03     MEMORY = 512 # RAM in MB
04     OS     = [
05        KERNAL   = "/boot/vmlinuz-2.6.32-24-generic", # Kernel to use
06        INITRD   = "/boot/initrd.img-2.6.32-24-generic", # initrd to use
07        ROOT    = "sda1", #Root partition
08        BOOT = hd #boot form harddisk
09     ]
10     DISK   = [
11        SOURCE   = "/opt/vmImages/ubuntu_server.img",  #location of source image
12        TARGET = "sda1", # mount as partition ]
13     DISK   = [
14        TYPE     = "swap", #swap drive
15        SIZE     = 1024, # size of swap drive
16        TARGET   = "sdb" #mount to partition
17     ]
18     NIC    = [ NETWORK = "Private Cloud" ] #connect to specified virtual network, several NIC can be specified, is you want to connect to several VNs

Save the VM description file to kvmVM.template, but don't forget to remove the comments, which have been inserted to explain the contents of the template and will produce parsing errors if left in the file.

To submit this virtual machine, use the openvm command. Just as the onehost command deals with host management, onevm deals with VM management

onevm [options] subcommand [parameters]

and uses the same set of command options (Figure 2). The subcommand arguments associated with openvm can be any of the following: create, deploy, shutdown, livemigrate, migrate, hold, release, stop, cancel, suspend, resume, delete, list, show, top, and history. See the OpenNebula documentation for more on these commands.

The create command allows you to submit a new VM to the OpenNebula cluster. This command requires the user to pass the VM template file. Once you submit a VM, you can watch its status with show. You will also know the ID assigned to the VM. To apply commands like shutdown, stop, cancel, suspend, and resume to a VM, just specify the VM ID.

OpenNebula allows a user to migrate a VM from one host in the cluster to another. With migrate, the VM is first stopped before it is redeployed on another machine, whereas livemigrate migrates the VM without stopping and redeploying it. To specify a migration, you need to specify both the host ID and the VM ID.

To list all VMs currently running, use vm list. To launch the virtual machine, type the commands in Listing 8, which shows the VM in the scheduling queue. The private cloud setup is complete.

01 nebula-user@nebula-cloud-server:~$ onevm create kvmVM.template
02 nebula-user@nebula-cloud-server:~$ onevm list
03   ID     NAME STAT CPU     MEM        HOSTNAME        TIME
04    0    kvmVM pend   0       0              00    00:00:07