One of the things any good sysadmin dogg wants to do is setup some virtual servers. Who wants to actually stand in front of a server to do that?! What if the machine is headless (no monitor attached)? Well, as long as we have SSH access to the machine, we can do everything from the dogghouse!
The first thing you want is to start with a nice server with a good OS installed. I’m starting with the CentOS 6 operating system, and a newer Dell Poweredge server. Obviously the more powerful the hardware, the better, but you will want something that supports full hardware virtualization, which is important for security, stability, and speed, among other things. Check in your BIOS for something like “Enable Virtualization.” Check with your hardware manufacturer for details.
After you have your initial server set up, you’ll need to add some packages for virtualization support. We’ll use the yum groupinstall command to get our necessary package groups, then install the bridge software separately, make sure everything is up to date, and finally restart the server so that we have the latest kernel running.
bash# yum groupinstall "Virtualization" "Virtualization Client" bash# yum groupinstall "Virtualization Platform" "Virtualization Tools" bash# yum install bridge-utils bash# yum update bash# reboot
When you reconnect to your host machine, make sure to forward some ports so you can tunnel to the VM’s console display! With putty, you can use a command like this:
putty.exe -L 5900:127.0.0.1:5900 -L 5901:127.0.0.1:5901 user@server
This will forward two ports for us, 5900 and 5901 to the localhost (the machine we’re connecting to) that we will use for VNC connections. Our first VM will use port 5900, our second VM will use port 5901, and so on. You can easily add more port forwards to your SSH connection to enable connecting to additional VM’s.
The first thing we need to do are set up some network bridges. There are a lot of different ways to configure network bridges, but we’re going to keep this one simple. Let’s say our machine has two physical network interfaces, eth0 and eth1. We’re going to assume that both interfaces are connected to the same network (so we can’t bridge them together), that we will use eth0 to connect to our physical machine (aka hypervisor, dom0, host, or manager), and that eth1 has no IP address. So, just create a separate bridge for each physical interface, br0 and br1.
Start by configuring eth0 (/etc/sysconfig/network-scripts/ifcfg-eth0):
DEVICE="eth0" HWADDR="12:34:56:78:90:ab" NM_CONTROLLED="no" ONBOOT="yes" IPADDR=10.0.0.100 GATEWAY=10.0.0.1 TYPE=Ethernet BOOTPROTO=none NETMASK=255.255.255.0 USERCTL=no PEERDNS=yes IPV6INIT=no SEARCH=mydomain.com DNS1=10.0.0.1 DNS2=10.0.0.2 DNS2=184.108.40.206 BRIDGE=br0
You could optionally configure eth0 to use DHCP, if desired. Be careful when remotely making changes to your active network connection; a misconfiguration could make you unable to connect!
Next, lets configure eth1 (/etc/sysconfig/network-scripts/ifcfg-eth1):
DEVICE="eth1" HWADDR="12:34:56:78:90:cd" NM_CONTROLLED="no" ONBOOT="yes" BRIDGE=br1
Restart the network connections to make sure everything is working so far:
bash# service network restart
Now that our physical interfaces are configured, lets configure some bridges! We can just edit the bridge scripts, so open the files and configure them already!
Bridge device br0 (/etc/sysconfig/network-scripts/ifcfg-br0):
DEVICE=br0 BOOTPROTO=none ONBOOT=yes TYPE=Bridge
Bridge device br1 (/etc/sysconfig/network-scripts/ifcfg-br1):
DEVICE=br1 BOOTPROTO=none ONBOOT=yes TYPE=Bridge
Restart the network services again, and we’re ready to go!
bash# service network restart
Check that our bridge devices exist with the following command:
bash# brctl show bridge name bridge id STP enabled interfaces br0 8000.842b2b03fa01 no eth0 br2 8000.0010186e94e8 no eth2
This shows us that br0 has interface eth0, and br1 has interface eth1.
Ok, so now that we have our host correctly configured, we need an OS to install! You can use a disk drive with the OS installation, install from something provided on the network, or go get a .iso of your favorite OS, and install from there. We’ll just use an .iso. Take a look at the following command:
bash# virt-install -n newbawx --vcpus=1 -f /home/vm/newbawx -s 60 -r 2048 --nonsparse -w bridge:br1 --vnc --accelerate -c /tmp/Fedora-15-x86_64-DVD.iso --os-type=linux --os-variant=fedora13 --noautoconsole Starting install... Creating storage file newbawx | 60 GB 04:43 Creating domain... | 0 B 00:00 Domain installation still in progress. You can reconnect to the console to complete the installation process.
First we use the virt-install command to create a new virutal machine named newbawx. We give it a hard drive size of 60GB (-s 60), 2GB of RAM (-r 2048), tell it to use bridge 1 (-w bridge:br1), and have it use a Fedora 15 .iso as a virtual CD drive to install from (-c /tmp/Fedora-15-x86_64-DVD.iso). Don’t forget the --vnc option to enable VNC access to the guest console. Notice that the os-variant is fedora13; that is because it is the highest os-variant parameter currently supported for Fedora!
After the virt-install command is run, we can optionally set the VM to automatically start with the host server:
bash# virsh autostart newbawx Domain newbawx marked as autostarted
Next, fire up your favorite VNC client and connect to 127.0.0.1:5000. You will be greeted with the installation walkthrough for your guest OS!
Also, if you wanted to delete and remove a virtual machine named newbawx:
bash# virsh destroy newbawx bash# virsh undefine newbawx bash# rm -rf /home/vm/newbawx
XML files containing your virtual machine settings are located in /etc/libvirt/qemu/.