- Vanilla installation of Xen v3.0.3 on
hypervisor.example.com. All defaults.
- Platform is CentOS 5. Paravirtualization is not supported on CentOS 6. It’s possible to make it work, but I think you should get a newer processor and run KVM if using CentOS 6 to save yourself the trouble.
Not meant to be complete.
|Type I Hypervisor||Runs directly on hardware. Virtual machines don’t know they’re virtualized.|
|Type II Hypervisor||Hypervisor (Xen) runs in OS (RHEL/CentOS). The virtual machines ‘‘know’’ they’re being run in a virtual environment|
|HVM (“Hardware Virtual Machine” or “Hardware-assisted Virtualization”)||Not entirely sure about this. Certain processor technology (e.g. Intel VT-x) allows “complete simulation of underlying hardware.” VMs don’t know they’re virtualized.|
||The hypervisor itself|
||A single virtual instance|
||Xen-provided tool to manage domU’s|
||A Red Hat-designed shell to manage VM’s. Differs from
||Management and provisioning tools based on
yum groupinstall Xen yum install python-virtinst qemu*
The first installs the Xen-enabled kernel, Xen daemon, virtualization libraries, etc. Make sure that (a) SELinux is disabled, and (b) that you reboot into the Xen kernel before doing anything else.
The First VM
- My VMs will be running CentOS 6. So I downloaded and loop-mounted the latest CentOS 6 ISO. I then offered the mount via HTTP for VM installation.
- I then created logical volumes for use as storage by the VMs. You can also format and use disk images.
Creating the VM
virt-manager is the easiest way to do things. You can do a
command-line install via
virt-install. Here’s a sample command that
creates a 64-bit VM called “devel1” running CentOS 6 with two virtual
CPUs and 1.2GB of RAM. Observe that I explicitly specify the MAC
virt-install \ --name=devel1 \ --arch=x86_64 \ --vcpus=2 --check-cpu \ --ram=1200 \ --disk path=/dev/xenspace/devel1 \ --mac=00:0C:29:1A:98:D5 \ --os-type=linux \ --os-variant=rhel6 \ --location=http://hypervisor.example.com/install/6/x86_64/ \ --debug \ --nographics
Once the VM is installed, it’s a good idea to save the kickstart files. Here’s a sample:
# Modified by Nikhil Anand install url --url http://hypervisor.example.com/install/6/x86_64/ lang en_US.UTF-8 keyboard us network --device eth0 --bootproto dhcp rootpw --iscrypted $1$9P2b0WZe$CSd.fBGCVjjUfzlZ6m5Rk1 firewall --enabled --port=22:tcp authconfig --enableshadow --enablemd5 selinux --enforcing timezone --utc America/Chicago bootloader --location=mbr --driveorder=xvda # The following is the partition information you requested # Note that any partitions you deleted are not expressed # here so unless you clear all partitions first, this is # not guaranteed to work clearpart --linux --drives=xvda part /boot --fstype ext3 --size=100 --ondisk=xvda part pv.6 --size=0 --grow --ondisk=xvda volgroup VolGroup00 --pesize=32768 pv.6 logvol / --fstype ext3 --name=LogVol00 --vgname=VolGroup00 --size=1024 --grow logvol swap --fstype swap --name=LogVol01 --vgname=VolGroup00 --size=528 --grow --maxsize=1056 %packages @base @core keyutils iscsi-initiator-utils trousers fipscheck device-mapper-multipath
If you ever wanted to reinstall the VM, you can now append a flag with the (HTTP downloadable) path to the kickstart file:
You can find if your processor supports HVM by issuing
egrep '^flags.*(vmx|svm)' /proc/cpuinfo
Xen offers the following:
It’s unusual (and crazy) to use all three on a given dom0 instance. The
default is bridged networking. The
brctl command is used to manage
In our case, the router hands out DHCP leases depending on MAC addresses. This is why I didn’t have to do anything other than specify the MAC address in a domU’s config:
vif = [ "mac=00:50:56:78:0a:1b,bridge=xenbr0,script=vif-bridge" ]
More exotic configurations are possible. You can, for example, specify
two virtual interfaces (
vif’s), with public and private IPs. In this
iptables commands become important, since you’ll
have to set up routes and masquerading.
/etc/xen/xend-config.sxp to set up these configs. For instance,
if you only had a routed config, you’d comment out every other
vif-script other than these:
#(network-script network-route) #(vif-script vif-route)
virt-install removes the
ramdisk lines from a domU’s
config file and adds this instead:
bootloader = "/usr/bin/pygrub"
PyGRUB itself will look for the first partition or LVM container that contain the kernel and init image.
I made an error of using the CentOS project-supplied kernel and ramdisk, which were good for an install, but useless when the domU was rebooted. They’re built specifically for installation :)
“Could not connect to localhost:8000”
You may see this when using
/etc/xen/xend-config.sxp and make sure these lines are uncommented:
(xend-http-server yes) (xend-port 8000) (xend-address localhost)
And restart the Xen daemon.
You’re supposed to be able to edit
/etc/sysconfig/xend, uncomment this
line and see logs in
Didn’t work for me.
“Guest name already in use”
virsh undefine <guestname>