My current project involves putting together a Eucalyptus mini-cloud so that we can evaluate it. I don't have access to hardware that supports KVM, so I have been trying to get Xen working with my Ubuntu installations.
I have no prior experience with GNU/Linux in general (we are making the transition from Windows to support this, so really no one on site has this background either). For a neophyte everything is pretty opaque.
Nonetheless I have now spent about 1 full week reading documentation and forum postings and attempting to get a Xen capable Dom0 kernel up and going and have been trying to start VMs with xm create (following a template I found for the config file on the Eucalyptus users site). I've been trying to use the pv-ops kernel from
the xen/master branch, following instructions I found here:http://bderzhavets.wordpress.com/2009/06/25/setup-xen-3-4-1-dom0-on-top-of-ubuntu-9-04-server-via-marc-a-dahlhauss-udev-patch/
I successfully compiled it and Xen 3.4 itself. I set up a new grub menu.lst entry:
title Xen 3.4 / Ubuntu 9.04, kernel 184.108.40.206
uuid b37xxxxx etc
module /vmlinuz-220.127.116.11 root=/dev/mapper/ubuntu--node--2-root ro console=tty0 max_loop=200
But when I boot from this it behaves very strangely, e.g. the username / password screen crashes and/or I have to enter my username / password several times. Eventually I can get into Ubuntu (so far),
but then various other things don't seem to be functioning (X11 randomly crashes, for instance). I have an Ubuntu VM image that I obtained from the Eucalyptus web site, specifically for Xen (it is based on the 18.104.22.168 kernel). I created a test_dom config file from a template I found on the Eucalyptus users forum.
So forging ahead, I then execute: xm create test_dom
It then replies:
Started domain ExampleDomain (id=1)
However, xm list then indicates only Domain-0 and /var/log/xen/xend.log indicates:
ERROR (XendDomainInfo:2012) VM ExampleDomain restarting too fast (Elapsed time: 4.028942 seconds). Refusing to restart to avoid loops.
The Eucalyptus forum also indicates that it should be possible to start this VM with virsh. But when I execute virsh version:
Connecting to uri: qemu:///system
Compiled against library: libvir 0.6.1
Using library:: libvir 0.6.1
Using API: QEMU
error: failed to get the hypervisor version
error: Unknown failure
Attempting to start with virsh create it fails immediately with an error indicating basically that xen isn't known.
I've tried a few other supposedly Dom0 compatible kernels I found from various places, but in each case Xen simply doesn't seem to be correctly set up some how and Ubuntu is flaky in various ways.
I thought I would now try to go through this procedure one more time with patches from Andrew Lyon and rebuild the 22.214.171.124 kernel (assuming I can figure out the makeconfig parameters). But as none of the attempts so far have been successful, I'm not optimistic.
Is it the case that I need to apply Xen patches to the kernel and then apply some sort of Ubuntu patches? Clearly Ubuntu seems unhappy with the vanilla kernels that I have built.
Most likely I have simply got myself into something I shouldn't have, and there is
some much easier way somehow to make this all work, I'm just missing it.
What do people recommend for someone like me just trying to get something going?