[one-users] Getting OpenNebula working on Fedora 17/CentOS 5

[Simon J Mudd]

Hi I'm trying to get OpenNebula working on Fedora 17 and CentOS 5
and have a few issues. It looks likes all the work for OpenNebula has
been done on Debian or Ubuntu. I've built OpenNebula successfully on Fedora
17 (from source) and now trying to resolve issues which prevent
the kvms starting up.

One issue I see is this:

Sat Jan  5 11:23:47 2013 [DiM][I]: New VM state is ACTIVE.
Sat Jan  5 11:23:47 2013 [LCM][I]: New VM state is PROLOG.
Sat Jan  5 11:23:47 2013 [VM][I]: Virtual Machine has no context
Sat Jan  5 11:23:47 2013 [LCM][I]: New VM state is BOOT
Sat Jan  5 11:23:47 2013 [VMM][I]: Generating deployment file: /var/lib/one/11/deployment.0
Sat Jan  5 11:23:48 2013 [VMM][I]: ExitCode: 0
Sat Jan  5 11:23:48 2013 [VMM][I]: Successfully execute network driver operation: pre.
Sat Jan  5 11:23:48 2013 [VMM][I]: Command execution fail: cat << EOT | /var/tmp/one/vmm/kvm/deploy /var/lib/one//datastores/0/11/deployment.0 mad06 11 mad06
Sat Jan  5 11:23:48 2013 [VMM][I]: error: Failed to create domain from /var/lib/one//datastores/0/11/deployment.0    [*] 
Sat Jan  5 11:23:48 2013 [VMM][I]: error: Cannot find QEMU binary /usr/bin/kvm: No such file or directory
Sat Jan  5 11:23:48 2013 [VMM][E]: Could not create domain from /var/lib/one//datastores/0/11/deployment.0
Sat Jan  5 11:23:48 2013 [VMM][I]: ExitCode: 255
Sat Jan  5 11:23:48 2013 [VMM][I]: Failed to execute virtualization driver operation: deploy.
Sat Jan  5 11:23:48 2013 [VMM][E]: Error deploying virtual machine: Could not create domain from /var/lib/one//datastores/0/11/deployment.0
Sat Jan  5 11:23:48 2013 [DiM][I]: New VM state is FAILED

This is on the CentOS 5 server. checking and indeed there is no /usr/bin/kvm binary.
There is the following:

[sjmudd at mad06 ~]$ ls -la /usr/bin/qemu
-rwxr-xr-x 1 root root 1797488 Sep  7  2009 /usr/bin/qemu
[sjmudd at mad06 ~]$ rpm -qi qemu
package qemu is not installed
[sjmudd at mad06 ~]$ rpm -qf /usr/bin/qemu
qemu-system-x86-0.10.5-1.el5.2
[sjmudd at mad06 ~]$ 

So I'd expect that configuring opennebula to use the qemu binary should be sufficient.
What's the best way to do this? (note the CentOS 5 PC is just running the node software,
the install was done on Fedora 17)

Also I had a similar issue on Fedora 17 and ended up doing something similar, namely
create a symlink to the expected kvm binary:

[sjmudd at fedora17 ~]$ ls -la /usr/bin/kvm
lrwxrwxrwx 1 root root 8 Dec  2 11:59 /usr/bin/kvm -> qemu-kvm
[sjmudd at fedora17 ~]$ rpm -qf /usr/bin/qemu-kvm
qemu-system-x86-1.0.1-2.fc17.x86_64
[sjmudd at fedora17 ~]$ 

This strikes me as rather broken, and I'd expect a way for OpenNebula to figure out
the location of the kvm binary by either looking for it, or based on the Linux distribution
that it's running on.

Also see the [*] above. Any ideas why I'm getting the double // in the path? Most certainly
a configuration error but I am not sure where to look.

Feedback would be welcome.

Thanks,

Simon