[Users] NAT configuration

[Dan Kenigsberg]

On Wed, Feb 05, 2014 at 09:50:04AM +0000, Sven Kieske wrote:
> I can confirm that vdsm at ovirt does work.
> 
> However, I have the strong feeling that
> the password in /etc/pki/vdsm/keys/libvirt_password
> is static for all installations.
> 
> And gerrit proves me right:
> 
> http://gerrit.ovirt.org/gitweb?p=vdsm.git;a=blob;f=vdsm/libvirt_password;h=09e60bce9bc401bb8943154f7cb9cb08bd0f49da;hb=refs/heads/master
> 
> So what is the purpose of authentication when that information
> is public?
> 
> I created a BZ for this:
> 
> https://bugzilla.redhat.com/show_bug.cgi?id=1061639
> 
> PS: I hope, whoever coded this, feels a little bit ashamed
> and perhaps buys a good book on writing secure code and reads it..

I feel ashamed, but not due to the "security" issue here.

Vdsm uses a unix domain socket to connect to libvirtd. That socket is
owned by vdsm, so that only vdsm and root can use it. There is no
security reason to use a password at all.

I am ashamed for caving in and adding an obfuscation layer, designed
only to deter local administrators from messing with libvirt under the
feet of ovirt. This little hurdle does not deter from messing with qemu
directly, but I suppose that qemu's command line does a good job anyway.

Red Hat support folks repeatedly claim that this hurdle is more
effective than putting a release note warning of the dangers in direct
libvirt access.

Dan.