[rhos-list] Nova-network v.s. Quantum in Openstack preview

[Perry Myers]

On 02/18/2013 08:34 AM, Eduardo Habkost wrote:
> 
> Thanks for the information. There's no need to test exhaustively each
> CPU flag, I just wanted to find a minimal test case where the bug could
> be reproduced by simply enabling one feature bit.
> 
> Note that the bug may be completely unrelated to support of feature
> flags on Nehalem or your host CPU. The problem is happening when KVM is
> already running guest code, not during CPU feature check/initialization.
> 
> I don't know yet what can be causing the issue, but I don't discard the
> possibility of a CPU defect. In either case, this bug is not
> reproducible on any of the machines where we have run our tests, so we
> need more details about your host CPU: the contents of /proc/cpuinfo and
> (in case you can install x86info) the output of 'x86info -v -a'

If it's a cpu defect...

Shixiong, can you at least reproduce this on multiple physical machines
or is it only reproducible on a single server thus far?

> Also, I still need to know the version of the kernel, seabios, vgabios,
> and gpxe packages.