KVM FYI

[Sean Estabrooks]

On Sun, 20 Jun 2004 15:56:26 -0400 (EDT)
Tom Diehl <tdiehl at rogueind.com> wrote:


> 
> I do not know where you work, but wheel mice are the norm everywhere I work.
> It worked before so as far as I am concerned some change in this latest
> kernel broke it. I have had this kvm for 3 years without a problem. It worked
> with RHL 7.3/8.0/9, FC1, FC2 until now. I just looked in bugzilla and saw the
> same suggestion. The problem is that for me not having my wheel work is
> just as bad as not having the mouse work. I am used to having it so if it
> does not work, I will waste time scrolling the wheel.

Sure, wheel mice are the norm.   It's just that KVM's aren't typically used
with machines that have a person in front of them all day long.  More often
they're used in a server room and only used for occasional access.   Given
the great remote access capabilities of Linux using a KVM is often not
needed for day to day use.

> I realize some of the software in the kvm's are crap but they are not going
> to go away. At the very least Red Hat should put some of them on the HCL and
> say these are supported. At least then I would have an idea what to buy. 
> The way things are now it is a crap shoot and even if you get one that works
> perfectly today, tomorrow some change to the kernel might break it.
> 
> If the one I have had never worked I would have sent it back. For obvious
> reasons that is not an option now.
> 
> One other thing I have noticed, with the old kernel switching between the 
> different machines with the kvm was instantaneous. With the new kernel
> there is about a 1 second delay. Something changed and it is not for the
> better.

So, you seem to understand this is a kernel issue and will have to be solved
by the kernel community.  Not sure  then why you went on and on about the
attitude at RedHat.   The 2.6 kernel has the mouse handling pulled into the 
kernel where previously it just passed raw mouse data to user mode applications.   

This has caused numerous issues, perhaps more than it has solved.   Recently 
there's been talk of adding a "raw" mode to the kernel to allow the kernel processing 
to be bypassed.   Perhaps this will happen, but the real answer is to continue
to improve the kernels processing of less-than-perfect mice implementations.
Getting upset with RedHat and accusing them of a bad attitude isn't likely to help.

Cheers,
Sean