A hardware KVM allows for the utilization of one keyboard/mouse with many independent computers. Generally the user is expected to turn a dial or flip a switch in order to transfer input from one machine to another. These instructions show how to set it up such that one can use a single keyboard/mouse combination seamlessly between computers, without the flipping of a switch, given a few key pieces of software and network availability.
This is specifically targeted towards users of a Linux workstation + a Windows laptop, where the laptop is pulling a dynamic IP from a wireless network. However, these instructions apply to many other similar setups with little to no modification. The targeted scenario is common enough in CS graduate lab/office environments (at the very least) that this document seems to be worth writing.
Step 1: VNC Server
VNC is kind of like X, but where state is preserved on the server. That is, clients connect, are allowed to view/modify the state of a workspace, but if the connection is somehow dropped (either purposefully or otherwise) the client can always reconnect and find the state unchanged due to the disruption. This is a useful paradigm to work in for those (like me) that rarely close applications, but tend to be physically mobile in the sense of working in many different areas with a laptop (ahhhh wireless).
So, with VNC you can run things like an emacs session, a browser with a number of tabs open, and a few multi-day experiments, all on a server (such as your Linux workstation). Then, as a client, you can connect to this workspace from elsewhere whenever you wish, for however long you want, and never again have to worry about needing to get all of everything re-setup each time you begin a new session.
All of this is only tangentially related to this How-to, as in this case we are not really using VNC in its "default" manner.
What you want to do is install a VNC server on your laptop and start it up. Be sure to include a decent password for authentication when you do so. There are multiple flavors of VNC implementations (it is an open spec), of which I have linked to two at the bottom of the page.
Step 2: x2vnc
Now, this part is pretty cool, and is the basis of our whole enterprise. Fredrik Hubinette, whom I have never met or interacted with in any way, has hacked together a neat VNC client with an important property. Namely, it is only one pixel wide.
Why is this important?
Well, it means that if you move your mouse to one side of the screen or the other (depending on where you have the client oriented), then as you come in contact with whichever screen edge, this client will grab your mouse and keyboard input and begin transmitting it to the server as any normal client should. However, while it is only one pixel wide, it is smart enough to realize the dimensions of the remote workspace, and only allows the mouse to escape from its control if you move it to the appropriate screen edge, server-side. The resultant effect is that if you have a laptop running next to a workstation, with the edges configured according to the physical placement of the devices, then you can watch your mouse pointer move from one screen to the other and back again, just like a dual display setup.
Grab x2vnc from my included links, install it. You should be able to connect to your laptop's server by typing in an IP address and your cleverly manufactured password. Doesn't work? Read the appropriate help doc (not on my pages) and figure it out.
Functionally this is as far as you need to go for the basic idea. The rest are candy items to make the whole thing especially sweet.
Step 3: Dynamic DNS
Now that we've setup the basic machinery, there are a couple of things to make this nicer. x2vnc allows one to enter either a raw IP address (123.456.789.000), or some string to be resolved via DNS lookup ("mylaptop.iamageek.org"). For laptops getting a dynamic IP over a wireless source, one needs to lookup your current IP each and every time you want to run x2vnc (hence the dynamic bit). Of course, many setups tend to be stale and constantly feed you the same address over and over, but that just makes it even more annoying when it actually does change.
Solution? Setup your laptop to dynamically tell a third-party its current IP, and have that third-party resolve some hostname in your honor. There are a number of ways to go about this, many involving geekery. I myself simply installed a client from NoIP that offers free and easy dynamic DNS services. I recommend the same for you, but do whatever you like. NoIP is linked below. The only issue I ever have with this is the need to sometimes manually run the client when I bring a machine out of the various standby modes in Windows. But then, for any others that might experience this same problem, you already know that plenty of other fun geek things hate standby also, so this is just par for the course really.
Setup some sort of dynamic DNS service for your laptop. I recommend NoIP, which is linked to below.
Step 4: Minor scripting
As we're all about laziness (ahem! *efficiency*) here, our setup at this point is begging for a small little script to bring everything together. This will get us to the point where each time you turn your laptop on, you will either hit a single button, or run a one-line script that sets everything up in a single go. Even better would be if the laptop informed the workstation of its local presence, thus saving us from even needing to do this. However, I'm not really into the workings of geo-spatial referencing via wireless signals, and bluetooth (though I love it) is generally not found on standard department purchased hardware. Any ideas on this point are welcome.
The following is a one line call to x2vnc that includes the name of your laptop, as well as a password file that you will have to create:
`x2vnc mylaptop.imageek.org:0 -passwdfile /somepath/someuser/.vnc/mylaptop.passwd`
I recommend either aliasing this, putting it in some script in a personal bin directory, or creating a Gnome/KDE launch button with this set as the command.