I forgot to include the relevant part of /var/log/messages:

    Jun 18 18:24:12 george kernel: drivers/hid/usbhid/hid-core.c: can't
    reset device, 0000:00:1a.0-1.2.2.4/input0, status -71
    Jun 18 18:24:12 george kernel: usb 1-1.2.2: clear tt 1 (00a0) error -71
    Jun 18 18:24:12 george kernel: drivers/hid/usbhid/hid-core.c: can't
    reset device, 0000:00:1a.0-1.2.2.3/input0, status -71
    Jun 18 18:24:12 george kernel: usb 1-1.2.2: USB disconnect, address 6
    Jun 18 18:24:12 george kernel: usb 1-1.2.2.1: USB disconnect, address 7
    Jun 18 18:24:12 george kernel: usb 1-1.2.2.2: USB disconnect, address 8
    Jun 18 18:24:12 george kernel: usb 1-1.2.2.3: USB disconnect, address 9
    Jun 18 18:24:17 george kernel: drivers/hid/usbhid/hid-core.c: can't
    reset device, 0000:00:1a.0-1.2.2.3/input1, status -110
    Jun 18 18:24:17 george kernel: usb 1-1.2.2: clear tt 1 (0090) error -19
    Jun 18 18:24:17 george kernel: usb 1-1.2.2.4: USB disconnect, address 10

After that the system is unresponsive and I have reboot with the power 
switch.  Those devices are:

Jun 18 14:47:40 george kernel: generic-usb 0003:04F2:0833.0002: 
input,hidraw1: USB HID v1.11 Keyboard [CHICONY USB Keyboard] on 
usb-0000:00:1a.0-1.2.2.3/input0
Jun 18 14:47:40 george kernel: generic-usb 0003:045E:0029.0004: 
input,hidraw3: USB HID v1.00 Mouse [Microsoft Microsoft IntelliMouse® 
Optical] on usb-0000:00:1a.0-1.2.2.4/input0


One last question, should I post text only to this list or are people 
happy with html formatting?

Joe


On 06/18/2012 07:42 PM, Joseph Areeda wrote:
> Greeting,
>
> This is my first post to this list, I'm hoping for some insight into a 
> vexing problem.
>
> My situation is 4 computers running various operating systems: Ubuntu, 
> Scientific Linux 6.2, Debian Squeeze, MacOS X lion, and Windows 7. All 
> except the Mac and one Ubuntu are multiple boot and used for cross 
> platform development and testing.
>
> I have a manual switch box with computers on ports 1-4 and a powered 
> USB hub with mouse, keyboard, scanner, microphone and a USB headset 
> adapter.
>
> There is also an HDMI and DVI switch box but they are not part of of 
> the problem. Together they give me a more flexible KVM switch. I can 
> watch a long running job on Monitor #2 connected to one system while 
> working on another using Monitor #1.
>
> When I boot up everything works fine. I can also switch between 
> systems freely. However if I leave one of the Linux systems 
> disconnected for a long while it doesn't respond when I switch back to 
> it. I have not seen this behavior with Windows or Mac. The Mac is 
> sometimes disconnected for days but Windows usually gets booted back 
> into Linux when I'm done with it.
>
> SL6 usually dies but I can usually ssh into Ubuntu and reboot 
> cleanly.  SL6 doesn't respond to pings or ssh.  As long as I keep the 
> switch box on SL6 it has run for weeks.
>
> Now I have tried 2 different USB switch boxes and when it doesn't 
> respond it doesn't respond even if I plug the hub, or the mouse and 
> keyboard directly into the usb ports on the system. I don't believe it 
> has anything to do with the KVM the OP mentioned or my switch boxes.
>
> Searching the web I found a comment that said a powered hub per system 
> worked with his USB KVM switch. I suspect we're seeing some sort of 
> USB timeout. I suppose I can get a powered hub per system but I built 
> these machines with 6 or 10 USB 2 and 2 or 4 USB 3 ports so I really 
> don't need them. The powered hub may however convince Linux that there 
> is something plugged into that port and keep it alive.
>
> Does anyone know of a reason for this or even better a fix for it?
>
> Thanks
> Joe