On 03/10/2015 08:15 PM, W. Steven Schneider wrote: > > Damn it, I hate using my phone for this! > > On Mar 10, 2015 6:15 PM, "Jason Adams" <adams...@gmail.com > <mailto:adams...@gmail.com>> wrote: > > > > On 03/08/2015 09:38 PM, Steven wrote: > > > I've got a set up between two towers where I use a KVM (KVMS?) > > > switch between them. The one running OpenBSD (snaphots and recent as > > > of this morning) seems step up it's CPU speed when I'm switched out > > > to the other computer. I'm wondering if I'm the only one seeing this. > > > > Can you ssh into the machine and then switch the KVMS away from it and see > > what "top" > > says is chewing up CPU cycles? > > > It appears that the privilege separated Xorg is demanding a high percentage > of CPU. I had Xorg > niced to -10 buy bringing it's back to a nice of 0 didn't change the behavior. > > Also, what about plugging in another keyboard? I had one old linux server, > > that had similar > > issues some years ago, and simply plugging in another cheap keyboard, and > > dropping it behind the > > table (never using it) allowed the switch (and the keyboard attached to it) > > to work without issue. > > > I might yet try that as a workaround if no solution comes to mind or is > forthcoming. :-) >
By the way, the same might be tried with some random mouse plugged directly into the offending machine. It sounds like xorg is expecting to read state from input devices, and failing that it works harder. But I've also seen issues with simply the absense of a monitor causing problems. Some kvm switches compensate for this by emulating DDC connection to the switched away-from machine. Here is a random example chose from a quick search: http://linkskey.com/product/ldv-302arc-2-port-dvi-usb-kvm-w-audiomic-plus-quickswitch-remote-button/ There are some KVM switches that handle this condition -- Those who do not understand Unix are condemned to reinvent it, poorly.
