Hi, Today I was contacted by a VMware colleague asking me why my test systems were using almost 100% CPU while waiting for a prompt. He was used to this behaviour for DOS guests :) Apparently the isolinux boot CD that is waiting for another test Linux installation is using about 1.8Ghz of the 2.5Ghz virtual CPU on a VMware ESX 3.0.2 host. Since I have 5 of these VM guests that are used for testing our Linux installation process and automatically boot into isolinux waiting for another testdrive (or to boot into the new installation) this is using quite some processing power while not doing anything most of the time. I was wondering if this was known behaviour ? And if it is something that we can fix ? Let me know if you need more information about our environment to troubleshoot. -- -- dag wieers, dag at wieers.com, http://dag.wieers.com/ -- [Any errors in spelling, tact or fact are transmission errors]
Dag Wieers wrote:> Today I was contacted by a VMware colleague asking me why my test systems > were using almost 100% CPU while waiting for a prompt. He was used to this > behaviour for DOS guests :)They didn't fix that after DOS -- one of my jobs is at a shop with many, many, many win2k VMs, and we definitely noticed the same kind of behaviour when they're waiting. And a new policy was born. - bish
Dag Wieers wrote:> I was wondering if this was known behaviour ? And if it is something that > we can fix ?This is known behaviour, and there isn't really anything we can do about it. The *only* device that we have to talk to that we can guarantee getting an interrupt for is the keyboard, the rest of the devices *have* to be handled by polling, via an idle loop. Vmware broke the PXE spec trying to "fix" this, and all they did was cause massive trouble elsewhere. I guess I might put a HLT in *as an option* (because people *will* complain that the serial port loses characters), but perhaps a better thing would be to not start the VM until you actually need it... -hpa