From mboxrd@z Thu Jan 1 00:00:00 1970 From: Rik van Riel Subject: Re: report stolen time via pvclock? Date: Tue, 09 Mar 2010 16:39:48 -0500 Message-ID: <4B96C024.5080605@redhat.com> References: <201003092147.38498.thomas@scripty.at> <20100309213009.GA26204@amt.cnet> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit Cc: Thomas Treutner , KVM mailing list To: Marcelo Tosatti Return-path: Received: from mx1.redhat.com ([209.132.183.28]:18101 "EHLO mx1.redhat.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751239Ab0CIVjw (ORCPT ); Tue, 9 Mar 2010 16:39:52 -0500 In-Reply-To: <20100309213009.GA26204@amt.cnet> Sender: kvm-owner@vger.kernel.org List-ID: On 03/09/2010 04:30 PM, Marcelo Tosatti wrote: > On Tue, Mar 09, 2010 at 09:47:38PM +0100, Thomas Treutner wrote: >> Hi, >> >> I'm referring to this patchset >> >> http://www.mail-archive.com/kvm@vger.kernel.org/msg23810.html >> >> of Marcelo Tosatti. It seems it was never included or even discussed, although >> it's nearly half a year old. I wonder if there is a good reason for that? I'd >> like to use the steal time for my VMs, as I consider it useful in some cases. > > There is a problem with it: stolen time is accounted separately (in > addition to) user/system/idle. > > And as you noted there seems to be lack of interest in the feature. More like a lack of time to implement it properly for KVM. I know it is a useful feature for system administrators, and should probably try to get it reimplemented so it works right and with lower overhead than full schedstats (I have some ideas on how to achieve that). Thanks for reminding me of this project :)