From owner-freebsd-performance@FreeBSD.ORG Fri Nov 19 01:55:46 2004 Return-Path: Delivered-To: freebsd-performance@freebsd.org Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125]) by hub.freebsd.org (Postfix) with ESMTP id 7DEB616A4CF for ; Fri, 19 Nov 2004 01:55:46 +0000 (GMT) Received: from rproxy.gmail.com (rproxy.gmail.com [64.233.170.200]) by mx1.FreeBSD.org (Postfix) with ESMTP id E2DAA43D49 for ; Fri, 19 Nov 2004 01:55:45 +0000 (GMT) (envelope-from aaron.glenn@gmail.com) Received: by rproxy.gmail.com with SMTP id b11so28916rne for ; Thu, 18 Nov 2004 17:55:45 -0800 (PST) DomainKey-Signature: a=rsa-sha1; q=dns; c=nofws; s=beta; d=gmail.com; h=received:message-id:date:from:reply-to:to:subject:cc:in-reply-to:mime-version:content-type:content-transfer-encoding:references; b=ldjvmGC2uWUeXfOUHje4jTUXgzSrZcTK0/Eie1HECNpGdkoVbhrhtj5BC0DnKLkL9KMGx5Z428jfyV4WfwtwE3e9R+kGZvV9wDFPWEI7d77w0XhSj5Yp83yEnHojjeVJYY12GM3R3xOcEEWQOzCmJwfkD8Z8VBSpUCVqsHDY4YQ= Received: by 10.38.71.67 with SMTP id t67mr189018rna; Thu, 18 Nov 2004 17:55:45 -0800 (PST) Received: by 10.38.151.56 with HTTP; Thu, 18 Nov 2004 17:55:45 -0800 (PST) Message-ID: <18f6019404111817553208789e@mail.gmail.com> Date: Thu, 18 Nov 2004 17:55:45 -0800 From: Aaron Glenn To: Sean Chittenden In-Reply-To: Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit References: <18f6019404111801471db5bbfd@mail.gmail.com> cc: freebsd-performance@freebsd.org Subject: Re: I've ran out of ideas X-BeenThere: freebsd-performance@freebsd.org X-Mailman-Version: 2.1.1 Precedence: list Reply-To: Aaron Glenn List-Id: Performance/tuning List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Fri, 19 Nov 2004 01:55:46 -0000 On Thu, 18 Nov 2004 15:29:14 -0800, Sean Chittenden wrote: > Are you using thttpd+sendfile(2)? If so, I'm going to guess you've > exhausted your sendfile(2) buffers. Can you run top to see what the > state of your thttpd process is? I've bumped into this numerous times. sendfile? I wasn't aware of that option. I'll look into it 11303 nobody -8 0 4124K 3516K biord 4:35 2.83% 2.83% thttpd This is with the previous sysctl changes mentioned, and a new kernel with HZ set to 2000 and DEVICE_POLLING enabled. Absolutely no change on the interrupts - still 12-14% cpu. > If you look through arch@ there's a patch from me to change the > behavior of sendfile(2) so that it returns with ENOBUFs instead of > blocking when your sendfile buffers are exhausted. I would've thought > that alc@'s work on sendfile(2) would've prevented this from happening, > but you may have just exhausted them in which case you're getting > killed by this. -sc > > -- > Sean Chittenden