From owner-freebsd-hackers Thu Jun 29 17:11: 1 2000 Delivered-To: freebsd-hackers@freebsd.org Received: from mail-04-real.cdsnet.net (mail-04-real.cdsnet.net [204.118.244.94]) by hub.freebsd.org (Postfix) with SMTP id 0450437C29C for ; Thu, 29 Jun 2000 17:10:52 -0700 (PDT) (envelope-from mrcpu@internetcds.com) Received: (qmail 95000 invoked from network); 30 Jun 2000 00:10:45 -0000 Received: from schizo.cdsnet.net (204.118.244.32) by mail-04-real.cdsnet.net with SMTP; 30 Jun 2000 00:10:45 -0000 Date: Thu, 29 Jun 2000 17:04:18 -0700 (PDT) From: Jaye Mathisen X-Sender: mrcpu@schizo.cdsnet.net To: hackers@freebsd.org Subject: Funky scheduler stuff under heavy I/O. Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-freebsd-hackers@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.ORG Noticing something a tad odd here, under 4.0-stable, as of 6/27. 8 identical drives, (all IDE, all connected to identical IDE PCI controllers). 8 parallel DD's started at the same type creating 2GB cycbuffs (as fast as & can put them in the background). (dd if=/dev/zero bs=1m count=2000 of=blahblah) The amount of data actually written to the buffs varies widely. both the Master and Slave drives on the first channel of each controller are about 200MB's ahead of the master and slave drives on the second channel of each controller, and the gap is growing. Also, the number of tps on the slower set is about 2/3 the # of tps on the faster set... Originally I was using vinum, and thought it was in there, but this is just going straight through UFS, no vinum inbetween... Any brilliant ideas? To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-hackers" in the body of the message