From owner-freebsd-current Sat Dec 11 12:36: 3 1999 Delivered-To: freebsd-current@freebsd.org Received: from homer.mkintl.com (242671hfc111.tampabay.rr.com [24.26.71.111]) by hub.freebsd.org (Postfix) with ESMTP id 6967515216 for ; Sat, 11 Dec 1999 12:36:01 -0800 (PST) (envelope-from joeo@homer.mkintl.com) Received: from localhost (joeo@localhost) by homer.mkintl.com (8.9.3/8.9.3) with ESMTP id PAA14143 for ; Sat, 11 Dec 1999 15:35:24 -0500 Date: Sat, 11 Dec 1999 15:35:24 -0500 (EST) From: To: freebsd-current@freebsd.org Subject: Re: Errors from the ata disk driver Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-freebsd-current@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.ORG Will the backdown to PIO mode be permanent till the next reboot of the machine, or will the driver be able to attempt to return to DMA mode after a timeout period. I'm only seeing these errors under really heavy disk activity (mutlitple nfs readers and writers plus rsync/mirror jobs to the vinum volume in question). On the vinum front, does vinum have the ability to retry writing a block (or whatever the correct abstration is) to the same device if an error like this comes up from the underlying device? Thanks, keep up the great work (Yes, I know I should be using scsi for something like this, but the price of a 200GB raid0 unit built out of IDE equipment compared to the same built out of SCSI hardware is quite shocking. Especially for a non-mission critical role.) On Sat, 11 Dec 1999, Andrew Gallatin wrote: > These are UDMA CRC errors. Whatever you do, DO NOT upgrade to a > recent current without reading the ata driver's commit logs very > carefully. > > The ata-driver has recently grown recovery code where it will try to > back down to PIO mode to fetch such blocks. As recently as last week, > the ata driver would lock a machine solid (unpingable, reset or > power-cycle required) when attempting to back down to PIO mode when > the drive in question was attached to a Promise Ultra controller. > > Soren knows about the problem & is going to fix it. > >------------------------------------------------------------------------------ > Andrew Gallatin, Sr Systems Programmer http://www.cs.duke.edu/~gallatin > Duke University Email:gallatin@cs.duke.edu > Department of Computer Science Phone: (919) 660-6590 > > To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-current" in the body of the message