Date: Mon, 4 Aug 2003 22:14:45 +0200 From: Wilko Bulte <wkb@freebie.xs4all.nl> To: Don Bowman <don@sandvine.com> Cc: "'aic7xxx@freebsd.org'" <aic7xxx@freebsd.org> Subject: Re: AIC7902 SCSI aborted command Message-ID: <20030804201445.GE13405@freebie.xs4all.nl> In-Reply-To: <FE045D4D9F7AED4CBFF1B3B813C8533702742114@mail.sandvine.com> References: <FE045D4D9F7AED4CBFF1B3B813C8533702742114@mail.sandvine.com>
next in thread | previous in thread | raw e-mail | index | archive | help
On Mon, Aug 04, 2003 at 04:10:15PM -0400, Don Bowman wrote: > From: Justin T. Gibbs [mailto:gibbs@scsiguy.com] > > > After ~200 hours of operation a PC spit out this message: > > > > > > Copied 18 bytes of sense data offset 12: 0xf0 0x0 0xb 0x0 > > 0x0 0x40 0xff 0xa > > > 0x0 0x0 0x0 0x0 0x47 0x3 0x8 0x0 0x0 0x0 > > > (da0:ahd0:0:0:0): WRITE(06). CDB: a 0 40 ff 80 0 > > > (da0:ahd0:0:0:0): ABORTED COMMAND info:40ff asc:47,3 > > > (da0:ahd0:0:0:0): Reserved ASC/ASCQ pair field replaceable unit: 8 > > > > > > The system has a single U320 15KRPM seagate 18GB drive in it. > > > No other messages are present. > > > > This means that the target saw a CRC miscompare on a packet. CAM > > retried the command and it succeeded. CRC errors are not unheard > > of in U320 configurations, but at only 200 hours of load, that is > > a little too often for comfort. > > > > I believe you mentioned before that you were using an extremely > > short cable in one of your configurations. Be aware that there > > is a minimum cable length suggested by the SCSI spec. This > > varies depending on the capacitance rating of your cable, but > > the shortest length allowed is 10cm (cables with less capacitance > > need larger spacing). If your cable is shorter than this, you might > > consider trying something longer. > > Me & my trusty ruler re-measured the cable, and straightened out, > it is 10.4" tip to tail, which matches the spec. I was foolishly > thinking of the folded length. It isn't too folded I hope? In the sense that you sometimes find flatcables stashed in miniscule cutouts in cases etc. Not too good for signal integrity either. -- Wilko
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20030804201445.GE13405>