From owner-freebsd-stable Wed Nov 4 11:24:09 1998 Return-Path: Received: (from majordom@localhost) by hub.freebsd.org (8.8.8/8.8.8) id LAA16893 for freebsd-stable-outgoing; Wed, 4 Nov 1998 11:24:09 -0800 (PST) (envelope-from owner-freebsd-stable@FreeBSD.ORG) Received: from sabre.goldsword.com (sabre.goldsword.com [199.170.202.32]) by hub.freebsd.org (8.8.8/8.8.8) with ESMTP id LAA16819 for ; Wed, 4 Nov 1998 11:23:57 -0800 (PST) (envelope-from jfarmer@sabre.goldsword.com) Received: (from jfarmer@localhost) by sabre.goldsword.com (8.8.8/8.8.8) id OAA24048; Wed, 4 Nov 1998 14:28:10 -0500 (EST) Date: Wed, 4 Nov 1998 14:28:10 -0500 (EST) From: "John T. Farmer" Message-Id: <199811041928.OAA24048@sabre.goldsword.com> To: dhw@whistle.com, freebsd-stable@FreeBSD.ORG, gaylord@gaylord.async.vt.edu, jim.king@mail.sstar.com Subject: Re: has this been fixed? Cc: jfarmer@goldsword.com Sender: owner-freebsd-stable@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.ORG On Wed, 04 Nov 1998 12:52:02 Jim King said: >There are published documents that are referred to as IDE standards. My >experience over the years has been that different drive manufacturers >interpret them differently, and newer standards tend to skimp on backwards >compatibility with older standards. I always cross my fingers when mixing >drives from different manufacturers, or drives that were manufactured more >than a year apart. > >Just last night, the simple act of tossing in a 2-year-old EIDE drive into >a system with a 6-month-old UDMA drive resulted in a *completely* >non-functional system, until I pulled out the EIDE drive AND reset the BIOS >configuration. I absolutely *hate* IDE. > Some days, SCSI is almost as bad what with the SCSI-2, ultra-SCSI, ultra2-SCSI, etc. I spent >5 hours Monday scratching my head (wasn't much hair left anyway...) after connecting a 5 year old Seagate 1.0GB drive to a new Adaptec 2940UW card. It would boot & run just fine, and a single user could access the drive, read, delete, and move files. If the user started a large database app on the server, then things "got interesting." Stray interrupts would be announced in the console log, the user would hang and soon the server would crash. Proved it wasn't the database by having other users try simple tasks. If 2 or more attempted access to the drive, it went down. Finally traced it to the fact that the Adaptec supplied drivers default the tag command queue size for every device to 16. Varying this number didn't change anything. Finally had to disable tag queueing entirely. (This was a Novell server but I have had similar things bite me under FBSD). Related to IDE, I had a client ask me why I placed the CD-ROM drive on the second IDE channel on the machines I built for him. Told him that it was for performance, both the machines and mine. Seperating the drives meant that I didn't spend time getting Brand X drive to work with Brand Y cd-rom drive... John ------------------------------------------------------------------------- John T. Farmer Proprietor, GoldSword Systems jfarmer@goldsword.com Public Internet Access in East Tennessee Office: (423)691-6498 for info, e-mail to info@goldsword.com Network Design, Internet Services & Servers, Consulting To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-stable" in the body of the message