Date: Sat, 7 Mar 1998 23:14:13 +0100 From: Stefan Esser <se@FreeBSD.ORG> To: jhs@FreeBSD.ORG, scsi@FreeBSD.ORG Cc: me@FreeBSD.ORG, gj@FreeBSD.ORG, Stefan Esser <se@FreeBSD.ORG> Subject: Re: your mail Message-ID: <19980307231413.32288@mi.uni-koeln.de> In-Reply-To: <199803071731.JAA27263@hub.freebsd.org>; from Julian Stacey on Sat, Mar 07, 1998 at 09:31:09AM -0800 References: <199803071731.JAA27263@hub.freebsd.org>
next in thread | previous in thread | raw e-mail | index | archive | help
On 1998-03-07 09:31 -0800, Julian Stacey <jhs@FreeBSD.ORG> wrote: > Hi SCSI people, Hi Julian! > Is it necessary to format a 4.3G SCSI disc on the same system + controller > ncr0 <ncr 53c810a fast10 scsi> rev 18 int a irq 10 on pci0:10 > that the disc will later run FreeBSD on ? > I believe it doesnt' matter which SCSI controller you use to format ? > (so long as it issues a sensible sequence of format commands, > & does actually format, The low level format is initiated by sending a standard SCSI command to the drive. But you normally should not do this, as has just been discussed in another mail list (-hackers ?). The MBR contains assumed geometry information, and if you don't use "dangerously dedicated" mode, then you have to make sure the information matches what the SCSI card's BIOS assumes. The NCR SDMS code is very flexible, it has a default geometry, but reads in the MDR and accepts a geometry that has been used by some other SCSI BIOS before (e.g. if an AH2940 was used in the installation process and the drive is moved to a 53c8xx later). > I think the last of the `format with the controller you will run with' > dependencies was back with the ST506 ? I never used one. The first small system with hard-disk that I used for some time had a 3.5" Rodime 5MB SCSI drive (connected to a Z80 single board card with the SCSI host adapter implemented by means of a modified parallel port). > I'd agree to `It'd do no harm, why not if you can', but > - My box with the NCR SCSI controller has no DOS partition, The ASUS SCSI controller floppies are for a dos base (& multiple flops, > so squeezing down to 1 DOS boot flop would be difficult). > - An adjacent system with Adaptec 1542c formats conveniently from bios Ctl. A > So I just used the target PC's power, & cross hitched a scsi ribbon to format. > It worked fine twice before, only got some bad sectors months later. Hmmm ... > I bought my first > "IBM DCAS-34330 S61A" type 0 fixed SCSI 2 > 1997.08.26, it developed bad sectors 1997.10.09, when I bought another > "IBM DCAS-34330 S65A" type 0 fixed SCSI 2 > to rescue the data to, when I reformatted both discs using the 1542C (I think). > The old S61A is now in a 3rd system (1542B), running fine, > neither are warm, both have power supplies with good DVM volt readings > (but I haven't oscilloscoped the power rail for noise , > but yes connectors are tight).. > > As this was the 2nd set of bad sectors with the S65A, I took it back to the > dealer (wasted time, Koenig in Munich Not reccomended !), I'd now like > to research deeper, & list both tables of bad sectors & see how full they are, > how many tracks dedicated to free sectors etc, & monitor the count regularly, > on both discs, & if there's a systematic failure mode on both discs, > to ignore the hostile Wintel oriented dealer, & report failure direct to > someone at www.storage.ibm.com, if these discs are exhibing some batch > oriented bad sector growth symptoms. Do you know about the S.M.A.R.T features of the DCAS drives ? There should be some information available from the IBM storage devices Web server. I haven't looked into this in detail, but it should provide you with information on the health of the drive and with early warnings in case of soft errors or marginal behaviour. > I read > man [0-9] scsi > /usr/share/misc/scsi_modes > /usr/include/scsi.h > & happen to have a Seagate SCSI booklet, but it's heavy going, > & a few more example commands would help both `man scsi` & me. > > BTW on my 2.2.5 system with the NCR controller, (with root on wd0) I did: > scsi -f /dev/rsd0c -m 1 -e -P 3 > AWRE (Auto Write Reallocation Enbld): 1 > ARRE (Auto Read Reallocation Enbld): 1 > [ so the disc _should_ be self correcting, > but obviously isn't, is running out of sectors ] This will only help, if errors develop slowly after setting the flags. The drive will take a (recovered) severe soft read failure as a signal to reallocate the block. But in case of a hard error, there is no way to rescue the old contents of the sectors, and the drive will flag the reallocated block as invalid, until new data is written to it for the first time. > TB (Transfer Block): 0 > RC (Read Continuous): 0 > PER (Post Error): 0 > DTE (Disable Transfer on Error): 0 > DCR (Disable Correction): 0 > Read Retry Count: 1 > Correction Span: 0 > Write Retry Count: 1 Hmmm, I've got 8 for both retry count values ... > scsi -f /dev/rsd0c -c "4 0 0 0 0 0" > SCIOCCOMMAND ioctl: Command accepted. > return status 1 (Command Timeout) after 2000 msCommand out (6 of 6): > 04 00 00 00 00 00 > No sense sent. > [ Don't know what's wrong The script /sbin/scsiformat adds a "-s 7200" timeout parameter. You may need to increase that number, depending on the capacity of your drive and the time per cylinder of the drive's implementation if format in the firmware. > Then I gave up & formatted with 1542C (with no problem). > So I'll now rebuild with this drive, & in another few months I expect it will > develop bad sectors again ..... Any suggestions, sample scsi systax ? Have a look at /sbin/scsiformat. I never tried it (don't want to format any of my drives, currently :) but others reported success ... > PS I'm fresh subscribed to scsi@, but no traffic yet, > so a cc: jhs@freebsd.org would make sure I see your reply ... Thanks ! Traffic on -scsi varies depending on the respective activity of the developers. Guess that the import of Justin Gibbs new generic SCSI layer will make this list much more interesting over the next few weeks :) Regards, STefan To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-scsi" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?19980307231413.32288>