Date: Wed, 28 Jun 2000 22:34:52 +0200 (CEST) From: =?ISO-8859-1?Q?G=E9rard_Roudier?= <groudier@club-internet.fr> To: Holm Tiffe <holm@freibergnet.de> Cc: freebsd-scsi@FreeBSD.ORG Subject: Re: BDR with Quantum VIKING (I) Message-ID: <Pine.LNX.4.10.10006282148440.2340-100000@linux.local> In-Reply-To: <20000628172406.A94663@pegasus.freibergnet.de>
next in thread | previous in thread | raw e-mail | index | archive | help
On Wed, 28 Jun 2000, Holm Tiffe wrote: > Hi, >=20 > yesterday I've upgraded my old dual Pentium/100 to an Athlon 750 > and this old Quantum disk makes me some trouble. >=20 > Here the dmesg output: [ ... ] > sa0 at sym0 bus 0 target 3 lun 0 > sa0: <TANDBERG TDC 4222 =3D07:> Removable Sequential Access SCSI-2 device= =20 > sa0: 4.807MB/s transfers (4.807MHz, offset 8) > Mounting root from ufs:/dev/da1a > da0 at sym0 bus 0 target 0 lun 0 > da0: <CONNER CP3200-200mb-3.5 4040> Fixed Direct Access SCSI-CCS device= =20 > da0: 3.968MB/s transfers (3.968MHz, offset 8) > da0: 203MB (415872 512 byte sectors: 64H 32S/T 203C) > da4 at sym0 bus 0 target 6 lun 0 > da4: <QUANTUM VIKING 4.5 SCA 880R> Fixed Direct Access SCSI-2 device=20 > da4: 40.000MB/s transfers (20.000MHz, offset 16, 16bit), Tagged Queueing = Enabled > da4: 4345MB (8899737 512 byte sectors: 255H 63S/T 553C) > da2 at sym0 bus 0 target 2 lun 0 > da2: <HP C3010M1 0B04> Fixed Direct Access SCSI-2 device=20 > da2: 10.000MB/s transfers (10.000MHz, offset 8), Tagged Queueing Enabled > da2: 1910MB (3912172 512 byte sectors: 255H 63S/T 243C) > da1 at sym0 bus 0 target 1 lun 0 > da1: <HP 2.13 GB 2nd 0582> Fixed Direct Access SCSI-2 device=20 > da1: 10.000MB/s transfers (10.000MHz, offset 15), Tagged Queueing Enabled > da1: 2033MB (4165272 512 byte sectors: 255H 63S/T 259C) > cd0 at sym0 bus 0 target 4 lun 0 > cd0: <TOSHIBA CD-ROM XM-3501TA 3054> Removable CD-ROM SCSI-2 device=20 > cd0: 4.237MB/s transfers (4.237MHz, offset 16) > cd0: Attempt to query device size failed: NOT READY, Medium not present > da3 at sym0 bus 0 target 5 lun 0 > da3: <IBM DCAS-34330W S65A> Fixed Direct Access SCSI-2 device=20 > da3: 40.000MB/s transfers (20.000MHz, offset 15, 16bit) > da3: 4134MB (8467200 512 byte sectors: 255H 63S/T 527C) >=20 >=20 > After some heavy disk activity the scsi bus stops working and after an > timeout the console prints this: >=20 > (noperiph:sym0:0:-1:-1): SCSI BUS reset detected. >=20 > and the machine is working again for a while. The driver messages are sometimes too verbose and sometimes too sparse. The latter probably applied here.:) Given your description, it seems that for some reason the SCSI BUS hung up. There are numerous reasons for this to happen. A device, for example, can hang while it owns the BUS for obscure reasons. When a SCSI IO times out, the driver first tries to synchronize with the SCSI SCRIPTS and if it succeeds within less than 10 seconds, it will try to send a ABORT (or ABORT TAG if command is tagged) message to the device. If the new delay of 10 seconds also expires without the abort to have occurred, then the driver resets BUS, chip and SCRIPTS, and requeue all the commands to CAM. (All that with quite sparse messaging by default. :-) ) By the way, I maintain some slightly broken hardwares and firmwares in by box, in order to check error recovery. I notably use a FAST-40 BUS with a dubious cable that locks or experiences a SCSI ERROR every 3 to 5 minutes when stressed with IOs and gives errors from time to time when used normally (btw using another cable fixes the problem). I never saw a single bit having been lost from the 2 disks plugged on this BUS (Cheetah + DRVS) that contains also some useful data (but not critical:) ). > I have now disabled tagged queueing for this device and an parrallel cvs = update > for ports and src seems to work without any stops now. (the cvs three is = on this > device) >=20 > I know, those Quantum disks are basicaly crap, but is there a known worka= round > to get those working with tagged queueing ? (I've got this drive for free= ) I donnot know about the VIKING (I?), but I have a VIKING II LVD that works fine for me with tagged command queuing. The drive accepts up to 63 tagged commands simultaneously without whining about QUEUE FULL as ATLASes in a unpredicable way. Note that the ATLASes are a lot more silent= =20 about QUEUE FULL when write-caching is disabled. > The disk seems to have the latest firmware revision from Quantum (880R). > The controller is an Tekram DC390 Wide SCSI and booth the se und the > wide connectors are used. But the drive seems to use SCA connector and I heard that some SCA->68 converters can make problems, especially with SYMBIOS chips. This should be investigated, in my opinion. > Does anyone know how stable the autotermination on those controllers work= s ? The auto-termination of the Tekram boards never has been reported to me to make problems and seems to be quite smart. > I can't therminate the upper 8 bits manually by jumpers or in the bios se= ttings. IMO, Unless it is broken, the Tekram board should have detected that the narrow part of the BUS is also used and have properly terminated the upper BUS lines at the controller. > The OS is an 5.0-current from June 2 2000, cvs update is running currentl= y... >=20 > Holm > --=20 > FreibergNet Systemhaus GbR Holm Tiffe * Administration, Development > Systemhaus f=FCr Daten- und Netzwerktechnik phone +49 3731 7812= 79 > Unternehmensgruppe Liebscher & Partner fax +49 3731 781377 > D-09599 Freiberg * Am St. Niclas Schacht 13 http://www.freibergnet.de/ G=E9rard. 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?Pine.LNX.4.10.10006282148440.2340-100000>