Date: Mon, 14 Oct 1996 18:19:26 +0200 From: se@mi.uni-koeln.de (Stefan Esser) To: taob@io.org (Brian Tao) Cc: se@zpr.uni-koeln.de (Stefan Esser), freebsd-scsi@freebsd.org (FREEBSD-SCSI-L) Subject: Re: Wonky controller or drive? Message-ID: <199610141619.SAA04657@x14.mi.uni-koeln.de> In-Reply-To: <Pine.BSF.3.95.961014114553.2494F-100000@rogue.io.org>; from Brian Tao on Oct 14, 1996 11:58:59 -0400 References: <Pine.BSF.3.95.961014114553.2494F-100000@rogue.io.org>
next in thread | previous in thread | raw e-mail | index | archive | help
Brian Tao writes: > On Mon, 14 Oct 1996, Stefan Esser wrote: > > > > Hmm, adding the 7th drive caused problems ??? > > No, not precisely. This is the history: sd1 was a 4GB Quantum > Atlas to begin with. That one died during a power outage about a > month ago. I replaced it with a 1GB drive in the meantime. That's > still seven devices. I only recently replaced it with a new 4GB > drive, since we were running out of room on the 1GB (Web and FTP > server logs, mostly). Ok. > The cable configuration hasn't changed, just that one drive. sd0 > and sd1 (1GB boot drive, plus the new 4GB drive) are mounted > internally in the PC (50-wire ribbon cable, about 40 cm). The > remaining drives are in two daisy-chained external enclosures, total > cable length of about 150 cm (two lengths of external cabling, two > lengths of internal ribbon cabling), out to the last drive. The > internal chain is terminated with jumpers on the Quantum. The > external chain has an active terminator plugged into the last > enclosure. Each enclosure has its own power supply and fans. One has > four bays (three drives) and the other has two bays (two drives). Hmm, I count 3 internal and two external cables ... I guess each internal cable is some 100cm, actually, at least I've yet to see a shorter cable being delivered with a system. (Well, if you made it from components yourself this is different :) > > I expect this to be caused by either a too > > long cable (for the transfer rate) or a problem > > with the power supplies. > > What is the maximum? 6 m or something like that? I should be > well within spec. I've seen problems with much less than 6m of cable, and I would not trust single-ended Fast SCSI with more than 3m. Don't forget, that those maximum length specs do only apply to SCSI cable of a given quality (impendance and capacitance), and your internal cables are probably far from optimal ... > > Well, my first guess would be the SCSI cable being > > too long (or not good enough) or the peak load on the > > power supply being too high. > > > > You can check the prior by using only slow transfers > > (async. or at most 5MB/s sync). If the power supply > > is at its limit, then you should be able to cause > > failures by increasing the seek rate (ie. do random > > seeks with little data actually being transferred). > > Hummmm... what could I use to do that? Running bonnie repeatedly > on the 4GB drive for the random seek tests? I don't think there is a I'd rather use multiple simultanously running "find" processes, startet half a minute apart in order to not take advantage of the caches ... > way on the NCR (unlike the Adaptecs) to specify the bus speed or > sync/async mode. Well, I use /usr/sbin/ncrcontrol for that purpose :) # ncrcontrol -s sync=5 limits the transfer rate to 5MHz. # ncrcontrol -t 1 -s async makes target 1 only use asynch. transfers. # ncrcontrol -t 3 -t 4 -s tags=0 -t 5 -s tags=8 disables sending tags to targets 3 and 4, while target 5 will receive up to 8 simultanous commands (instead of the default of 4 for a device supporting TCQ). # ncrcontrol -i T:L Vendor Device Rev Speed Max Wide Tags 0:0 Quantum XP32150 576D 10.0 10.0 8 4 1:0 DEC DSP3053LS X442 10.0 10.0 8 4 4:0 TOSHIBA CD-ROM XM-3601TA 0725 4.0 10.0 8 - 5:0 HP C1533A 9406 10.0 10.0 8 - # ncrcontrol -v -p 1 total XP32150 DSP3053L CD-ROM X C1533A transf. disconn interru ---- ms transfer ---- t/s kb/s t/s kb/s t/s kb/s t/s kb/s t/s kb/s length exp une fly brk total pre post disc 26 174 26 174 0 0 0 0 0 0 6853 33 0 26 0 9.6 0.4 0.4 5.8 43 153 43 153 0 0 0 0 0 0 3644 48 0 43 0 9.3 0.9 0.0 7.0 40 188 40 188 0 0 0 0 0 0 4813 55 0 40 0 10.8 0.2 0.0 4.5 30 168 30 168 0 0 0 0 0 0 5734 34 0 30 0 15.0 0.0 0.3 12.3 73 363 73 363 0 0 0 0 0 0 5092 72 0 73 0 20.4 0.5 0.1 16.2 (Disk activity caused by "find / -print".) There is a man page for ncrcontrol, BTW, and "ncrcontrol -?" and "-s ?" give some information on supported options. Regards, STefan
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199610141619.SAA04657>