fmbutt@cloud9.stratus.com (Farooq Butt) (06/12/89)
Here is a little story which may be of interest to comp.sys.sun users. Please direct any bouquets/brickbats to me personally. It was a dark and stormy day outside Boston when our 3/280S finally arrived. True to our graduate student heritage our group hurriedly ripped off the various wrappings and cartons and proceeded down the ominous path of bolting together the various components we had ordered. In order to save our company some greenbacks * , we had ordered a highly ...>ahem<... nonstandard configuration: * $55,000 to be exact. 1. 3/280s (i.e. CPU board, 8M memory, 2nd ethernet, ALM and "fridge") 2. Xylogics 753 controllers (up to 4 spindles) 3. Xylogics 472 tape controller 4. VME <--> Multibus Adapter (Sun supplied for 472 ctlr) 5. Random 3rd party 6U -> 9U adapters for Xylogics 753 ctrlrs 6. Randomly supplied CDC 92-185-02V tapedrive 7. Randomly supplied 781Mbyte Fujitsu M2372K microdisks 8. 8M auxilliary memory from Clearpoint 9. Cabling for drives (both tape and disk) from Arrow Electronics We knew the task ahead of us was not going to be an easy one so we girded up our loins. Noted social commentator and Sun guru extraordinaire Frank Bull had apparently hooked up Fuji M2372K's before and armed with his list of DIP switch settings (posted to sun spots some time ago) we started preparing our disks to be "Sunified." The server was powered up and after some minor EEPROM surgery we had it sending output to ttya and had hooked up a tvi950 terminal to it. Everything looked great! The diagnostics showed that all the internal tests were being passed and that the system seemed to be working OK. Unfortunately at this point only the boards we had actually purchased from Sun were installed. Feeling happy that our server's brains had not been fried by Joe's delivery agency we got four burly guys together to help install our CDC tape drive. CDC apparently believes in the value of 'good ol' American Steel' and their tape drives weigh in at about 160 lbs. Herniated and happy we switched on the tape drive and lo and behold it worked, the internal diagnostics proclaimed that all was OK. After that victory we installed the Xylogics 472 ctlr into its VME<-> Multibus adapter and plugged it into the VME bus. MUNIX came up OK but it would not see the tape ctlr no matter what we did. In desparation we called Sun genius Mr. Jim Dion of Sun Microsystems who duly informed us that the VME<->Multibus adapter supplied by Sun had 12 (count em' !) banks of DIP switches and that we had almost certainly misconfigured them. As it turned out since we had ordered our tape ctlr from Xylogics and our VME<->Multibus adapter from Sun, Sun had conveniently assumed that it was supplied on a "spare-part" basis and supplied absolutely NO documentation with it ! With the help of Mr. Dion we managed to configure the 472 correctly and actually saw it appear as a device during the boot sequence. The Clearpoint memory board was installed quickly thereafter and gave us no problems. Feeling elated by our victory over the forces of darkness, we looked around at ourselves and saw that it was time for the "real-man" stuff. Enough of this quiche-eating messing about with devices that are known and supported, we had to now venture into the unfamiliar terrain of unsupported/unacknowledged devices....in short, we had to hook up the Fujis. Well, we know others had done it and that made us feel good. Little did we know what horrors awaited us. Hooking up the Xylogics 753 boards in their 6U<->9U adapters (the xy753 as opposed to the Sun 7053 is basically the same board but it is much smaller) was simple. A quick glance at the MUNIX boot sequence showed that all was well in kernel-land and that we had successfully fooled SunOS into thinking that there was a 7053 on our VME bus. Hooking up the cables was a little weird since we had not done it on Micro-SMD devices such as the Xy753 ctlr but other than that it was not particularly difficult. We went with one disk to be safe and thought that our task was done. Power on....controller shows up but no disks....panic. Oh oh, we realized, silly us we had forgotten to put a terminating resistor pack (55ohmsx4) on the drive, so we dug up four resistor ICs and installed them on the drive and powered up again. No change. I made a frantic call to Frank Bull who indicated that I may have mis-set my jumpers on the 753 ctlr. The 753 was pulled out and Frank went over the jumpers with me. Everything looked OK. What *could* be the matter? In desparation, we checked the A & B cables for connectivity, they seemed fine. The answer, we felt must lie in the drive if the ctlr is showing up on the boot bus-probe and if both its DIPs and the cables are OK. Another frantic call to Frank who patiently went over the disk drive DIPs with me. They seemed to be set just fine. What could the answer be? With the 50 lb disk on my lap cutting deep gashes into my thighs, I suddenly knew !! It *must* be the jumpers on the drive !! I checked the drive jumpers against Frank and sure enough a couple were out of whack. I modified them to suit what I was told works and plugged the drive back into the chassis. No dice. This was getting desparate and if I didn't get this beast up I know that both my partner and myself would be given the figurative Soweto-Burning-Radial-Tire necklace by our company for having had the audacity to think that we were a third-party system integration house. What were we to to do? The computer is OK...it boots fine it sees every board that is plugged into it. It sometimes has a little trouble telling how much memory it has installed but that's OK since MUNIX comes up just fine off tape. But you can't be too safe so I called Mr. Sun Genius Jim Dion in again to swap what we felt was a flaky CPU board on a brand-new machine. The swap took place and our minor memory flakiness went away. The disks, I was sure at this point, would now be "seen" and we would light up the Benson & Hedges and ride off into the sunset. No such luck. I was now at the end of my tether and I called Frank in Arizona again with a sob story. We went over the facts: Good computer (since it sees every device, boots MUNIX, passes DIAG and "feels" OK), good ctlr (since it "shows up" on a bus probe), good cabling (checked by an ohmmeter) and presumably a good disk. This last fact was a tad shaky so in order to be absolutely certain we plugged the disk in question into a Stratus computer and it worked beautifully. This was getting strange. Well, we re-checked all the DIPs and jumpers but after calling Fujitsu we found out that the on-board jumpers control rev. level and shouldn't be touched. Apparently the reason my jumpers and Frank's had disagreed was because of board rev. I changed all the jumpers back to what they were and tried to see if that made a difference. No way Jose, same as always, the computer came up, did a bus probe, we saw the 753 ctrlr but no drives appreared. Back to the drawing board. It was at this time we did the smartest thing of all, namely to call in a hardware type person who came armed with a scope and logic-analyzer. It took our hardware nut about 3 hrs to determine that the chips on the disk were oscillating wildly, indicating that the drive was not receiving control signals from the 753. But how could that be? The cables were fine and the 753 seemed to check out OK. Well we plugged in our spare 753 hoping that maybe, just maybe, our 753 had been brain-damaged somehow and that replacing it with another would help. Same as ever, the darn disk drives refused to show and the drive chips oscillated. After much logic analysis of the drive we moved our probe to the first diode on the 753 which is supposed to be live at -12V. There seemed to be no juice. OH NO!!!! We ripped off the front panel of the Sun and quickly went to work probing its power supply. YEP!! NO -12V DC !!! We had found the guilty party, apparently none of the Sun-supplied equipment was in the least bit bothered by the lack of -12V or even -5V. The only piece of equipment that needed it and did not get it was the 753. It had used the +5V to drive its "brains" and had actually come up as a legit device but it needed -12V to control our disks. We changed power supplies and our Sun worked like a champ. I won't forget that one in a hurry. Besides being a generally humiliating experience it taught a software engineer that computrons are not the only things and that sometime electrons are also important! Thanks to Bob Wise, Pete Clemson, Jim Dion and Frank Bull for putting up with my/our exasperation. Farooq Butt fmbutt@cloud9.UUCP (508) 460-2798 - Stratus Computer, Marlboro MA.
grunwald@flute.cs.uiuc.edu (06/17/89)
yeah gods, a blast from the past -- several years ago, we bought a Cadlinc (another marketer of proto-Sun-1 boxes) system sans disk & installed a CDC drive that sounded like Ohare Intl. Same symptoms; ctrl visible, everything looking good, can poke into ctrl memory & see if say `no disks''. I call up the dude at Cadlinc, who sez, ``ok, lets walk through this. Take your borrowed Voltmeter & measure the -12V supply located beside the fan....what do you mean, there's nothing there.'' In order to save money, unless you bought a system with disk, they didn't install a -12V supply. Ain't like wonderful. -- Dirk Grunwald -- Univ. of Illinois (grunwald@flute.cs.uiuc.edu)