back to article Who needs the A-Team or MacGyver when there's a techie with an SCSI cable?

Welcome back to Who, Me?, The Register's regular feature covering confessions from readers who just wanted to get the job done. "Bobby" (for that is most definitely not his name) is the man in the comfy armchair this week, ready to regale us with misdeeds from the previous century (or it might have been this one – it was a …

  1. ColinPa

    Lonely machine needing company before it would work

    I was told this in the bar at a conference, so cannot vouch for its 100% truth....

    In the corner of a big machine room were a couple of machines. They moved one (as part of an upgrade) and the other machine stopped working. They moved it back again and the other one continued to work, so they left them both there. They found out a few years later that there was a maze of cables under the raised floor, and one cable was joined to another ( and stretched tightly). When there was no weight on the raised floor the the contact was broken. When there was weight on the the floor tile the cable relaxed 1 mm and the connection was made and the machine worked.

    1. elhvb

      Re: Lonely machine needing company before it would work

      I had it the other way around with a network cable where the center and the outer shell would short whenever someone walked atop the cable gutter.

      As for Frankenservers, I had to build one when the only available hardware wasn't having a power supply big enough to also give juice to the disk cabinet. I has been running for a few weeks waiting for the actual replacement server to arrive. And yes, this was SCSI as well.

      1. Anonymous Coward
        Anonymous Coward

        Re: Lonely machine needing company before it would work

        Ah yes. The old "Don't step on that floor tile..."

        <Ker dunk>

        "...it's loose and causes a head crash on the disk next to it"

        Oh well. It only used to take 30-40 minutes to completely boot up a 2966 into CME from a cold start, and we had nothing better to do on a wet Saturday in Liverpool anyhow.

        1. Andrew Norton

          Re: Lonely machine needing company before it would work

          Sounds suspiciously like a certain place on Old Hall st

    2. 0laf

      Re: Lonely machine needing company before it would work

      I think stories like that are surprisingly common.

      1. P. Lee

        Re: Lonely machine needing company before it would work

        Has anyone here not connected the disk in one system to the interfaces in another?

        Maybe I'm old...

    3. Anonymous Coward
      Pint

      Re: Lonely machine needing company before it would work

      I was told this in the bar at a conference, so cannot vouch for its 100% truth....

      I read this in a Reg comment, so ...

    4. Soruk
      Boffin

      Re: Lonely machine needing company before it would work

      On the theme of cables behaving badly, the ISP I worked at sent me and one other to the data centre in early 2014 to fit a SIP to E1 converter box to connect to our then new call centre provider's Avaya kit. It absolutely HAD to be working before we left, and of course we needed an E1 crossover cable, and that's not the sort of thing we had lying around. Did we have any RJ45 jacks or a crimp tool there either? Nope.

      So while my colleague configured the box I googled pinouts on my (now deceased) HP Touchpad, and calculated the adjustments required to modify a straight Ethernet cable. Again, no tools available so the cable was literally hacked into using a house key, appropriate wires cut and rejoined, twisted together and held together and insulated with electrical tape. It was ugly as hell and had no right whatsoever to work let alone exist, but it did work. Our call centre was online.

      A few years later it was decided a proper lead would be crimped and taken to the DC. Sadly, this replacement cable refused to function as intended, or even remotely close to it, so the old hack-job cable was reinstated, with service restored.

      The cable was finally retired in early 2019 along with the E1 kit when we changed call centre operators who interconnected with us using SIP.

  2. Anonymous Coward
    Anonymous Coward

    I used t build Frankenservers, not by choice.

    A company i used to work for as sales account manager used to put Frankeservers into company's.

    It wasn't by choice the workshop manager, who will remain nameless used to go out to site to spec up machines and pass them to the sales accounts mangers to price up the servers. He would hit the roof when we went back with branded kit. He would insist we built into a Cheiftec tower, with a Asrock K7s41 motherboard (that's right the cheapest mob we had that would traditionally loose its A drive) and IDE drives. His reasoning was that we could get the parts off the shelf from the parts warehouse (we sold to trade and retail also and had a sales counter). Wasn't until he left the company that the sales manager was able to change policy and we sold HP's from there on in. Oddly enough the number of server warranty issues we had fell dramatically shortly after the regime change. #OddThat

    1. Anonymous Coward
      Anonymous Coward

      Re: I used t build Frankenservers, not by choice.

      We took over a company who were absolutely certain that they had a proper server because it was in a really big (beige, unmarked) case and their previous IT guy said it was fine... mATX motherboard, IDE drives, no redundancy anywhere....and a horrible grinding noise periodically. Lasted just long enough to get P2V'ed, thanks goodness. I sympathise.

    2. chuBb.

      Re: I used t build Frankenservers, not by choice.

      Not quite a frankenserver, but i have been responsible for many drive cage expansions facilitated by zip ties and electrical tape in tower servers. Was very effective for some 10k rpm screamer drives that used to over heat as they were built into a fully populated cage with no thought of heat mangement, pulled every other one out and suspended in 5 1/4" bays, doubt that was every changed can only assume its been decomissioned now

      Have also been responsible for politicoservers, aka the big boss had to have the big box. So that led to putting a mATX mobo into a full tower case and rebuilding the (thankfully modest) server into the mini tower. Actually worked out well as the case was small enough to fit on a high shelf and never got fiddled with

      1. Stuart Castle Silver badge

        Re: I used t build Frankenservers, not by choice.

        "doubt that was every changed can only assume its been decomissioned now"

        Don't assume.. I used to manage an Apple xserve in one of our server rooms. It was used to manage our fleet of apple hardware, but when Apple terminated the Xserve, and effectively moved out of the server business, I immediately put in a requested, with a detailed plan that included the decommissioning and removal of the Xserve, and the secure destruction of it's HDDs (as required by company policy), along with the plans for introducing the new system. This was back in about 2011 or 2012. I went into that server room last week, only to find the Xserve still in position in the rack, unplugged, but otherwise untouched. As was it's replacement. This was still plugged in, but was also supposed to have been decommissioned, because it itself has also been replaced. (I actually don't know why. I was off when the decision was taken to replace it).

        1. tcmonkey

          Re: I used t build Frankenservers, not by choice.

          I evicted a couple of 486 PC-based telephony appliances from the bottom of a comms rack less than 12 months ago...

          1. Anonymous Coward
            Anonymous Coward

            Re: I used t build Frankenservers, not by choice.

            At one of my previous employers I was involved in a bid to virtualise the various systems that supported the NOC of a major Telco.

            This would have been in 2016, and there was stuff in those racks that date back more than 20 years. Point solutions running on ancient versions of Unix made by companies long since vanished on hardware that still had RISC processors. Kit so old that the operations procedures manual included instructions on how to hold a seance and the closest thing that they had to a bug fix was the occasional exorcism.

            Ok, so I exaggerate, but it was all pretty decrepit.

          2. Anonymous Coward
            Anonymous Coward

            Re: I used t build Frankenservers, not by choice.

            Funny that. I worked for a 500 employee company that had NT servers in a rack running their voicemail system. Since they had call groups, voice mail was fairly important. In 2010 I designed a replacement cloud based system that we could pay for by disconnecting four of our 8 T1's. I never got buy in and left in 2012. Yesterday I discovered they are still blissfully running those NT servers...

    3. big_D Silver badge

      Re: I used t build Frankenservers, not by choice.

      One company used to sell newish servers with old RAID controllers, because that was all SUSE from 2000 supported, in 2015! Finally the supplier couldn't supply any more 2000 vintage RAID controllers and the devs were forced to update their software to run on something a little more modern...

    4. Fungus Bob

      Re: I used t build Frankenservers, not by choice.

      I used to build Frankeneverything, but that's because I worked at a place I shall only call "Chinaman Joe's PC Emporium" to protect the guilty. The owners changed suppliers more frequently than most folks change underwear.

  3. David Robinson 1

    SCSI

    If I remember correctly, you had to sacrifice a chicken over the terminator using a silver-bladed knife with a black pearl handle in order for it to work properly.

    1. Anonymous Coward Silver badge
      Pint

      Re: SCSI

      Correct, as long as it was on the fifth Thursday of the month.

      1. Rich 11
        Terminator

        Re: SCSI

        ...and you've got the right sort of terminator.

        (Possibly even a flesh-coloured one.)

        1. elhvb

          Re: SCSI

          Pah. All of you forgot your book of IT spells...

        2. Down not across

          Re: SCSI

          ...and you've got the right sort of terminator.

          Aside from the necessary sacrifices, termination is nearly always the issue. If at all possible avoid passive terminators, and use active terminators.

          Occasionally cabling (expcially external) is an issue, always beware cables with kinks in them.

          (Don't forget the young goat...)

          1. Will Godfrey Silver badge
            Unhappy

            Re: SCSI

            It's not for nothing it's usually pronounced "scuzzy"

            1. Toastan Buttar

              Re: SCSI

              I always pronounce it that way. That's why "an SCSI" in the headline sounded all wrong to me.

          2. Kiwi
            Mushroom

            Re: SCSI

            and use active terminators

            Oh trust me.. Last time someone wanted me to install a SCSI setup using the cheapest 2nd hand no-name components he could find, I was all for "active termination".

    2. big_D Silver badge

      Re: SCSI

      Best deal ever!

      A friend had bought a PC and stuffed a SCSI controller into it and 3 HDDs, 2 x 40MB Quantum Fireball and a 20MB something else... Only it never worked. He got so frustrated, he ripped it all out and when I was visiting him, he was sticking IDE drives in the PC. I set the slave jumper on the second drive for him and he said I could take the SCSI cr*p with me.

      Looked at it when I got home, set the terminator properly on the SCSI card and it booted straight up! My Cyrix 133 PC absolutely flew with those drives attached.

      1. Anonymous Coward
        Anonymous Coward

        Re: SCSI

        Used to work for a company years ago that had two offices. Head office, beancounters and sales had the nice kit. (Netware 386(?))

        The smaller technical office weren't allowed to buy stuff, so had to make do with what we had laying around, so we had to make do with what was laying around. Despite this we had a network with more functionality than head office - basic stuff like faxing from the desktop! (WfW IIRC)

        Did also build a NT server out of bits and pieces, including a 4x SCSI drives in a RAID 5. Went back a few years later to find that my server was still working away in the corner, but the "proper" server (IBM) had problems with its disk drives that had to be reseated occasionally!

        Gone on a tangent - anyway, had a Franken-desktop with SCSI. Worked fine(ish) with Win98. Tried to upgrade to NT4 and the install program wouldn't work. Long story short - didn't have the termination set on the drive, so no idea how it worked.

    3. MJB7

      Re: SCSI

      Chicken, chicken!?!

      "SCSI is *NOT* magic. There are *fundamental technical reasons* why you have to sacrifice a young goat to your SCSI chain every now and then." - John Wood

      https://odetocode.com/blogs/scott/archive/2004/09/22/scsi-is-not-magic.aspx

    4. Anonymous Coward
      Anonymous Coward

      Re: SCSI

      Never had much problems with SCSI termination, but maybe I just didn't deal with it as much as other people.

      Basic rule - terminate at the end of the cable, not in the middle. If using resistor packs, make sure that pin1 goes in the right hole!

      1. Kiwi
        Coat

        Re: SCSI

        Basic rule - terminate at the end of the cable, not in the middle. If using resistor packs, make sure that pin1 goes in the right hole!

        But what if it's easier to hold the plug upside down? Shouldn't it then go in the left hole?

        Should I be saying things like this with all this talk of "termination" in the air?

        Perhaps I grab my coat now.. It's the kevlar one with the cammo motif...

    5. macjules

      Re: SCSI

      You forgot to add the required sacrifice to the Goat-headed God of the ID Switch. This involved ensuring that certain makes of scanner always had their ID set to the end of the range (7) or (3) if you had another scanner already there. You would try and ensure that an external HD should be the next in line (1) and that removable media such as Optical or Syquest drives were on (2), (4) or (5). Once that was done and you had had the terminator blessed by a defrocked Pagan priestess then you could recite the mantra of the Goat-headed God of the ID Switch ("Please don't screw up in front of my client") and switch on the devices and then the computer.

      1. Tim99 Silver badge
        Unhappy

        Re: SCSI

        No. The rule was don’t mix scanners and disks NOMATTERWHAT the relevant suppliers’ technical documents said. The use of sacrifices and incantations appearing to produce a working system was just the universe screwing with you.

        1. macjules

          Re: SCSI

          Nope, you definitely could mix devices. Many very high-end drum scanners (Crosfield Magnascan etc) required that you had a high capacity external HD to scan to, ideally on the same bus.

          1. Martin an gof Silver badge

            Re: SCSI

            Not had many problems with SCSI, except with my mum's old Mac, where you could connect either the scanner or the external CD burner, but not both at the same time, no matter what IDs they were set to nor which order they were on the chain. Without a terminator (always worth a try) they simply refused to work at all. With a terminator (yes, it was in the right way around - though I have vague memories that one of the devices had a switchable terminator built-in) some horrid burning smells started from one or the other and I never waited long enough after power-on to find out if they worked, for fear of letting the magic smoke loose.

            Individually they were fine and worked well, just never together.

            And these with components bought at the same time from a Mac-specialist who couldn't offer any useful advice.

            M.

            1. SImon Hobson Bronze badge

              Re: SCSI

              In a previous life, I used to supply and support Macs, as well as various Unix boxen - so naturally got to configure plenty of SCSI setups. Strangely I rarely had termination problems.

              One interesting problem I had was when a client wanted a tape backup unit, but we couldn't boot the Mac with it connected. What happened was the tape drive would lock the bus up while it did it's reset - so the Mac would start up, reset teh SCSI bus, then proceed to look for devices. Trouble was, by the time the tape drive was ready, the Mac had given up and in an attempt to contact devices, reset the bus again.

              So for ever, that particular setup needed the tape drive powered on after the system was fully booted. That was particularly annoying as it was in a nice case with room for an additional HD.

          2. davidp231

            Re: SCSI

            See also: Laserwriter IISC

          3. Criggie

            Re: SCSI

            I managed to get two physical hosts on the same scsi bus. It was a P133 running linux and a mac LCII. Both could see the Mac's internal drive and a Syquest EZ135 external drive and a second external Microbpolis 4GB drive.

            But the Mac couldn't see the other controller on ID6, and the PC couldn't see the mac's controller on ID7.

            https://criggie.org.nz/scsi/

            1. Monkey&Typewriter

              Re: SCSI

              Wow that brings back unpleasant memories. Many moons ago I ran MS Exchange on some early NT4 Beowulf clusters like that: server<-->SCSIcable<-->driveArray<-->SCSIcable<-->server with the SCSI adapters providing termination at either end. Both nodes had read-write to the common array with a single lock file as the quorum resource to prevent resource conflicts.

              It ran, but reliability was horrid. I eventually just disconnected one of the cluster nodes and ran on the remaining standalone node... with proper termination of course.

    6. Evil Auditor Silver badge
      Thumb Up

      Re: SCSI

      Damn Dave! You should have told me 30 or so years ago. Stupid me was fiddling with addresses, controllers, drivers and their load sequences - without much success. When all was needed was a bloody chicken sacrifice.

      1. Kiwi
        Pint

        Re: SCSI

        When all was needed was a bloody chicken sacrifice.

        I presided over many a chicken sacrifice to get kit working..

        Usually in the guise of "you cook me a nice roast chicken dinner while I mess around with your kit"

        1. zb
          Happy

          Re: SCSI

          I said that many times but there were never any computers around.

    7. Anonymous Coward
      Anonymous Coward

      Re: SCSI

      > If I remember correctly,

      And you were cool, or at least thought you were) when you referred to "Arnies" rather than terminators.

    8. hmv

      Re: SCSI

      HVD SCSI required goats.

    9. davidp231

      Re: SCSI

      "If I remember correctly, you had to sacrifice a chicken"

      I believe Space Corps Directive 68250 would apply also.

    10. Andy Denton

      Re: SCSI

      Never had a problem (aside from the odd dodgy terminator) with SCSI. A mate of mine who still runs the business we started decades ago told me he retired the last Netware server from one of our customers last year. It was a 486 DX2/66 with 16Mb of RAM, a 2GB SCSI HDD connected via an Adaptec 1542CF SCSI Adapter. It had almost 6 and a half years of uptime showing when he finally shut it down.

  4. richardcox13
    Alert

    Snuggle the zombie server up to the live one, remove the facing side panels and snake a SCSI cable from the working machine to the powered disk cage of the otherwise mostly dead server.

    And it worked? Not possible, any rearrangement would need to include the requisite sacrifice to the Trickster God of SCSI Buses.

  5. John70

    The thing died. Naturally, "well ahead of the replacement schedule."

    Machines do seem to know when their time is coming to an end and decide to accelerate their demise as a final F. you.

    1. Wellyboot Silver badge

      All machines, Back in the '90s I had an ultra reliable Pug-405 diesel (6yr old, 170k miles & serviced every other month) that only had a pair of 3-point seatbelts in the rear. As SWMBO was rapidly growing with sprog#3 I said we'll change car to one with more seatbelts, lo and behold two weeks later on my way to work the engine internals rearranged themselves at 70mph. After fitting a 2nd hand engine, it sold for net £0.00 instead of being an (admittedly low) trade-in.

  6. Sgt_Oddball
    Gimp

    Bless..

    I've got this sort of hackery going on my home server right now. Bought a second 8 drive cage with sas and power cables (already had an expander card fitted to no problems there). Connected the power cable to find its 6 inches shy of its required power (the cables themselves being specc'd to run only one cage off the same cables). Cue old molex to EPS cable torn down for parts and soldered (IT'S SOLDERED GODDAMMIT NOT SODDERED.... *bloody yanks*) in with heatshrink over the splices to prevent shorts.

    Been fine for the last year with that. Oh and the new rechargeable batteries I soldered to the array cache backup. Those where bodged too (I'm not spending £100 on something I can juryrig for a couple of quid and some shrink-wrap).

    Gimp because I'm probably by this point a ducker for punishment.

    1. KarMann Silver badge
      Headmaster

      Re: Bless..

      "(IT'S SOLDERED GODDAMMIT NOT SODDERED.... *bloody yanks*)"

      Hey, these days, I'm just glad when it isn't 'soldiered' or 'solider'.

      1. chuBb.
        Boffin

        Re: Bless..

        Unrelated but the greatest bit of engrish i ever came accross was in a manual for an aliexpress hakko knock off soldering iron.

        "Warning: Man, Child or Dog, could become serious bum, if touch soddering tip" <-- well yes

        1. Anonymous Coward
          Anonymous Coward

          Re: Bless..

          Worked with some one that managed to burn his ear with a soldering iron.

          He was smoking a ciggie, and had the soldering iron in the same hand, and when he went to suck* on it....

          * not a smoker, so not sure of the correct terminology!

          1. Doctor Syntax Silver badge

            Re: Bless..

            "not a smoker, so not sure of the correct terminology"

            I suppose his terminology was quite vigorous.

          2. The Oncoming Scorn Silver badge
            Terminator

            Re: Bless..

            Both myself & the still not quite ex Mrs Oncoming Scorn, worked at a electronics factory (Mentioned elsewhere).

            First Day H&S guy would tell of a story about an assembler who liked to to test her soldering iron was hot enough by holding it just inside her mouth which worked fine until the day she was startled & burnt the inside of her mouth.

            Myself, working in a hotel bedroom at a Trade Show (Hammersmith Novotel & still incredibly drunk\hung over) getting C64 cartridges to work at a trade show, soldering iron plug left unplugged (Having sobered up enough to twig that the units we had collected the previous night had been assembled in the cases the wrong way around). I left the room so the cleaners could remake the beds\deal with the bathroom (That part of the blame was the Chilli eaten at the Chieveley Services on the M4 the day before) & allowing me to get some food.

            On my return I picked up the iron by the shaft to discover (After a few seconds lag - Still hungover) the cleaner had also unplugged the mini bar for the hoover & plugged the iron back in on completion. Burnt human flesh really does smell like pork.

            1. Anonymous Coward
              Anonymous Coward

              Re: Bless..

              I worked with someone who burnt her lip with a soldering iron. The iron got caught on the holder, and when she pulled it free, she had a bit too much follow through.

            2. Olivier2553

              Re: Bless..

              I recently burned my lip with an hot lighter: after lighting up tens of incense stick for some ceremony, the lighter was hot and I did not want to put it back in my pocket so I started blowing on it. Then I tested it on my lip to see if it was cool enough, it was not.

              The picture illustrating the article also reminds me a anecdote during last Chinese new year: we were about to perform when one of my friend had the sole of his trainers starting to peel of. No, I had no glue, no tape, nothing and it was too late to try to find a solution. All I had was an handful of bandages so he had to do with a Frankenshoe.

              No IT angle on that I am afraid as this is part of my life completely devoid of computer stuff.

        2. Antron Argaiv Silver badge
          Thumb Up

          Re: Bless..

          Part of my training regimen for the new interns, that is:

          "Always pick up the soldering iron from the end with the cord, never from the other end."

          Thankfully, I no longer need to give them a mnemonic to remember the resistor color code :-)

          (good thing, as many of them are now female)

          1. Alan Brown Silver badge

            Re: Bless..

            "Always pick up the soldering iron from the end with the cord, never from the other end."

            When I was _MUCH_ younger, I was working in my "shed" (a space at the end of the garage) and the soldering iron slipped off the bench for some reason - not really thinking and only seeing the movement out of the corner of my eye, I reached out - and caught it by the hot end.

            An Antex X25 leaves a lovely linear burn across the palm that takes weeks to heal properly and can take a lot of explaining when you're 14

        3. ma1010
          Alien

          Re: Bless..

          ALL YOUR BUM ARE BELONG TO US!

          1. Kiwi
            Coffee/keyboard

            Re: Bless..

            ALL YOUR BUM ARE BELONG TO US!

            Do you know how hard I needed to fight just to limit my comment to this???

          2. chuBb.

            Re: Bless..

            Move zig

    2. chuBb.
      Pint

      Re: Bless..

      Missed a trick, while i commend your hands on approach to soldering your own battery packs, it sounds like you neglected to put an inline connector on the leads makes life so much easier when you can just unplug and go...

      Have a pint regardless, just dont drink it; its isopropyl to clean the flux off the PCB ;-)

      1. Sgt_Oddball

        Re: Bless..

        I reused the original charging circuit so it comes with inline connectors anyway.

    3. This post has been deleted by its author

      1. Dale 3

        Re: Bless..

        Thanks for the video. It explains Oddball's righteous indignation at SODDERING!

        1. John Brown (no body) Silver badge

          Re: Bless..

          Yeah, especially for a US product. A soddered BUTT connector. Surely someone in marketing must've spotted that faux pas and silently sniggered to themselves.

          As for the product itself, I could see a small market for those people who might need to join wires once or twice in their lives, but it strikes me as a dumbing down for the masses who want instant gratification without the learning curve,

      2. Sgt_Oddball

        Re: Bless..

        I'll give them credit for the water seals at either end but usually I just go with good old fashioned electrical tape on the connection to waterproof it.

    4. Sequin

      Re: Bless..

      Don't forget that you must use juicy lead based solder - none of that naff ROHS lead free junk!

      1. Martin an gof Silver badge

        Re: Bless..

        The Lead-free stuff is ok (ish) for electronic "hobby" soldering if the following conditions are met:

        • you use a good quality solder - there are several different formulations of "Lead free" out there, all of which work in subtly different ways. In particular while Tin-Lead is properly eutetic, Lead-free isn't (quite) in my experience. PbSn will suddenly and quickly solidify, but Lead-free seems to take a bit longer, therefore
        • be very careful not to move any components while the melted solder is cooling
        • you have an iron that can get hot enough - so probably 50W as a minimum and temperature-controlled. Lead-free needs quite a lot more heat than Lead solder, more so than it appears from the specification on the label
        • never mix Lead and Lead-free solder, even to the extent of swapping soldering iron tips if possible. It won't always obviously "fail", but the joint is never as good if you have one type contaminated with the other so it pays to have a good idea what was previously used if re-working something

        Currently having fun soldering Copper water pipes with Lead-free. Lots and lots of flux and oodles of heat from a really good quality blow torch, seems to work well...

        M.

        1. Kiwi
          Mushroom

          Re: Bless..

          Currently having fun soldering Copper water pipes with Lead-free. Lots and lots of flux and oodles of heat from a really good quality blow torch, seems to work well...

          Would that stuff ever be allowed on gas pipes?

          I've used up my swear quotient for the day so I won't describe what I think of the majority of the lead-free solder I've seen (if anything earned the term 'SODder...)

          When they started using it in mobos, about 2 years later we started seeing lots of failed boards where you could pin it down to a 'dry joint" in the form of a cracked joint. For whatever reason, the (IIRC zinc-based) stuff would go brittle fairly quickly, leading to failures.

          Not just mobos either, but TVs, audio gear, and probably all sorts of other consumer electronics. So much stuff destined for the landfill years earlier than it should've been. Had lead solder bee used, many of these devices could've seen more than 10 years (some more than 20) of faithful service.

          Given the number of components that'd have to be removed, "reflowing" was generally far more expensive than the cost of replacing.

          Another ill-conceived or ill-implemented 'pollution solution' that made things worse, not better.

          1. chuBb.

            Re: Bless..

            Main cause of the rrod on xbox 360 lead free bga pellets going brittle and failing due to thermal shock (same with ps3, bodged enough back to life with a paint stripper gun and some bent bits of steel to act as nozzles on the hot end, reflowed after approx 35 secs at full pelt heat low fan)

      2. Anonymous Coward
        Anonymous Coward

        Re: Bless..

        That sounds like a Big Clive quote.

    5. Kiwi
      Pint

      Re: Bless..

      (IT'S SOLDERED GODDAMMIT NOT SODDERED.... *bloody yanks*)

      Hear hear!

  7. -tim
    Coat

    SCSI Hacks?

    A friend had an Amiga with two SCSI cards. There was a program on the Amiga that looked like a block device to a PC that thought it was the "master" on the chain. The result is he could rapidly backup and reboot the PC much the way the virtualization allows today as well as doing RAID0 style merging of small Inexpensive (aka rejected from other projects) disks. When he ended up with a Sparcstation, it too went on the SCSI chain and it allowed him to work on the same files with the PC or Sun without all those expensive network cards as long as he remembered to only have one system active at a time.

  8. Huw D

    I'm now curious as to whether "Bobby" was working for an Accountancy firm in the early 90s as I'm sure that I recall one of the regional IT Managers telling a story like this.

  9. Jaspa

    IDE not SCSI but the hack worked

    Many moons ago, in a Woolworths not far away I was presented with a dead EPOS Server smack in the middle of Christmas week.

    Our "Senior" Eng had requested I assist by bringing a spare IDE cable as his "didn't fit" and it was hampering the tape restore process on the ageing 386.

    The Manager was impressed that we turned up mob handed in his hour of need but was becoming increasingly frustrated by the lack of progress and the queues of Shoppers waiting to pay for Pick n' Mix and the latest CD single releases .

    We were faced with having only the newer cables with the unused pin blanked out. The kit was of an age before these so Senior's didn't fit was accurate.

    Senior is calling anyone and everyone in the hope of locating a cable not too many miles away while I'm trying to keep the manager placated.

    We were stood in the store right next to the sewing section and in a flash I had brainwave. Inviting the Manager out for a long overdue cig I asked if I could borrow a sewing needle as we left the store. Spark up said cig, heat needle and melt the offending offending plastic away. Hey presto, one working cable.

    Senior went on to watch paint dry as the restore worked its magic and I wandered off feeling smug .

    For those not of a chronological disposition ...

    http://www.mikeshardware.com/howtos/howto_connect_ide_hd.html

    1. Anonymous Coward
      Anonymous Coward

      Re: IDE not SCSI but the hack worked

      I would have got a pair of cutters and cut off the pin!

      1. A.P. Veening Silver badge

        Re: IDE not SCSI but the hack worked

        That would work also, but lacks finesse.

        1. vogon00

          Re: IDE not SCSI but the hack worked

          Ahhhhh... the finesse here was getting the overdue ciggie...sorting out the cable was just a bonus :-)

      2. The Oncoming Scorn Silver badge
        Paris Hilton

        Re: IDE not SCSI but the hack worked

        In the stress of the moment, I'd be concerned about cutting the wrong pin, voiding warranty & causing a new problem.

        Easier to mod the cable.

      3. Jaspa

        Re: IDE not SCSI but the hack worked

        Not wanting to risk further damage to the already lacking DR kit I erred on the side of caution.

    2. Patrician

      Re: IDE not SCSI but the hack worked

      Yep, been forced to do that quite a few times in years gone by.

    3. John Brown (no body) Silver badge

      Re: IDE not SCSI but the hack worked

      "Spark up said cig, heat needle and melt the offending offending plastic away. Hey presto, one working cable."

      My solution was to poke through the plastic blank with my smallest jewelers screwdriver. I didn't especially think of it as a clever hack, just getting the job done by hook or by crook. But then I worked for a cheapskate company and bodgjing stuff was just the way we had to work due to lack of proper kit and parts. It wasn't unusual to leave site with a PC sporting two multi-I/O cards because eg the IDE controller had failed on the original and I had a "spare" with a failed serial port. A quick play with jumpers to enable/disable the relevant parts and the customer had a working machine. Since I didn't like bodging and I knew the order for a fully working replacement would never complete, I would usually enable the "new" card such that the customer got an extra printer port and/or serial ports, which helped salve my conscience somewhat and often pleased the user.

    4. EVP

      Re: IDE not SCSI but the hack worked

      Man, that summons up flashbacks from the darkest dungeons of my memory. Burning black plastic of female IDE connector didn’t smell good. I can still bring back its sharp and toxic malodour. The fumes made your eyes to water, too.

      Unless my memory fails me badly, some cables had on pin 19 a miniature plug which could be fished out.

    5. Kiwi
      Pirate

      Re: IDE not SCSI but the hack worked

      Spark up said cig, heat needle and melt the offending offending plastic away.

      Done that a few times :)

      Also learned that the plastic covering the hole was often a thin film that could be punctured by the pin with sufficient force, maybe a little starting encouragement in the manner you describe...

      --> Closest to sharp pointy things we have 'ere.

  10. Chris King

    This one obviously needed a chalk pentagram in the computer room...

    My predecessor in a previous job set up a real nasty Frankenserver...

    Not one, not two, but THREE SCSI controllers - one for disks, one for a CD-ROM and one for a tape drive.

    The one with the tape drive wasn't properly configured, so any attempt to make a tape backup would result in the machine locking up solid.

    There was a sound card in there for some reason as well *boggle*.

    Oh, and all this lot (complete with long SCSI cables) was sat in a ATX desktop case with two 5.25" Micropolis SCSI disks - so it got pretty toasty.

    Did I mention this was the site's primary DNS, DHCP and web server ?

    1. Doctor Syntax Silver badge

      Re: This one obviously needed a chalk pentagram in the computer room...

      Micropolis. There's a name from the past. Not that I was ever aware of using any.

      1. Chris King
        Mushroom

        Re: This one obviously needed a chalk pentagram in the computer room...

        If you never felt the searing, dry heat like that of the sun itself, you probably never got that close to a Micropolis disk. (Or a Quantum Fireball for that matter)

        1. Down not across

          Re: This one obviously needed a chalk pentagram in the computer room...

          If you never felt the searing, dry heat like that of the sun itself, you probably never got that close to a Micropolis disk.

          They did run rather hot. May have been contributing factor for them suffering from seized bearings requiring switch kick/"tap" with a mallet to restart them if they cooled down (say you needed to move a machine(.

    2. joshua.wells

      Re: This one obviously needed a chalk pentagram in the computer room...

      “There was a sound card in there for some reason as well *boggle*.”

      Back in the year 2000, I was hired as a tech at a small private school. They were sold a poorly implemented Citrix solution by a well-know local tech firm as a solution to not upgrading their desktops. When they couldn’t get sound out of the desktops, their solution was to put sound cards in the Citrix servers, because the desktops were running sessions off the server, and if the server didn’t have sound, how could the desktops?

    3. hmv

      Re: This one obviously needed a chalk pentagram in the computer room...

      My memory of it is somewhat distant - I was a PFY at the time and now I'm getting to be a grumpy old fart. But I recall an ancient Sun SS20 with something like 5 SCSI controllers each with an insane tangle of leads coming out to a mixture of external disk drives, tape drives, and one CD drive (which wasn't really needed).

      I was tasked with moving the damn thing from one side of the DC to the other. Spent an afternoon labelling everything up before unplugging everything, moving it, and (hopefully) plugging everything back. Seemed to power up okay :)

    4. Alan Brown Silver badge

      Re: This one obviously needed a chalk pentagram in the computer room...

      "Not one, not two, but THREE SCSI controllers - one for disks, one for a CD-ROM and one for a tape drive."

      GIven the utter shittiness of a lot of pc-grade scsi tape and cdrom drives, that may have been a wise choice - your experience seems to bear that out - they really didn't play nice with others

      And a hell of a lot of scanners assumed they were the only device on the bus

      1. Nick Ryan Silver badge
        Stop

        Re: This one obviously needed a chalk pentagram in the computer room...

        All this reminiscing is bringing back a lot of repressed memories. You bastards.

        Next we'll start hankering after the glory days of needing a sheet of graph paper to work out how to combine various bits of PC hardware with their individual IRQ options into one workable configuration. I probably still have a bag of jumpers from this time. Prior to the similar juggle of loading DOS drivers such that there is enough base memory left, and suitable EMS or XMS facilities for the PC to be able do anything remotely useful after starting up.

        1. Anonymous Coward
          Anonymous Coward

          Re: This one obviously needed a chalk pentagram in the computer room...

          Ah, IRQs. That one got me into trouble...

          When I was a kid, we had a 486DX33 with Win3.1 on it. My brother and I really wanted something better than PC speaker sound, so we bought a sound card and popped it in the case. Upon bootup, Win3.1 played that lovely startup noise over the speakers. Easy success! Ok, let's try a game, double-click the icon - what do you mean, file not found? Hmm. Try another - same thing. Uh-oh. Ok, File Explorer to take a look at... wait, File Explorer not found?

          In our youthful, inexperienced exuberance, we didn't know about IRQs, and how important it was to not duplicate them, like using IRQ5 for both the sound card and the hard drive. The Windows startup noise scrambled the c:/windows directory. We frankensteined together a half-functional Win3.1 installation, but our parents were pissed. My dad ended up installing Win95 on it instead.

          For that machine's fiery demise, look at my other comment here posted a few minutes ago...

  11. Anonymous South African Coward Bronze badge

    Had a Frankenserver as well.

    For backup storage (like a NAS) using Server2003. Yes, it was a long time ago.

    It performed the job well, but near the end of its lifetime its SAS RAID card was corrupting data, so I chucked the whole thing.

  12. Anonymous Cowerd
    FAIL

    "an" SCSI cable?

    We all know it's "a" SCSI cable.

    1. Killfalcon Silver badge
      Headmaster

      Re: "an" SCSI cable?

      Depends how you pronounce it. You'd use "an" if the word starts with a vowel sound, but "an" for a consonant sound even if the letter is different. See "a united front" for example, or "an SSD drive".

      So if you say "Skuzzy cable" it's "a" as you lead with a consonant sound. If you say "Ess-Cee-Ess-Eye" then you're wrong.

      1. The First Dave

        Re: "an" SCSI cable?

        What do you mean: "if you say skuzzy" ? That is the only way to say it.

        1. Anonymous Coward
          Anonymous Coward

          Re: "an" SCSI cable?

          "That is the only way to say it."

          Well, to be fair, there's usually an expletive before the 'S'.

          1. EVP

            Re: "an" SCSI cable?

            Thank you for good laughs, I grant you on my own authority title of the wittiest comment of the day!

      2. John Brown (no body) Silver badge
        Headmaster

        Re: "an" SCSI cable?

        "You'd use "an" if the word starts with a vowel sound, but "an" for a consonant sound"

        You have one too many n's there.

        1. Killfalcon Silver badge

          Re: "an" SCSI cable?

          ...dammit.

  13. Anonymous Coward
    Anonymous Coward

    I could tell a story about a machine with many SCSI-to-IDE adapters...

    ... but I still have nightmares thinking about it. Maybe another time....

    1. Kiwi
      Pint

      Re: I could tell a story about a machine with many SCSI-to-IDE adapters...

      ... but I still have nightmares thinking about it. Maybe another time....

      Maybe.. But first, please have many of these -->. At least enough till the memory fades.

      And then... Er, what were we talking about again???

      (There's some things I just don't want to know, and if I have to squander the life savings buying someone enough booze so they forget.....)

  14. Anonymous Coward
    Anonymous Coward

    Frankensolutions

    They always stay for a LOOOONG time.

    The last frankensolution I had the privilege to hit my nose against was an AS400 dual DCs solution.

    Setup: 2 DCs, each with one set of servers and one disk array. Classic.

    We never could figure out how data replication would work between active and passive DCs. The system

    team would tell us it's the storage array and the storage array would tell us there is no such replication.

    And of course, documentation was lost in the mist of time ... And we didn't have AS400 skills allowing us to

    dig into the OS. Only an old local team had and the spoke only french anyway.

    End of the day, we had to move the passive SD and of course stop the whole thing (storage + systems), again,

    PASSIVE side. It turned out that:

    - storage replication was host-based

    - it was one-way only (you switch to passive ? Then you stay on passive ... forever

    - it was set up so as to stop the ACTIVE if passive was off. Nice DRP you have, here !

    We had to speed up like crazy the move so as to restore service on the active ...

    ""We ended up running that 2-servers-in-1 for about nine months before we were able to finish the migration."

    Lucky you, then !

  15. SonofRojBlake

    Frankensolutions

    Tourist here - chemical engineer, not IT. I can tell you Frankensolutions are distressingly common in the chemical industry. They usually follow a script:

    PHB: "Can the plant do X?"

    Engineer: "MMmmmm... yes. If we can get some of that stuff over there to here. I'll just draw up the..."

    PHB: "Just run the pipe from there to there and see if it works, we (i.e. YOU) can engineer in the elegant solution afterwards."

    Engineer: "Well... OK."

    ...

    Engineer: "It works, kind of, in the temporary configuration, with the valve operated manually instead of by the DCS and with the pipework connection hanging in midair. I've drawn up a design, control philosophy and project plan and a budget for making the alteration permanent. Could you sign here?"

    PHB (who only heard the first two words of the previous paragraph): "Good. Can the plant do Y?"

    Rinse and repeat until the place looks like a plate of spaghetti and the only people who can remember why have retired.

    1. H in The Hague

      Re: Frankensolutions

      "Engineer: "MMmmmm... yes. If we can get some of that stuff over there to here. I'll just draw up the...""

      That worked so well at Flixborough :(

      1. SonofRojBlake

        Re: Frankensolutions

        IIRC the problem there was that there WAS no "I'll just draw up the...". They didn't "draw up" anything, unless you count literally marking out a rough outline in chalk on the floor of the engineering workshop.

        That incident invented modern modification control procedures. It's disappointing that newer chemical industry graduates I've met have never heard of it.

    2. imanidiot Silver badge
      Mushroom

      Re: Frankensolutions

      Rinse and repeat until someone forgets to close a valve and 3 square miles turn into a superfund site, surely?

      --> The not entirely uncommon result of chemical plant SNAFU's -->

      1. The Oncoming Scorn Silver badge
        Mushroom

        Re: Frankensolutions

        & not to dissimilar from Pharmacuetical Chemical engineers from the sound of it, but at least you didn't bring a UXB that was dredged up with sand for "Infill" onto the site.

    3. Loyal Commenter Silver badge

      Re: Frankensolutions

      You don't work for Union Carbide do you?

    4. EVP

      Re: Frankensolutions

      Ah, that finally explains me why chemical plans look how they do.

      Not that my piece of mind has increased...

      1. SonofRojBlake

        Re: Flixborough

        I'm old enough that my university course went heavily in on Flixborough and other similar incidents. What's really distressing is that I mentioned the name to a recent (<5 years ago) chem eng graduate and they'd literally never heard of it. I felt like telling him to get off my lawn.

        I elided the bit where the PHB's request is put through design safety checks so Flixborough doesn't repeat. The issue - *my* issue - is that bit where you get the safe proof-of-concept design working, and that just becomes the production plant.

        @EVP - another reason chemical plants look like they do: poor attention to detail in construction contract negotiation, restricted budget and unrealistic timescales. This leads to:

        - no money to build a proper 3D model of the plant

        - no time to prepare isometric drawings

        - a piping contractor who is paid by the number of welds they do when site running pipework.

        This can lead to some... "creative" pipe routing, requiring many more bends and supports than you might have imagined are necessary.

        1. Kiwi
          Devil

          Re: Flixborough

          This can lead to some... "creative" pipe routing, requiring many more bends and supports than you might have imagined are necessary.

          Thanks for the memory of one of the more dishonest times in my life :(

          I often was asked to give tours of one of the plants I was at. I often suspected this was the reason why some of the piping had more bends than would be necessary.. If asked I often mumbled something about old pipework/structures/other stuff that had been there when the pipe was made but was since removed, and we'd have to shut down for a while to straighten the pipe.

          At least now I can be more sure that yes, it was built in the common fashion of government-funded sites - a part of Muldoon's "think big" projects - though one not widely publicised as such.. I'll leave it to the imagination as to why it wasn't trumpeted as a government success...

    5. Kiwi
      Pint

      Re: Frankensolutions

      PHB: "Can the plant do X?"

      Engineer: "MMmmmm... yes. If we can get some of that stuff over there to here. I'll just draw up the..."

      PHB: "Just run the pipe from there to there and see if it works, we (i.e. YOU) can engineer in the elegant solution afterwards."

      I think there's a very good chance we know some people in common.

      I won't say where... There are some things that should never again see the light of day, or be mentioned in polite society or even on El Reg..

  16. deadlockvictim

    SCSI

    Well, I, for one, am a big fan of SCSI, partly because I never had to support it professionally. I was the friend of the manager's/owner's daughter who knew about these things and generally the problem was either duplicate SCSI IDs, either taking off or putting on a terminator or unplugging the SCSI chain and putting the various devices back on again. Now I never dappled in the hard stuff (servers), just soft stuff like Macs and their often well-behaving scanners, CD-ROM, hard-drives and zip-drives.

    What I really liked about SCSI was the universality of it.

    First of all you had a bus that was as fast externally as it was internally. SATA & eSATA have this property too.

    Secondly, you could attach all manner of devices to it. Indeed almost all peripherals (except printers) were available with SCSI ports that could be plugged into the computer.

    Thirdly, they often (or usually) had a driver inside in the SCSI bus. This meant that SCSI devices were not platform specific.

    Fourthly, SCSI lasted a long time (1985-2005) and there was a lot of backwards-compatibility. It also meant lots of adapters and knowing what all of the terms meant. It meant that you could stick in a 300GB U320 server HD into a lowly Macintosh SE and have obscene amounts of storage space that made for mindboggling default file sizes in HFS (Hierarchical File System — Macintosh's file system from 1985 to 1998. It has 64K allocation blocks...). It also meant that you could plug in a SCSI device from 1985 into a machine from 2002 and use it.

    1. Elfoad Regfoad

      Re: SCSI

      > Now I never dappled in the hard stuff (servers), just soft stuff like Macs and their often well-behaving scanners, CD-ROM, hard-drives and zip-drives.

      In the mid-90s, a friend had a Mac with an internal hard drive and a printer and connected via SCSI that was experiencing intermittent hard drive failures. She brought the Mac to the a local shop and it tested fine. She went home and the problems returned. She went back to the store with her Mac and this time she brought printer (but not the scsi cable for the printer). Using the shop's SCSI cable for the printer it all worked at the shop, but when she went home, it still was having problem. Once again she went to back to the shop with the Mac, the printer, and this time she brought the printer cable. At the shop, they were able to verify that there was a problem with the hard drive. It turns out that one of her cats had nibbled on the scsi cable leading to the printer. Replacing the printer cable with one that didn't have tooth marks fixed the hard drive problems.

    2. Donn Bly

      Re: SCSI

      Secondly, you could attach all manner of devices to it. Indeed almost all peripherals (except printers) were available with SCSI ports that could be plugged into the computer.

      Actually, printers were available too, especially lasers. On the low end you had the LaserWriter SC but on the high end there were a number of large lasers printers.

      1. J. Cook Silver badge

        Re: SCSI

        Yep. I knew a lady that did short print runs that had one, because none of the printing companies around her at the time would take such a small job...

  17. Anonymous Coward
    Anonymous Coward

    I know I did this.....

    ...but I can't for the life of me remember why.

    I had to get the data off a production server and do a temporary smoke-and-mirrors job before restoring it all again. But we had no suitably large drives available and I recall we had to be quick. I ended up having to hit the man page of the mdadm command, and strung together a series of USB backup disks (this must have been around 2000, maybe 2002) To my astonishment, a suitable large, and most importantly, stable, disk array appeared and we got away with it. It must have been en embarrassing incident if I've blotted it out of my memory so effectively. Production continued on that string of disks until we put the main server back together again.

    1. Anonymous Coward
      Anonymous Coward

      Re: I know I did this.....

      In a past life we found that our disk monitoring alerts hadn't being going to our admins, with the result that we'd got various mdraid failures across the estate and, horror of horrors, one box with multiple partitions showing faults.

      Our admins wanted to throw the lot away and restore all our users from backup (a lot can happen in 24 hours), to which I said "fuck that".

      That was the day I blessed the foresight/whateverism of the mdadm developers and found I could extend RAID1 partitions to 3 drives instead of just 2, putting in a new drive and syncing the various partitions off whichever disk was still error-free and active for that location.

      One partition, though up, wasn't syncable due to errors on the remaining active drive; thankfully this was /var and with a quick bit of tar abuse I'd got everything bar a single irrelevant logfile onto the brand new disk.

      Mirrored -that- again, and with a couple of false starts getting the right root drive set in cramfs I was able to get things solidly functional with minimal downtime.

      I was -very- happy with myself for that.

      ...and never want to be in that situation ever again.

  18. redwine

    An SCSI?

    Who on Earth pronounces each letter?

    1. Loyal Commenter Silver badge

      Re: An SCSI?

      Presumably, the same people who spell out "http://www." when they tell you a web address...

    2. TomPhan
      Devil

      Depends on how much you want to wind up someone.

    3. vogon00

      Re: An SCSI?

      Depends on who you are speaking to.

      I was discussing some new infrastructure with a previous boss a long while ago. I used the my (actually *the*) usual vernacular term, to be met with the reply of 'Don't do that, put decent disks in instead'.

      I sh*t you not.

    4. Olivier2553

      Re: An SCSI?

      In French, I do. In English it is more "skassi".

  19. sawatts

    HOWTO Move Your Server

    Back in the late 80s or early 90s now, at a MOD R&D Establishment. Back when things were fun...

    A new graduate decided he needed to move a server/workstation from one side of the room to the other. However he discovered that it was running and he didn't have the password. He was at least wise enough not to just pull the plug on it.

    He didn't pull the plug.

    He somehow used bare ended cables to keep the plug fed with power while it was moved from wall outlet to extension, and trundled the machine across the room.

    Amazingly this appeared to work and, for once, no one died horribly - but someone had to explain to him all about three-phase power.

    1. vulture65537

      Re: HOWTO Move Your Server

      At Uni I was reading a book which covered three-phase power (it turns out to be important for X-ray machines) and found in the book a warning that if the fuse blows you must not replace it with a nail.

      1. H in The Hague

        Re: HOWTO Move Your Server

        "... found in the book a warning that if the fuse blows you must not replace it with a nail."

        Sound advice - a bolt works much better (say M12 or 1/2"). So I've been told :)

        1. Anonymous Coward
          Anonymous Coward

          Re: HOWTO Move Your Server

          My guess is that they are assuming the engineer maybe unaware of the size of said nail of fuse in relation to typical loads leading to overheating and premature failure.

          Given a large enough stock and range of bolts and nails, as any competent engineer should be expected to have "available"*, a correctly sized nail or bolt will do in a pinch.

          Note: when acquiring a suitable part from your "available" stock, try and avoid taking load bearing components. Unless its a load bearing component and then just ensure you have considered the consequences...

        2. EVP

          Re: HOWTO Move Your Server

          I remember (shame on me) MacGyver using a tin foil covered chewing gum wrapper for the same purpose once to get out of a tricky situation.

          Did he bad, yes? I want to believe that he was conscientious enough and replaced his kludge with a proper bold after dealing with the crooks (and leaving behind the misty-eyed lady he saved).

        3. ICPurvis47
          Joke

          Re: HOWTO Move Your Server

          Yes, and the thread acts as a series of cooling fins as well.

      2. Dave K

        Re: HOWTO Move Your Server

        That's not a nail, its just a special fuse with a 200A rating...

        1. A.P. Veening Silver badge

          Re: HOWTO Move Your Server

          That's not a nail, its just a special fuse with a 2KA rating...

          FTFY ;)

      3. Grumpy Scouse Git

        Re: HOWTO Move Your Server

        Back when I was a boy engineer, I heard a tale from the older guys of a machine that kept blowing a half amp fuse, irrespective of which part had been replaced to try and fix the machine. Solution was to replace half amp fuse with the biggest one they could find, turn the machine on and stand well back until the actual broken part began to smoke!

        1. Anonymous Coward
          Anonymous Coward

          Re: HOWTO Move Your Server

          These days we have thermal cameras. But the procedure can be much the same!

        2. Kiwi
          Mushroom

          Re: HOWTO Move Your Server

          Solution was to replace half amp fuse with the biggest one they could find, turn the machine on and stand well back until the actual broken part began to smoke!

          When I was doing my apprenticeship that was a fairly common diagnoses method.

          Of course, sometimes the actual faulty part wasn't the one smoking/glowing (ever seen a tiny little signal diode handling more juice than it wants? I think I know how super-bright LEDs were invented!), but a nearby part drawing on (or in the case of a tranny, putting out) more A (or V) than was considered ideal for the circuit involved.. (And in fire-fighting parlance, some of them became "well involved").

      4. Alan Brown Silver badge

        Re: HOWTO Move Your Server

        " found in the book a warning that if the fuse blows you must not replace it with a nail."

        No warnings about making especially sure not to hold onto said nail?

        We jest about this, but I've seen the results of what happened when someone tried to use a nail in a ceramic fuse holder - it stuck out the ends and the inevitable happened when he whacked it back into the panel without bothering to turn the supply main off first

        Luckily for him he was a poor conductor and survived but one of the side effects of being well and truly zapped is temporary loss of control of bodily functions - including sphincters.

        1. EVP

          Re: HOWTO Move Your Server

          I call your story a crash course on dangers of electricity. He, or anyone else attending to the course, won’t try it again.

          Seriously, good that he survived. This case was probably something that happened impulsively. Every now and then you see death-trap installations by “electricians” which require serious planning and trouble to build. Why do it properly, when you can kill yourself, or someone else?

        2. Robert Carnegie Silver badge

          Re: HOWTO Move Your Server

          Some "co workers" need to be led by the hand. Some really should not.

    2. ICPurvis47
      Boffin

      Re: HOWTO Move Your Server

      When my eldest daughter was studying Electrical Engineering in sixth form, she spent the Work Experience section of the course at the local Electrical Manufacturing company, working on synthesizing Variable Voltage, Variable Frequency (VVVF) 3 Phase output from a fixed frequency single phase supply, using IGBTs. When she went back to school, she was told that, as far as the curriculum went, three phase did not exist, and that she was to forget all she had learnt in industry because she would fail her GCSE exams if she didn't. She went on to take Electrical Engineering at University.

  20. theblackhand

    IDE in servers

    <random dump of useless information>

    Before PCI-bus IDE HBA's, ISA IDE HBA's required an interrupt per channel which limited the number of hard drives you could use (you could usually manage two-channels of 2 drives each once you added a few network cards) and generally multiple hard drives were the only way of expanding capacity in the days where the biggest hard drives weren't big enough. From memory, the Netware IDE drivers weren't great either although that may have been a hardware bottleneck rather than just a driver issue.

    SCSI HBA's allowed upto 14 drives per HBA...more than enough.. And the drivers were better. If you could get updates.

    </random dump of useless information>

    1. Kiwi
      Boffin

      Re: IDE in servers

      (you could usually manage two-channels of 2 drives each once you added a few network cards)

      I remember showing a "computer expert" friend my BBS machine once.. Inside was 2x2IDE drives (2HDDs on one and 2CD drives on the other, one a 4 or 6 CD caddy), and a further 5 or 6 SCSI drives on there.

      He as adamant some fakery must be going on because he'd never been introduced to SCSI and as far as he was concerned a PC could only ever manage 2 HDDs. When I started in computing he taught me all he knew.. Strangely, within a couple of months I knew far more than he did, despite his self-promotion as "one of Wellington's pre-eminent experts".

      Know any good baggage handlers? I see I have some more junk to get rid of!...

  21. vulture65537

    Long ago somewhere in this galaxy there may have been someone who tinkered with the inside of a PC so much he got to leaving the screws undone and the case open a few centimetres. Then there was contact from a near neighbour (who had not seen indoors at this place) complaining "You're running a PC with the case off which interferes with TV signal and is illegal". The case went back on and there was no more complaint.

    1. Anonymous Coward
      Anonymous Coward

      Move along ..... nothing to see here !!!

      vulture65537,

      Most of my kit at home has *most* of its casing open, most of the time !!! :)

      It saves so much time when you are reconfiguring things or 'trying things out'.

      I also have 10 hard disks running in one Server case from its powersupply with the sas cables plugged into a contoller in another Xeon Server with lots of memory, result 'Franken VMware Server'.

      Works just fine and the Disks are so much cooler being in their own case without motherboard + CPU etc.

      :)

      1. phuzz Silver badge

        Re: Move along ..... nothing to see here !!!

        I don't think the side of my computer has ever been screwed shut all the way.

        I might give one of the thumbscrews a half turn to stop it rattling, but that's the limit.

        1. Kiwi

          Re: Move along ..... nothing to see here !!!

          I don't think the side of my computer has ever been screwed shut all the way.

          I've seen wooden beer crates (and other boxes) used as cases before. One person used their glove box (back in the very early days of making an MP3 player 'portable'), and another had done some fancy work with mounting mobo's under the seat in a minivan, LCD screens in the backs of seats - enough kit the kids could do some network Doom/Quake on long trips.

          Once you get the traditional closed beige (or black or white or eggshell or...) box out of your head and start to think "where else can this go?" you can come up with all sorts of interesting ideas. Just needs enough airflow in the right places for cooling...

  22. Eric Kimminau TREG

    Upsidedown Folio

    In 1995 I became the webmaster for a very colorful computer company based in Silicon Valley. On my first trip out to meet the team and see the servers and infrastructure I would be responsible for I sat down with the senior systems administrator and asked "what kind of issues have been occurring and what kinds of changes would he like to see made to make his life better and our sites more reliable?"

    His response surprised me. "Well, the first thing I would do would be to repair "upsidedownfolio" and do some general cleanup." Whatthehellwasupsidedownfolio?

    Turns out the primary support website for the entire company was called Supportfolio. It was apparently hurriedly deployed on a new server a couple of years previously and when it was racked, it was racked upside down. Whenever they had to work on it they would slide it out on its rails and had to work on it from underneath. Ergo, the server was racked upside down. It was apparently the joke of the administration team and it became a rite of passage for the new guy to get assigned to work on the system if it ever had an issue. WTF!

    Needless to say we scheduled a change for the following weekend and I personally flipped the server over into correct orientation.

    1. Robert Carnegie Silver badge

      Re: Upsidedown Folio

      I would upvote twice if (1) I could and (2) the story had ended with the server turned over and promptly not booting until turned upside down again. I know that isn't what happened, but... wasn't that your backout plan?

      By the way, am I misreading the story that it was YOUR decision that the upside-down server should be not upside down any more? Your colleague wanted to repair it... wanted YOU to repair it... while lying on your back??

      (Did they have one of those body-sized skate board things?)

  23. Anonymous Coward
    Anonymous Coward

    Sounds Familiar...

    This story reads remarkably like an adventure of mine at a large "document management" company in the 90's.

    We had Northgate servers (remember them?) with EISA BUSs running Netware 3. The server in question had dual EISA SCSI cards with the drives, if I remember correctly, in a RAID1 configuration. It was tasked with being the main software repository for a very large and important project.

    It was "protected" by a low-end, consumer-level, desktop UPS. I wasn't the one who built this disaster in waiting, but, inherited it when their full time SysAdmin went on maternity leave and never returned. As I normally do, I wrote down what I thought were areas that needed attention and gave it to management. Who promptly told me I didn't know what I was talking about and, besides, they couldn't afford to address what they said were non-issues.

    One night we had a nasty rainstorm with the accompanying power outages. I came in to find the server dead. Literally. Mains power had been up and down all night draining the UPS. With no battery available the UPS happily passed whatever electricity was available to the server. Consequently the server would begin to boot then the power would go out. It did this for hours until the power supply and motherboard gave up and promptly went to Silicon Heaven.

    As management had quashed all attempts to buy replacements for anything, there were no spare parts. The only EISA based machine was my Unix admin's desktop PC. About an hour later, with the development team complaining nonstop and their manager being a exceptional pain in the ass, I had disassembled the Unix guys PC, inserted the SCSI cards in it, snaked the SCSI ribbon cabled between them, grabbed the EISA config files, and booted the whole mess from a 3.5" floppy disk.

    It worked. I was lucky that the PC had more RAM than the server and that everything came up. I photographed the entire mess and made sure management knew what what had happened. The server stayed in that state for over a year.

    The IT manager, a particularly slimy little bastard from Shanghai, took the credit for the fix. And I got blamed for the fiasco.

    1. Alan Brown Silver badge

      Re: Sounds Familiar...

      "It worked. "

      You fool.

      The BEST way to turn "there's no money for that" into "what do we need to do to sort this?" is not to pull that kind of bodge in a company that's too far up themselves to listen to warnings.

      And you can add "fire the IT manager" to the list of fixes.

      1. A.P. Veening Silver badge

        Re: Sounds Familiar...

        And you can add "fire the IT manager" to the list of fixes.

        Agreed, but above his pay grade (it was his manager).

  24. Wexford

    Late 90s, we did a similar thing. Late 90s, two servers in tower desktop FF, one with a CD-ROM drive and the other without. They wound up snuggling like lovers with the CD-ROM drive ribbon dangling between them like a strand of spaghetti in Lady and the Tramp.

  25. Kiwi
    Devil

    Months???

    Months?

    The man's dreaming! MONTHS?????

    I know a machine running still that 10 years back I told them was "a kludge guaranteed up until the moment this conversation ends. If it's still running when I leave the carpark consider yourself lucky. Get the data moved!". And it's still going..

    But... That's not the worst...

    In the factory we had a hulking great rectifier. Energy levels that would give a dictator wet dreams, and internals that gave electricians wet daymares. Mor'n 20 years before I saw it the original cooling fan sucked it's last bit of dust, and was replaced by an old fan from the roof. The rest of the fans were pulled from the roof and tested, working ones put into storage. Oh, replaced? I mean sat on top of the thing above the old fan. I doubt any of the original internal wiring was left by the time I first saw the thing, and the original circuit diagram was a memory in the head of some poor fellow down the local asylum - the last electrician to try and match the paper with reality.

    Despite bypass cable after bypass cable, despite cooling fan failures, despite a build up of dust and - well you can imagine what gets into the air in a metal plant - this thing just kept chugging along. Replacement cost was near the valuation of the business, and it had to be kept running as we needed it going.

    For the life of me I cannot fathom how it ever passed an OSH or any electrical safety inspection. No clue how the firm kept some pretty extensive insurance policies in place with this thing in the workshop, and that's not considering it being powered up. I think it was so terrifying that it burned itself out of the memory of those who only saw it on occasion. Perhaps they thought it was a 'joke' or diversion, something not really running but with an old neon indicator light inside to give it it's characteristic operating glow (which, tbh, could've been one such light deep inside - or perhaps an old valve - or perhaps something inside was getting toasty).

    In time I kinda grew fond of it. If anything was going to send me home early, this would've been it. My own personal death machine..

    It's probably still running somewhere, unhitched and miles from any known power source, all it's wires burnt and salted, no sign of any cooling fan or other moving parts, yet it hums and glows and whirs as if all it's bits were factory-new.

    If you hear screams in the night, it's memory has visited my dreams...

  26. Anonymous Coward
    Anonymous Coward

    I Love SCSI!!!

    Call me old fashioned but I love SCSI! The amount of today’s ‘techs’ or ‘sysadmins’ that have no idea how to properly configure a SCSI chain is laughable! Most know how to plug in a SAS/SATA drive as it requires next to no configuration whatsoever, but place a SCSI or IDE drive in front of them & they get confused. Most wouldn’t even know what the additional pins are for next to the data & power connections on a SATA/SAS drive also! I still use an Adaptec 39320UW controller in one of my home servers & have never had any issues with connecting any devices to it. Only time I would change it would be for a M.2 backplane that could run multiple NVMe drives off it - but that will never be available so looks like I will be running the SCSI chain until it finally dies. The same server does also have a hot swap SAS backplane running on a PERC/5 controller that I have populated with large SATA drives purely for the storage capacity (instead of running multiple servers & making the electricity company executives much richer than they already are!)

    1. Nick Ryan Silver badge

      Re: I Love SCSI!!!

      Unfortunately it wasn't just a case of "The amount of today’s ‘techs’ or ‘sysadmins’ that have no idea how to properly configure a SCSI chain is laughable!" - it was the damn hardware developers that seems to posess a terminal lack of understanding when it came to SCSI. Usually, but not exclusively, limited to PC targetted components rather than those that were intended for use in a variety of platforms.

  27. Stevie

    Bah!

    I have only one hing to say about all your tales of woe: BORING!

    Not one tale included the phrase "flames shot out".

    What a sad day in I.T. In my day magic smoke was accompanied by gouts of mundane flame, or it didn't count. You'd be laughed out of the lounge bar of the Brewer's Elbow if you said "A bit of smoke came out and that was that" over a Friday lunchtime pint.

    And as for these reports of otherwise completely ignorable "noises", well, I don't know where to start.

    Don't know you're born. Fought two wars. Rationing. Mafeking. Etc.

    *grumblegrumblegrumble*

    1. Anonymous Coward
      Anonymous Coward

      Re: Bah!

      "Not one tale included the phrase "flames shot out"."

      Then here's one for you!

      When I was a kid, we had a 486DX33. Started with Win3.1, but after a botched sound card addition by my brother and me (that's a different story), we got Win95 on it. Worked fine for years, eventually replaced, the 486 tucked into a closet.

      Fast forward some 10 years or so. My brother pulls it out of the closet and hooks it up, just to see if it still works. When he booted the computer, the screen showed a blinking cursor, and that was it. Until a sheet of flame shot out the front of the computer! Turns out a small capacitor in the back had gone bad, and the resulting spark had ignited the dust in the case.

      1. Stevie

        Re: Bah!

        Excellent! This is more like it!

        I've had more pyrotechnics while vacuuming the carpet than some of the other tales being trotted out.

    2. Kiwi
      Flame

      Re: Bah!

      Not one tale included the phrase "flames shot out".

      Flames? Dunno about that...

      My first PSU insertion... I'd somehow got a pile of mobos and some machines in cases. The "expert" I mentioned earlier had got me started the day before, but I was still barely beginning. One machine I had was a NEC one and instead of having the PSU lead split into 2 equal portions (IIRC 4 and 4), this one had 3 and 5, and had been unplugged (whether I unplugged it or it was someone else I cannot recall, but quite possibly me). When it came to connecting it there was a problem, either plug could go on either end and there was NOTHING to say which. So I looked at a known working machine in the pile, noted all the lack wires were in the middle, and set the NEC up the same way. And fired it up..

      I dunno if flame was involved. I do know some tantalum caps split and released a decent amount of sparks, and some other components transformed themselves into black scorch marks on the board - there were sparks a-plenty and shrapnel (in the "utterly harmless' and sub-millimetre range I expect but still shrapnel), smoke, some noise..

      But I don't think there was any flame.

      Thankfully, I was not looking down at the machine from overhead. I don't recall why not, but I can tell you that ever since, when I turn on an un-cased machine of uncertain explosiveness, I am out of the firing line. And even many machines where I am certain they will not explode for that matter.. You never know if some lost screw from previous surgery will be sitting in a bad place after transport.

      1. Stevie

        Re: Bah!

        Sparks count some.

        But flames are the sine qua non of an exciting tale of Man Versus Machine in A World Gone Mad.

        I blame all these low-voltage CMOS electronics. Can't get a decent blaze out of three anna bit volts. Stands to reason.

      2. Nick Ryan Silver badge

        Re: Bah!

        The prevalance of cameras on mobile phones and me remembering to take a photo or five of something before disassembling it has saved me on a lot of occasions.

    3. Anonymous Coward
      Anonymous Coward

      Re: Bah!

      Not quite flames, but I have no idea why the thing didn't overheat in a very terminal way...

      Back in my first student house one of the guys in the house asked a friend of mine in the house who was going to a computer trade show to buy him a maths co-processor so he could pretend he knew how to use it for his maths dissertation (he was writing a program to display graphs). I'm fairly sure that he disn't quite say those words but we knew that it was what he meant... Maths co-processor? Yep, if you wanted one and your motherboard had the slot you could add one to your speedy 16Mhz 286 processor.

      My friend went to the trade show and, for somewhat less cost than the guy was prepared to pay for a separate 16Mhz 80287 maths co-processor chip, he brought back a board with both an 80386 CPU and an 80387 maths co-processor on it.

      Unfortunately the recipient was an obnoxious twatt and despite him getting a much better and cheaper solution he had guests around and wanted to impress them with how assertive he was. So told us that the 80386 board would not do because it was not what he asked for and how dare we provide something different to what he, the fountain of all such knowledge, had asked for? So we went out and picked up a second hand 12Mhz 80287 maths co-processor chip, charged him the price of a new 16Mhz 80287 one and slammed it into his system. It ran at 16Mhz rather than 12Mhz and really rather hot but somehow it worked and we told him that this was normal and because the maths was hard, the co-processor was working really hard too.

      Later I checked his code and he hadn't even included the relevant maths libaries therefore all the maths in his entirely hopeless dissertation application was performed using the CPU for all the calculations...

  28. ShortLegs

    Ah, SCSI

    Many moons ago my nickname was "SCSI Al", or more likely "Scuzzy Al". Absolutely loved SCSI, and from 1996 every box I had was SCSI based. Usually based around "God's Own Controller", the Adaptec 2940UW until U160/320 came about (I had one U2 controller, but really skipped that generation), and some form of RAID, be it Compaq SMART-2DH or LSI boards.

    *never* had an issue with termination, mixing devices (even 8-bit devices on a 16-bit bus - just install the narrow devices at the end of the chain and use the correct terminator).

    The only problem I ever experienced wasn't really SCSI related. Large *cough cough* multinational, migrating from Netware 3 to Netware 4. B and I working late to migrate data from old to new servers. Large company, IT recognised as mission-critical and *no* expense was spared; Compaq shop through and through, mahoosive quad PPro boxes, loads RAM, mirrored RAID5 arrays 5x9GB UW drives (largest capacity one could buy at the time), spare components AND servers in server room "just in case".

    Its the last weekend of the project, and we migrating over the main site file-and-print server and backup, which also provided Bindery services for network login. And the darn migration fails. Every. Single. Time. The new server runs out of disk space, despite having a [slightly] larger partition.

    Those of you who remember Netware 3 and Netware 4 are nodding in anticipation. NW3 defaulted to a 4KB block size, although 16KB could be specified - but with a large number of small (<16kb) files, or filesize not a multilple of 16, disk space would be wasted. NW4 used a block size dependant on volume size. IIRC we had 32KB blocks, and no sub-block allocation...

POST COMMENT House rules

Not a member of The Register? Create a new account here.

  • Enter your comment

  • Add an icon

Anonymous cowards cannot choose their icon

Other stories you might like