If we can't find a working SCSI cable, the company will close tomorrow Welcome again to On-Call, our Friday fumble through memories of jobs on which things didn't go as planned. Or sometimes went in ways it's not possible to plan. This week, meet “Jean” who in the early noughties scored a gig as “a fairly new-to-the-game support engineer for a shifter of overpriced household furniture.” Jean …
I imagine that Jean has prospered. I too have found myself unbending pins in the middle of the night while the administrators with no hardware nous were reading logs and rerunning diagnostics. Understanding how the kit works from mechanical, optical and electrical viewpoints will always give you an edge. Nowadays, for example: scratched faces on fibre interconnects? Been there, because I used to work, in the 90s, in a company where they used to cut and terminate their own fibres for ATM and Fibre Channel. Do you have a microscope and polishing kit to hand?
"Do you have a microscope and polishing kit to hand?"
If they're cutting and terminating cables like that, probably not.
And they probably "tested" the cable by plugging both cables into a barrel connector at one end, then shining a torch down the other end.
This is what happens when you cheap out and subcontract networking installations to the local sparkies, and he subcontracts the fibre installs.
Sadly the people with this sort of experience are getting rather long in the tooth these days. I'm retiring later this year.
Youngsters marvel at you as you fix a dodgy Cat-5 cable connector by cutting off the old one and applying a new one with your trusty crimp tool.
I did this at one company. The IT Director said that they'd received a quote of £1500 to relay the whole effing cable. All I did was replace the connector and test the cable using my Ethernet tester. With the cable checking out, I plugged it back in and the archiving could be done.
Got a 12-bottle case of Verve for my 30 minutes of work (the company was a wine merchants).
It might be worth getting a list of what bits of kit people carry with them 'just in case' and what bits have saved the day more than once.
It might be worth getting a list of what bits of kit people carry with them 'just in case' and what bits have saved the day more than once.
Penknife. Pure and simple, basic Swiss Army Knife. Specifically one of these.
Ok, it won't cope with fibre termination, but it'll do just about anything else from emergency punch-down resetting to dealing with rack bolts to "field modification" of (say) ceiling panels. I won't pretend it's as good as the "proper" tools, but it stays permanently in my pocket and has saved me more times than I can say.
M.
I too had a proper SAK when I was working.
I was in Arizona once when I was faced with a pile of documents, all stapled, that needed unstapling and reassembling, so without thinking I got out the SAK and started to remove the staples. Suddenly I was aware of these Americans staring at me, and one asked how I had ever got that through the airline security. This was two years pre 9/11, and I told them I had checked and it was perfectly legal. These were guys who had gun racks in their pickups, go figure.
I always have a rather large pocket knife. It gets put in my luggage when I travel, then back in my pocket when I reach my destination. Same with the RJ-45 crimper. I have watched people in our engineering department marvel at me putting a new end on a piece of proprietary test equipment cable. "We just spent $160 on a new one of these. I didn't know you could just replace the end." And when you need a crossover cable in a pinch, it's simple to just whip one up in the exact length you need.
The company I used to work for was going through so many telephone handset cords at their call center that I bought the manager a RJ-10 crimper and a box of connectors and showed her how she could replace the "staticky" cable end in a few minutes for 10 cents instead of spending a few bucks per cord to replace a dozen a week.
I never did a lot with my electronics education other than being a hobbyist, but it does also often serve me well in IT.
When are you ever going to need a Philips screwdriver these days?
When are you ever going to need a corkscrew? It's getting very difficult to find corked wine these days.
The advantage of the Philips is the T-shape the penknife makes when you open it. Although it's easier to unscrew rackbolts using the big flatblade, sometimes they're so stiff you need the extra leverage that the T-shape gives, just to get them started.
Oh, and a torch, as someone else mentioned. In my case it's a simple Petzl Tikka headtorch. I've had a number of others, but this thing is tough and the batteries last well. The light is bright, but not blinding and the beam is wide-ish, rather than the useless spot you sometimes get with cheap LED lamps.
M.
Swiss Army knife is SO useful. Last year I had the screen on a Nexus 7 Mk 2 replaced, but later found the rear camera did not work. Opened it up, took it all to bits and reconnected the camera with just the one tool.
Mind you, the original repair guy, in China, had made me blink when he opened the tablet by running his thumbnail round it!
It might be worth getting a list of what bits of kit people carry with them 'just in case'
Leatherman multitool (Charge TTi), PB Swiss tool roll, one of those nicely bright LED flashlights (single 18650 cell. And a headband LED light.
Got a DS25 going again, twice, and oodles of DS10's.
For PC's I have the Lowa Tibet size 46 (11)
"All I did was replace the connector and test the cable using my Ethernet tester. With the cable checking out, I plugged it back in and the archiving could be done."
A cable run that long would surely end in a back box and not a plug (jack)? So yes I'd have done the same but I'd have got a drill, plugs and screws out first and a flourish with a punchdown tool.
Then I'd have got some trunking ...
"It might be worth getting a list of what bits of kit people carry with them 'just in case' and what bits have saved the day more than once."
A live Linux CD to prove it really is Windows that's fscked up after all the other diags and tests have proved the hardware is good but the customer is still complaining that the audio or whatever still doesn't work and yes, you really do need to get back onto your own IT people to fix it. (ie re-install it since most likely there is OS corruption because the user probably pulled the mains plug out on a live system one time too many)
Agree with what you're saying about the live Linux CD, but I was rebuilding a used PC for someone with a new Windows 7 installation and found that it would hang during installation. A live 64-bit LInux CD would work fine, as would trying the 32-bit install of Windows.
In the end, it turned out that the machine had some hardware problem, but apparently the Linux disc and 32-bit Win disc didn't tax the machine's resources like installing 64-bit Win 7.
I do. But last time I thought an audio 3.5mm > phono cable was playing up, I spent ages trying to 'fix' it. No joy. I took the input panel off the back of the speaker and discovered that the coiled wire thingy had fallen off the crossover PCB. Dollop of polyurethane and a resoldered contact and the whole set up was right as rain.
Well, the good thing is that even not knowing the a coiled wire thingy was, I could see that it wasn't where it was supposed to be! Hmmm, the groove in this lump of solder matches the wire that comes off this strangely free thing....
Similarly, one of the pink donut things on a motherboard of mine was brown and dirty looking, unlike its friends. Diagnosis was the motherboard was, and I do believe this is a technical term, fucked.
From the days filling out service forms (when online forms didn't exist), we had a box where we had to enter our reason for replacing an item, regardless of how obvious the problem, and one of my colleagues gained fame for actually using the following phrase repeatedly and it not getting spotted by management - "Field Unit Checked, Known Error Detected".
> Woah, slow down with all your technical jargon! :-D
I had similar buying a new mic. Turns out it was the sound card drivers. Branded ones just so happened to crackle and pop at the point with a mic you'd suspect hardware faults. Unbranded (direct from the chip manufacture not the badge supplier) work a treat.
One could not. The inductor stores energy. A permanent magnet stores a magnetic field but it isn't an inductor.
An inductor stores energy proportional to current and a capacitor stores energy proportional to PD across the terminals. Short a capacitor, get a magnetic field. Open circuit an inductor, get an electric field. The EM field gets transformed.
When I came across this in A level physics, it just seemed so beautiful and elegant I immediately wanted to be an electronic engineer.
"....always suspect the cables." But also be careful to test them as they are installed (usually curled up and held by Velcro because the SCSI cable is like 15 feet long when the server is only six inches from the drive). I had an engineer who insisted one pair of SCSI cables were good and refused to replace them, because he took them out of the rack and laid them out straight to test them. I suspected they had been curled too tightly, beyond minimum curvature of the cable, and straightening them was making the damaged wires inside the cable contact and pass his test, but the minute they went back in the rack and were wrapped up the break opened. He said that was unlikely, but when he wrapped them up and tested them they both failed. Lesson learned - test as you mean to use isn't just for software.
/Beer, 'cos that what cabling war stories are meant for!
Related to test as you use, and cables. Not my story, but I can't remember whose it was...
Call to helldesk that network connection is out. Tech come up to fix and everything works as he stands there testing it. Next day, same thing happens. He walks to desk & everything works as the guy who works there sits back to give him space. Day 3, same again.
This time he asks the guy to demonstrate and as he sits at his PC & starts to work, out goes the network connection. After much troubleshooting the fault was traced to the network cable, that a dodgy installer had run straight across the floor under the carpet... It was positioned such that as the guy sat at his desk, the wheel from the chair ran up against it & over time it had cracked a conductor. So when he sat there it was pushed open, but once the chair moved back to give the tech space, it reconnected.
Backups are important. I would be at the head of the queue to tell you this. In fact, I'd probably have come in half an hour before, said my piece and long gone by the time the queue formed.
But not this important! I mean, they're backups, for Bob's sake! Of your actual data you actually need to run. Only to be needed -- and usually found not up-to-date/functional -- when the live copy fails.
"But not this important! I mean, they're backups, for Bob's sake! Of your actual data you actually need to run. Only to be needed -- and usually found not up-to-date/functional -- when the live copy fails."
The thing is, the people demanding the backups belong to the insurance company. Basically, they're not going to pony up for "being bloody stupid," and not having a viable backup to them is considered "being bloody stupid," since live data crashing without a backup basically means you're screwed.
Indeed this is a story that resonates a lot with me as well.. spending hours into the night in a comms cabinet at a new office just the other side of the office park to the main HQ, configuring a router where the link would stubbornly refuse to come up and everyone else had given up.
Checked the newly installed cat5 drop to the downstairs comms room where the fibre terminated.. and the so called "professional" network engineer has punched two wires into the same bloody pin.. sigh..
Always check the cables.. yes..
Icon because, well, Friday!
Nope. There's a lot of connections that aren't, and those can be way more requiring of dark rituals (goats, black candles and pentagrams in the case of SCSI, for instance)
Surprisingly (or maybe not) large proportion of SCSI issues are down to termination (or lack of it). Often due to some people not understanding that the bus needs to terminated. Particularly mixing narrow and wide SCSI can easily trip people up.
Then there are the devices with internal/auto termination that doesn't. Been caught up by that enough times that I get very suspicious when no external terminator is present. Likewise I prefer active terminators to passive ones given a choice.
Years ago, had a PC with an internal SCSI drive. Worked fine with Win98 (said it was years ago!) A few months later, tried to upgrade to NT4. Couldn't even install it! However, once I installed the missing resistor-pack terminator thingy on the drive it worked...
"....cabling....." I'd suggest that's because cabling is usually done on the lowest bid because manglement don't realise that there is a vast difference in quality between cable manufacturers, let alone cable layers. The other contributor to such issues can be no rack cabling standards in use.
"He told us that if we don’t have a database backup by the end of the third night, the underwriters for the company insurance would pull our cover (including employer's liability) essentially putting the company out of business".
I've never heard of such a thing, do you have a link to the text of such a policy? I mean normally senior management are totally disinterested in the intricacies of IT backup maintenance. In fact most backup media out there are unrecoverable. Which won't be discovered until they have a major disaster.
"Agreed. Why would you even tell your insurers you had a backup issue?"
Well because if your trying to claim for "business losses due to losing all my data" then you'd have to tell them that, no ? Oh wait you mean Lie . i gotcha .
Yessir - we did have a backup but the burglars stole that as well.....
As the originator of the story, I am indeed "Jean", I can tell you that when a company is putting something like £1m (Sterling) into a 6 month project that will transform their underlying data backbone, processes etc, then the beancounters and directors all, will take the quite sensible risk mitigation strategy of taking out Project Insurance to cover any eventualities. Of course, going through your existing supplier to get a better deal is good for the beancounter.
Add to this the fact that the head beancounter * was a whiny bitch, and a director of the company to boot, I'm sure the phrase due dilligence would probably be his mantra.
Anyway, since this took place in the middle of the project to migrate every single remote server and their databases into the head office "datacenter", the DB backups were of paramount importance. ,
Anon to protect my identity and the identity of the shoddy furniture shifters.
* The head beancounter once wanted to edit a few pictures he took on his digital camera, so bought a Photoshop to do it.
> As the originator of the story, I am indeed "Jean"
Three days to migrate a data center and you brought in on the first day? Sounds like the place was being run under what I call 'crisis management'. That is the staff being continually kept in the dark, senior management engage in information hiding and projects being undertaken at the very last minute under the direct supervision of the senior clueless CIO, who only got the job cause he went to college with the owner. If so - then if you ever come across such a business in the future - run like hell.
"He told us that if we don’t have a database backup by the end of the third night, the underwriters for the company insurance would pull our cover (including employer's liability) essentially putting the company out of business".
If the insurers are putting in clauses like that, then your bean-counters probably cheaped out on the policy and accepted such exclusions to keep the premiums down.
"Which would explain why the bean-counter was getting twitchy and piling the pressure onto someone else".
But three days ? How did they cope with things like Christmas, unless they paid someone to come in and perform/check backups ?!
I remember a magazine in the nineties had a frebie on the cover: a mouseball-sized textured sphere with a hexagonal shaft.
Yep, it was actually a tool for cleaning the gunk off the rollers inside mice with a power drill.
Urgh, the horror of mechanical mice!
"How did they cope with things like Christmas, unless they paid someone to come in and perform/check backups ?!"
My system does it over that period, tells me that it's suceeded and sends distress calls if anything goes wrong.
Christmas is easy.
When I was in sixth form, there was one computer where the mouse wasn't working.
I looked at it briefly, realised that the cleaners had unplugged it and bent a pin when re-inserting it (it was a 9 pin serial mouse).
Teacher looked over and saw I'd got the connector in my hand and asked what the problem was. He said "just get another one out of the cupboard", by which time I'd already got my multi-tool pliers out of my pocket and straightened the pin... "No, it's alright, I've fixed it already"
Mullered the socket, probably. One of my pet bugbears was the Art & Design students stretching their legs out under the desk and catching the wires from the keyboard and mouse that inevitably ended up dangling down back there. Pulled out of the ADB socket, caught on the excitingly chunky moulded casing and to all visual inspection appeared to still be plugged in.
The mouse might have used a 9-pin D connector but connected via a proprietary card in the PC, aka "bus mouse". Most of the ones I saw back in the late 80's/early 90's were Microsoft InPort devices with round plugs, but I seem to recall some other brands used D plugs/sockets - and some of them used male/female the other way round to serial ports so they didn't get mixed up.
Nope, very early mice were indeed serial devices using 9-pin "D" connectors (I have a genuine Microsoft Serial Mouse in a box in the loft somewhere).
Most of the ones I saw back in the late 80's/early 90's were Microsoft InPort devices with round plugs, but I seem to recall some other brands used D plugs/sockets
Even the MS bus mouse card had room for a DB9 connector, but on all the ones I've seen it wasn't fitted, and no cutout in the slot bracket. There's a Siemens-branded Logitech bus mouse around that has a (male) DB9, but the matching card has gone missing. They probably switched from DB9 to mini-DIN because a bus mouse plugged into a CGA or Hercules card doesn't quite work like one would want.
And the first mouse I bought was a bus mouse with a mini-DIN plug, with a bus-to-DB25 converter block and a DB25 -DB9 pigtail.
A 9-pin D connector on a serial mouse has a female connector (therefore no pins to bend). Maybe you mean PS/2?
ummm... connectors have two pieces... a socket and a plug... the other piece of the connector in this case has pins and getting a new mouse out of the cupboard would not have fixed the problem ;)
"I assume you meant there weren't any computers at your school. Unless you were in sixth form over seventy five years ago! :-)"
There were no computers at my grammar school in the late 1960s. Timetable was written by the lower 6th maths set after exams. We did have Brunswiga machines. In those days, IBM etc. did not have the same influence on school budgets that Apple seems to now.
"There were no computers at my grammar school in the late 1960s"
Circa 1971 someone "kindly" donated a computer that had been decommissioned from their workplace.
It was delivered in bits so what could have been an exciting project was thwarted by the cost of getting it put back together again by an engineer familiar with that kit.
It was still in bits a couple of years later when I left.
> > When I was in sixth form, there weren't any computers!
> I assume you meant there weren't any computers at your school. Unless you were in sixth form over seventy five years ago! :-)
Seeing as the IBM PC didn't come out in the US until November 1981*, and in the UK the next year, I fear that your date calculations have a very large rounding error.
* I had the edition of Byte to prove it, but it got nicked...
"When I was in sixth form, there weren't any computers!"
When I was in the 6th form the computer was a very nice old gentleman called Arthur Postwick, a retired bank clerk. He computed the tide tables up the hill at the observatory.
We used to chase geese at lunchtime to pull a feather to make a pen to write with.
Whose next?
Coat: mine's the one with the soldering iron and solder in one pocket and a set of tank cutters in the other/
Yeah, as a teacher I can see his train of thought:
"Oh, that's my break time gone while I fix it, cos there's no chance if I call IT"
"Maybe I can fix it in class, I'll give the kid a spare in the meantime so he can get on with his work"
"Here you - oh you've fixed it already! I wish all my students were like this..."
One of the worst things that ever happened to me was defective cat5e cables. I got a call saying the server I replaced is not working right what did I do. Of course not working right is not at all vague. I found out that it's not connecting to the net work. No data is being passed. The LED link light is lit. I grab another cat5e cable. Still a no go. I use my cheap cable tester and it says the cable is good.I call it in thinking it's the NIC. I get a new NIC and plug it in same problem. I'm like hmm. Now seems like good time use the fluke cable tester I just bought. Bad cat5e cable. I go to my car use a known good cable and life is peachy. So I thought. I get a call saying the server is not working right. I get there and they replaced the cable I used with their own. I asked why did they replace the cable I used. They said it didn't match the color scheme. Fair enough. I pull out the fluke tester and yep bad cable. Their jaws dropped. Turns out they got a really great deal on a lot of 10 cat 5e cable from china.
Ok so it was not so bad for me but it was for the client.
Had a contract years ago putting new kit into offices around the country, where I was informed at 16:00 on a Friday afternoon, that one of the other sites had reported some new Cat5e 110-RJ45 patch cables which were incorrectly wired at the 110 end. Only the 3m ones, mind... which I knew I needed about 80 of for work going on over that weekend to install and patch new PCs.
Grab a handful of cables at random, grab the tester, and a few minutes later I've about a dozen cables and each one's dodgy. Bugger. Cue several hours in a cold dark office with 120 patch cables (I did them all, just in case, and I think I only found 3 wired correctly), gently prising the fragile plastic clips on the 110 end to remove the cap, rewiring correctly and then replacing the cap. I recall leaving to drive the 130 miles home at around midnight. At least I got paid!!
Not sure who this was more embarrassing for but I worked with a contractor who was a bit of a god on the systems we supported. Due to a small problem with renewing his contract he moved on and took the opportunity to try a different path by opening a shop. Knowing his IT he installed IP-based security cameras but despite testing fine on the bench they would not work when plumbed in. I went down to help him and made up some fresh cables which we initially tested out by laying them out on the floor and all worked fine. Fed the cables through and now all was perfect.
Tested the original cables and they were 'less than good' so I asked where he got them......came the reply B&Q.....I went home
"A good bit of armoured 10Base5 with a vampire tap on the end was far more durable"
But there were older networking standards in widespread use before they came along.
My "bit of fun" with network cables relates to cheap 10Base2 terminators with centre pins that would disappear inside the terminator (bad soldering).
Finding a dodgy terminator on such a cable isn't the easiest of tasks as what you see tends to be intermittent.
... IBM Token Ring. Ah, the horrors you young uns never saw....
The Devil's "network"
I worked for 3Com, designing Token Ring switches...evil things, and almost impossible to get through emissions testing, what with all the ports switching at the same time.
The problem: Twisted pair Ethernet (10BASET and 100BASET) was eating Token Ring's lunch. So...Token Ring over CAT5 was a must! Worked OK for 4Mbit TR, not so much for 16Mbit. It was right on the edge of working, but good luck trying to get more than about 200 nodes on a ring. All kinds of "jitter control" mechanisms were tried, but 16Mbit Token Ring was not destined to be "a thing".
Ethernet over twisted pair was done with a carefully crafted waveform which minimized radiation and took into account the characteristics of CAT3 and CAT5. Token Ring couldn't do that, because of the requirement that jitter be minimized, so nice sharp edges were needed.
"I worked for 3Com, designing Token Ring switches...evil things, and almost impossible to get through emissions testing, what with all the ports switching at the same time."
Evil things... That described a lot of the 3Com kit I had to herd. Especially the NetBuilder II and a LinkBuilder MSH that was the core of our network - I dreamed of using their replacement (a fully-laden Catalyst 5500) to crush them both, but they had to leave the machine room intact because they had trade-in value.
I once had a holiday which was a bus tour of the old cities of Morocco. (recommended, BTW.)
One morning at breakfast, the American lady next to me - whom I hadn't previously spoken to - said that her camera wasn't working.
So I asked what was wrong, and she said it was just completely dead. It had been fine the previous day.
So I asked if I could have a look. Took out the batteries, cleaned all the terminals and connectors, and put it back together. It worked perfectly. She was utterly amazed and extremely grateful.
So we started chatting.
"What do you do?", I asked.
Said she, "I design satellites for Hughes".
In fairness I doubt satellites suffer from much oxide build-up on the battery terminals :-D
Your story did give me a flashback though - remember when RAM used to get an oxidation build up and introduce random memory failures? Re-seating RAM was one of the first tricks I would perform on a dodgy PC but don't think I've had cause to do it for years.
Was that a real thing that doesn't happen anymore, or just a placebo effect we used to experience?
I just re-seated two RAM modules in an old desktop that I have been using. Random crashes ever since I dusted it off and loaded W10. For 6 months it suffered from random lockups only under stress, and it even passed memtest a few times. I thought it was a bad driver for the MB raid chip.
Before that, the guy that gave it to me had tons of problems, and when I initially plugged it in sparks came out of the back of it. Replaced the power supply. When I tore apart the original power supply I found that the socket for the cord was connected using only cold solder joints. Too bad I had torn it apart to the point where it was not repairable.
Not sure about oxidation - I thought it used to be that RAM modules in the "black plastic rectangle with silver legs all along the long edge" type, chips basically, were heated while the PC was in use and cooled when it was turned off, and basically slowly wriggled out of their sockets. A carefully placed thumb press on the centre of each black bit, not touching the legs, fixed that.
Later: a laptop's own UEFI memory test was apparently fooled by its memory cache or something, passed. Booted with SystemRescueCD (Linux) and that memory test said, yup, a dead module. That was DDR-something.
"black plastic rectangle with silver legs all along the long edge" type, chips
DIL chips, Dual In Line. The worst type was SIP modules. Like SIMMs or DIMMs but instead of the edge connector we're all familiar with, it was a single line of pins incredibly easily damaged, bent or snapped.
basically, were heated while the PC was in use and cooled when it was turned off, and basically slowly wriggled out of their sockets.
Yeah, we called it "thermal creep". I went to one many years ago, symptoms described as "intermittently crashing 2 or 3 times per week". Took the lid off and it was almost unrecognisable inside, everything buried in a thick layer of black dust (industrial site) and half the RAM chips just fell out when I tipped it on its side to start cleaning it. It was a machine tool works so maybe that black dust had a high metallic content and kept it working.
It was definitely a real thing. Tin-plated SIMMs and sockets were the worst, gold-plated ones slightly better. Putting gold-plated SIMMs into tin-plated sockets (or vice versa) was not a good idea, it invoked two camps of angry gremlins - those of physics and those of electrochemistry.
Physical weakness was mostly down to the SIMM socket design, where socket contacts touched only one side of the SIMM. And two measly latches in the socket had to hold the SIMM in place with enough force to secure all 36 or 72 contacts.
DIMM design is much better because of symmetricity - socket contacts are applying force evenly from both sides of the DIMM. Also, JEDEC gave out an edict saying "thou shalt not skimp on gold when plating contacts". Or something like that.
Recent example. Home PC locking up and blue screening complaining about BIOS corruption and graphics drivers.
Firmware updated. Still the same.
Drivers upgraded. Still the same.
Took the lid off and nearly died from dust bunny inhilation. Removed the graphics card and cleaned it out along with the rest of the PC. Reseated the card. Fixed.
Always check the hardware....
"But what if it's locked down and you lack the key?"
Presuming it's your kit (or you can get permission to cause cosmetic damage)
Option 1: If it's a padlock break out the bolt cutters, If it's a key lock built into the case see if it can be jiggled around/forced with a screwdriver
Option 2: Drill out the lock, or cut out the casing around around the lock.
If it's not yours and you can't get said permission, declare it not your problem and walk away until the key is produced.
"If it's not yours and you can't get said permission, declare it not your problem and walk away until the key is produced."
They block the door and reply, "You're not leaving until it's fixed."
"So you're basically saying I'm contractually-bound to fix a computer I have no permission to access and cannot leave until I do so?"
"Yes. And before you say this is illegal or the contract unenforceable, we know some good friends high up in the police forces..."
"If it's a padlock break out the bolt cutters,"
No need - you can break most small cheap locks with a couple of spanners: https://www.youtube.com/watch?v=1jJP0CcuJyE
"If it's a key lock built into the case see if it can be jiggled around/forced with a screwdriver"
find ballpoint pen of approximately the right size. Heat the end to soften the plastic and jam it into the lock. Wait till it cools and then turn. Usually works (and works for bike locks too). https://www.youtube.com/watch?v=R0q7Bnp8ZvY
Alternatively - keep a selection of case keys.
It's much easier to invoke the SEP field generator though.
I've had that.
There have been many occasions where I've upgraded the graphics card but I cant get the damn thing to work!
4 hours later, after many reboots, driver reinstalls, hair pulling etc etc, and getting close to midnight (as I know I wont sleep cos my brain will be trying to work out why its not working) it dawns on me - I haven't plugged the power supply cable into the card - or, on later cards, only 1 of 2 have been connected!
Went into my local motorcycle dealer to check on how they're doing fixing my bike...
The service manager is bitching up a blue storm because Windows keeps crashing. It's an ancient Dell box and I can hear the fans at full tilt enough to launch a 747, so I ask him if I can fix it.
[angry glare] "How long?" "Probably 5 minutes"
Of course it's absolutely packed with crap it'd inhaled in the past 10 years in a non-enclosed shop. I actually had to pull felt-like mats of crap out with my fingers. I grabbed some compressed air, and a month later he's still telling me "you know, that machine hasn't crashed since! it used to crash all the time!"
Now in the rare times my bike needs work, it's first in the queue and out in an hour. He did an ECU recall in 15 minutes for me.
"Took the lid off and nearly died from dust bunny inhilation. "
Careful with that if there are smokers around - nicotine infused dust bunnies can be nasty (they're concentrated enough that it can be absorbed through the skin).
Don't forget to warn the user that "thou shalt not put machines on the floor or (worse) under a desk, as they suck in every bit of airborne dust around them"
I had a similar issue with an IBM AIX box a couple of years ago; after a reboot, my data partitions didn't come up. Clearly a software issue. Nope. RAID? Nope? Dead disks? Nope. After a couple of engineers and various support staff at the end of a phone line had pondered on it for a couple of days, engineer three and I were at our wits end. Looking at the affected drive cage, we noticed the thick power lead that snaked all the way to the back of the case, out the back via a connector then back in at the other side of the case. And it was loose. Plugged that back in and bingo... I can see why it's done like that, but as a Wintel guy, I'd never come across anything like it before. Neither it seems had engineer #1. His company lost that support contract.
One customer had (has?) SCO at their sites, years ago.
We were asked if we could interface the head office mail system with the store SCO systems (connected by dial up modems using a proprietary file transfer system, no IP, apparently no e-mail MTA).
Possibly the correct way to do this was UUCP, but I'd decided (not being terribly experienced in Unix itself at the time) to use Sendmail with a hand written set of routing/re-writing rules, the proprietary file transfer, and a custom written central routing program. Somewhat fiddly with just a loaned SCO box and the O'Reilly Sendmail manual, but it was an interesting exercise and was completed. Remote installation was required for Sendmail, too..
Just as the project was almost complete, the customer came back and said 'we've discovered that there is already a proprietary SCO e-mail server on the systems, so we don't need to transfer and install Sendmail. Could you use that instead?'
That'll be a 'no', then.. Buggered if I was going to rewrite all the custom routing rules, and the SCO mail system probably wasn't up to it anyway.
... like ya'll seem to think.
I have a client with a (currently) 2013 Dell Precision SCO Linux tower server doing inventory and POS duties, with WLAN Windows XP/then Win7SP1 full workstation terminals since at least 2005. Surprised to see SCO Linux back then IIRC. Whole system gets security updates/software updates on a quarterly basis, too.
Somebody is maintaining it, I'll check on Monday what level the server is at.
Hello 404, I was writing about SCO Unix not the Linux version, and I have no idea of what is going on now but I remember they had some big customers like McDonald's. The SCO customers we had, around twenty, all converted to the NT during those years when Microsoft was smiling. Wasn't that difficult as we used PROGRESS for our systems. But it wasn't easy for the customers. I think all of them had to upgrade their new hardware soon because the damned thing got slower. Some customers suddenly found unwanted people messing around in their systems and worst of all they still needed help, something they did not expect because it was Windows.
I worked for a company that owned a lot of restaurants, mostly Pizza Huts. A large majority of those ran on SCO on either IBM or NCR hardware. Very stable OS, but user friendly to work on, it was not. NCR used to build a helluva reliable box though as far as hardware goes.
...just resting?
That SCO Linux started life as Caldera Linux. Before Caldera conceived this bright idea to rename itself to "The SCO Group" and start suing world+dog for things that they didn't own, especially for Unix bits written by the original SCO aka Santa Cruz Operations.
TSCOG is not SCO. Or should we use newSCO and oldSCO?
It sure isn't.
Just last week I had to rebuild a SCO OpenServer 5.0.nobody_knew box for a customer.
He's now on a Athlon 1.4G (instead of the old P2-450); has 512M RAM (instead of 64M); and now we're having problems with handshaking on his serial printers *grins*
And a new, fully legal copy of 5.0.7 registered and installed on his new server.
RwP
"All of the various SCSI types I've worked with had the pins in the device or on the controller"
That's odd. HD-68 was easily the most prevalent connector type for internal cabling. From Fast/Wide (appeared 1995 or so) through Ultra320. How did you manage to miss it? OTOH it may be a good thing, as SCSI cabling has always been an unholy mess. Good protocol though.
Working for the power company a while back when I came in Monday morning a little late to find a pile of help desk tickets for the engineering shop. Mice not workimg, computers not powering on, monitors not displaying, printers not printing, cofee maker down, etc... They thought they must of had a lightning strike or power surge.
Drive out there to find that the "safety team" over the weekend has decided to tie wrap all the cables of every device and in the process manged to randomly pull loose various connectors of two hundred plus machines.
Years back, I worked for 3D Systems, supporting their old resin/laser-based stereolithography systems. Customer had just taken delivery a of shiny new SLA machine, and the elevator stage had a mind of its own, with the stepper motor bucking and intermittently running the opposite direction as commanded. Two regional support techs and their senior supervisor visited the site, replaced nearly every component in the subsystem, but still the problem persisted.
After three weeks of this, the customer was ready to crate it up and send it back. So - I'm dispatched with a one-way ticket to Oregon to figure it out. First thing I did was unplug/check/reseat each connector on the harness involved. Sure enough, in manufacturing one of the pins in a molex block had been improperly soldered to its wire, connected by just a few strands. Sometimes the motor got correct current to all windings, and sometimes it didn't. 15 minutes with a crimp and portable soldering iron and all was well. Cabling is always a playground for failure.
I too had to fix a CNC machine - every so often, but not always, the machine would engage its brakes (by design it locks itself rigid if it loses contact with the stand alone XP PC it shipped with, for safety reasons) and though it would resume the tool path would have been knocked off kilter. The cables were innocent this occasion. I swapped out its Pentium 4 CPU for a faster Pentium 4 HT*, and the problem never came back. Faith restored, we could leave it on a 30 hour job and go to the pub.
In all likelihood, the original CPU would have been up to the job, were it not for Windows XP sometimes deciding to do something you haven't asked it to do, thus momentarily distracting the CPU from the one thing we asked of it.
*I happened to have this CPU lying around ever since the pick donutty thing on its previous motherboard turned brown.
SCO wasn't bad at all, and I compare with Solaris, HP-UX, Aix and For:Pro. (Any For:Pro people out there, Monaco anybody.). Of course SCO came on PC hardware and wasn't up to the more severe hardware for other Unix systems. The occasional unproblematic compile for more semaphores and that was about all. HP-UX had in fact more odd problems with some of the cell commands.
I still feel sorry for the good people at SCO who lost their jobs due to that one idiot who dreamt of a billion bucks out of nothing.
Ps. the SCSI cables for Sun were expensive indeed.
usb <-> ps2 dongles (2 at least)
usb serial port cable (both a 6 foot and a 16 inch one)
a Canadian tire magnetic multi slot screwdriver and a box with - 6" hex drive extender and 72 bits for the damn thing. Covers just about any screw head I've ever seen. (including those damned triangle ones that every oriental aftermarket tat assembler thinks are secure) The bits are from PrincessAuto.
1/4 and 1/2 socket drivers with 4" 6" 14" and 24" extenders, 48 sockets < that lot tend to stay in the spare wheel well of the car >
mini electric screwgun (with charger) < again, in the trunk most often >
Cable end crimping gun and I try to keep a dozen or so ends.
I actually have a parallel port cable, a serial to serial cable, and two USB2 A->B cables.
and not one but 2 spare 64Gb USB thumbdrives.
an 8' tested good cat6 cable. <helps if your TESTER is pooched. And I've met at least one>
I personally carry a 64Gb USB key that has ISO images for a *baseline* recovery image for *all* the OS's I support. And spare CD and DVD disks.
I ran across the Sn/Au issue on sockets (root of the "pull and re-seat" voodoo dance) in the 1970s. Researching it, I found a paper from the 1960s, and then a paper from the 1950s. Apparently the issue of the Sn/AU high-resistance eutectic is just one of those things that our (technical) society forgets regularly, perhaps as a consequence of the "up or out" policies of companies regarding technical people. When the only way to pay your adjustable rate mortgage or get your kids through school is to move from the "technical ladder" to the "management ladder", most just move, and thus every decade (including apparently the 90s with SIMMs, and probably the 80s as well) the "common knowledge" gets lost.
this story must be OLD, that sounds like early 90s technology replacing mid 80s...
re: tools, in my pocket at all times is a Leatherman "Juice" (a smaller/slimmer leatherman) which has a phillips driver as well as pliers/cutters, AND a little AAA powered Fenix LED flashlight which is way more than bright enough AND tiny enough that I always carry.
in my pickup is always a pair of 'truck boxes', one with a reasonable set of 3/8" metric and imperial sockets (hey, I'm in da USA, land of the ... oh wtf do they call that stuff now?) and combo wrenches, and the 2nd with an assortment of electrical tools including my all time favs, a bent pair of needlenose from Knipex, which are way better steel than the current USA brand name made-in-China junk. Said bent needlenose are the 'perfect' pin straightening tool.
I remember much frustration as the interface changed from Centronics to SCSI-2 50-pin connectors and bent pins caused so many problems. Usually it was a Jr. Admin forcing the cable to connect and leaving me wondering why I'm suddenly getting so many SCSI errors. I have to say though, those little pins were resilient. Even after being smashed in sideways, most of the time you could bend the pin back, carefully reconnect the cable, and it would work again.
Nice read, brought back a lot of memories.
Some years ago we had a probloem with a PICMG enclosure. Two new expensive switches, with fibre uplinks, literally "went mad", showing a very erratic behaviour. Calling support didn't help. On site technician replaced them, same issue. The networking colleagues were giving up (I was responsible of some applications running on the computer boards). After a session of "RTFM" I told my boss and colleagues it should have been a grounding issue - but nobody believed it could have been something so simple. I got a cable connected it to one of the grounding screws, and then put it around a cooling water pipe. The switches started to work without issues...
"....grounding issue...." Yeah, it is surprising how often someone decides it's a good idea to save $5 per rack by not buying the Earthing kit from the rack vendor. We had a partner who kept causing delays on a project because one of their servers kept having RAM and HBA failures. Because it had only been happening for six months or so, their admin insisted it was zinc whiskers coming off the tiles in the old datacenter flooring - cue two extensive vacuum-cleanings of all the racks and underfloor space, plus all the old floor tiles being replaced (not cheap!). Nothing changed, to the point where my employer threatened to sue them if they couldn't solve the problem. A quick chat with the vendor who had switched out the components told me it wasn't zinc whiskers but static damage, which the admin ignored because it didn't fit his pet theory. So what else had changed in around six months before?
A new set of vending machines had been installed in a corridor behind the DC. The carpet tiles in that corridor created static. Prior to the new machines, the staff had to go out of the building to get food, using an anti-static station when coming back into the DC. The new vending machines were accessible by cutting through the DC, the staff stocking up on snacks and static, then traipsing back through the DC and often touching the un-Earthed racks on the way through.... One box of Earthing kits and an additional Earthing station later and problem solved.
This post has been deleted by its author
I ran across a story of how a man's PC kept rebooting every time his toilet was flushed. The tech thought the guy was crazy, but verified that indeed, the PC would restart every time the toiled flushed. It turns out the line voltage was low and the power supply was weak. When the toilet was flushed, the man's well pump would come on to make up for the water pressure drop, which would drop the line voltage just enough to cause the PC to restart.
Then there's this story I've always loved: http://catb.org/jargon/html/magic-story.html
First proper job was supporting OS/2 Warp 4 workstations on a Token ring network, talking SNA back to an IBM mainframe. So all this Windows using TCP/IP over Ethernet is something of a breeze.
We used to have a problem whereby every so often the entire floor we were sat on would just fall over - I remember standing in front of the Madge Networks "Ringswitches" and watching them go into full-on Christmas-tree mode. The relays used to make a great noise when they tripped. Cue many hours of playing hunt-the-knackered-balun under people's desks. One ring per floor, so when one device failed, 300+ people fell off the network. Nothing like a bit of segmentation.
I also remember another issue - encountered as they were upgrading to NT4. Turns out that the new IBM servers had been installed with dual NICs inside them. The primary was cabled to a dedicated 100Mbit HSTR ring, but the second was simply connected to the user ring on that floor, so when the HSTR switch failed (which was about once a month), then all of the servers would fail back to the 16Mbit floor ring, massively overloading it and reducing the entire floor to a standstill. And this was in the day when network management was very much an afterthought, so the usual diagnostic process involved staring at the box and hoping for divine inspiration!
In terms of tools:
1) Ethernet cable joiners - straight-through and XO
2) Assorted screwdrivers
3) Tweezers
4) Scalpel
5) RJ45 plugs and crimps
6) Leatherman (with belt-holster, because the ladies love a belt-holster)
7) Domain-admin password, written on post-it
8) Various network patch cords and fibres
9) Mini-Maglite
10) Vanilla netbook purchased from PC World to get around USB-lockdown
11) USB-to-serial cable plus assorted console cables
12) Aftershave, just in case
The closest I can come to that is visiting a site that was having random errors all the time. I checked physical connections and all looked fine. I could attach my PowerBook G4 to the network and could get out and see the other computers.
After struggling for a bit, I decided a "tcpdump" (looking at the raw packets) might shed some light on the subject. After manually inspecting traffic for a bit, I would occasionally see a blurp of Cisco packets that seemed out of place.
I pulled out my trusty 4-port Linksys Ethernet hub from my backpack and swapped a few of the computers over to that connection. Sure enough things started working. The office had to deal with only a few terminals being available until I could get Cisco to replace the switch which had apparently gone TITSUP.
Oh yeah, there was another time I had to work on restoring a backup at a courthouse over a weekend, including being locked inside the building by the County Clerk. Although I eventually prevailed, that was a very long manual restore process. The error was our fault, so I explained it to the CC and she thanked me for not bullshitting her and taking care of the problem.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2025
