Heard the one where the boss calls in an Oracle consultant who couldn't fix the database?

Over the months, On Call and Who, Me? – El Reg's weekly columns of tech support cock-ups – stack up quite a few submissions that are too short to whip into a single piece. To help fend off those post-Christmas blues and remind you that you do indeed need to return to work soon, we've compiled a list of winning one-liners and …

[Anonymous Coward]

A "proof of concept" project was part of a big government IT bid process. To stress the comms network the customer employed a consultant who brought a terminal protocol emulator load generator made from many microprocessors.

In one crucial test their consultant declared we had failed. He pointed to the stand-alone network monitor that apparently showed a corrupt transmission from our system. I looked at it - and pointed out that the monitor had shifted the bits by one place. If you looked at bit level you could see the poll was valid and the emulator had responded correctly.

To prove the point I showed him how to write the displayed hex values as bits on a sheet of paper - then regroup them into octets one bit shifted. It took him a while to grasp the principle - particularly of sending an octet's LS bit first. He apparently knew nothing about the bit level of comms protocols.

On another occasion we were failed for not always seeing responses. We showed him on the monitor that it was his emulator not replying - presumably as it had reached its throughput capacity well short of the test's intended maximum traffic.

I earned my pay on that project.

[DJV]

"I earned my pay on that project."

More like you earned the consultant's pay! I hope his pay incorporated an idiot tax!

[Mr Benny]

Years ago I was working for a large US bank that needed a network link to a stock exchange that used a proprietary protocol over TCP. So far so good. The problems started when the exchange - whose servers were all unix based - gave us a test system written by outsourcers unrelated to the team who wrote the live system and it ran on windows. Alarm bells rang, and sure enough it had more bugs than an entymology exhibition. We probably wasted 2 or 3 weeks in total deciding if a given bug was ours or the test harnesses. Eventually we persuaded the exchange that it was a POS and they gave us access to a proper system to test against. Not that I was complaining, my fixed term contract was extended because of their screwup. :)

[Anonymous Coward]

This is a common theme. Do you trust the test stuff? On a project once I wrote a simple tool to dump ethernet frames as hex and it was very useful, if a bit long-winded, to debug the higher level protocols. A colleague was writing a protocol analysis tool - a sort of austropithecus to wireshark - but we could never trust what it was telling us, even after a lot of work to debug it. So many months of work on it was abandoned.

AC as I don't want to embarrass him.

[Anonymous Coward]

"[...] but we could never trust what it was telling us [...]"

That was true of many commercial network monitoring tools - especially the ones with "expert" analyses. I wrote my own 30 years ago - which could be quickly modified when a new problem was encountered.

While its high level abstractions usually proved correct - I always double-checked them at frame level. Tedious but necessary.

While people trusted my results in situations where the commercial tools had failed - most wouldn't spend any time learning how to use the tool. They preferred to get accreditation on a commercial tool which could be parlayed into a pay rise - or a lucrative job change.

[TRT]

RAIB report 19/2018 Collision at London Waterloo.

A test desk had been added to the signalling control room in order to simulate inputs into the signalling system during major works on the track there. Someone at some time later added an extra wire into the system without the correct procedures being followed because the test desk no longer reflected the works they had just carried out. This provided a current path that shouldn't have been there to a point motor which deflected a passenger train into an engineering train.

[Alan Brown]

"This provided a current path that shouldn't have been there to a point motor"

There's a lot to be said for NOT using analog control systems to control boolean systems - "phantom paths" is something that shows up time and time again in control system catastrophes from aircraft to nuclear power stations to railways and more.

(That's quite apart from ensuring that the displays show the REMOTE device actual status, and not the LOCAL switch position - which was one of the fundamental failings at Three Mile Island which allowed the meltdown to happen. Switch said valve was closed, valve was not closed, no feedback to say switch and valve didn't agree.)

[Chairman of the Bored]

TMI failure

You're quite right that the Power Operated Relief Valve (PORV) indicator screwup was a big part of the failure chain. But there was one indication available that could have provided a clue: PORV valve tailpipe thermocouple readout was undoubtedly high, BUT the design of the control room was so jacked up it might as well have been under a filing cabinet, protected by a jaguar.

Apparently there were so many alarms going off the line printer spitting them out fell behind by hours.

Perhaps not the best user interface design?

[Anonymous Coward]

I got something similar when a Linux group tried to put the blame on my Windows software for the low throughput of a system. I was able to show them most of the time my software was almost idle waiting for input - and using test inputs it scaled fairly well.

It turned out their software took the Unix philosophy too literally. They had three distinct processes doing very atomic separate tasks ("Write programs that do one thing and do it well!" - OK, but you have to stop well before single functions applications...) and sharing data through disks files ("in Unix everything is a file!" - without understanding the difference between a generic file API and an actual disk file) - it was much more than disk caches could handle, and they spent just a lot of time waiting on I/O - I wouldn't really wanted to be the poor disks heads as the single, slow disk was truly overwhelmed.

Of course they used nor debuggers nor profilers - true hardcore programmers don't need them... they just try to blame someone else. And let's not speak about the dreadful error management...

[Doctor Syntax]

"sharing data through disks files"

Instead of pipes?

[Anonymous Coward]

As I wrote they misunderstood the 'in Unix everything is a file' meaning. For them, a file was a disk file. The fact you could access other mechanisms with the same file API didn't touch their minds. Moreover, there was no need of three different processes, they could have shared data in memory directly. And locking at the whole file level they serialized more than was necessary.

They just added a layer of complexity to follow blindly some Commandments.

Even funnier, they didn't catch the issue because they mostly tested each process separately, so disk access contention didn't happen.

[Alan Brown]

"Even funnier, they didn't catch the issue because they mostly tested each process separately, so disk access contention didn't happen."

Latencies are fun aren't they?

Even without disks in the way, if you have a TCP/IP comms link running near its limits, latencies can go from "tolerable" to "oh shit" in the blink of an eye.

The PROBLEM arises when looking at usage statistics because they're invariably 5-minute averages and the instantaneous peak values in each sampling period is never reported.

As a rule of thumb, if a WAN link his about 30% on 5-minute averages then it's probably spiking close to 100% multiple times during that period and the only way you're going to detect that occuring is to measure roundtrip ping times a few times per second.

64 byte pings can go from 14ms to 3200ms and back to 14ms in a 10 second period without tripping any alarms under high/spiky load conditions and it only takes a couple more percent to have it go out to 15,000ms pings and start having your helldesk lines light up.

This is why QoS (or using priority queuing) is important. Dropping low-priority stuff to the back of the queue can avert the phone calls or at least allow people to continue with important work.

It's even worse with satellite links. Back in the dark ages (1994) with an entire country on a 128kb/s link, overall monthly GB more than doubled when it switched from Sat to submarine link, but more importantly for the users latencies on interactive services dropped by 90% (www is not interactive)

[Alien8n]

Talking of testing...

I've told this before but it's still worth a retelling. Harking back to my engineering days we had a separate testing room for testing fibre optic transceiver modules. Back then most networks required 2 cables, one for transmit and one for receive, the company I worked for did some fancy optical engineering with shaped silicon that allowed the same fibre to be used for transmit and receive at a fraction of the cost of the standard setup. However, when testing outside of the cleanroom we would get spikes in test failures. Turns out that these spikes coincided almost exactly to 20 seconds before receiving a text message on an Orange networked phone. Solution, no phones allowed within a metre of the test equipment.

As an aside, the test equipment in the cleanroom also had a 1 metre clearance area surrounding it and a big sign saying not to cross into that area if fitted with a pacemaker. Reason that equipment never failed? All phones were banned from the cleanroom.

[Anonymous Coward]

You would be amazed how many IT people under 35 don't understand the concept of bits / bytes / octets / hex (except for network people). They never had to learn that stuff.

[Mike Lewis]

I remember being turned down for a job as a games programmer. They hired a COBOL programmer instead who didn't know what an interrupt was and didn't believe them when they tried to explain.

[Uncle Slacky]

About 10 years ago I had to explain the concept of big-endian/little-endian to a ~25yo French engineer with a 5-year degree under his belt, so the age limit checks out.

[jake]

Round about 2000 I started seeing "programmers" that didn't know what the heap and the stack are on a fairly regular basis. Nowadays it's normal for the youngsters to have many gaps of that nature in their education. I fear we are losing something very important that is going to prove to be almost impossible to get back.

(In a similar vein, a few months ago I had a child CEO ask me why I'd have a need to thump a TV with a screwdriver, but then CEOs have always been clueless ... well, most of 'em anyway.)

[Nick Kew]

@jake - Round about 2000?

Those of us whose education pre-dates computer science as a supposedly-serious degree subject[1] started with all sorts of gaps in our knowledge. Noone explained to me heap vs stack, I just figured it out on the job sometime back in the 1980s.

[1] Including some who later taught the subject, and whose graduates emerged qualified for things their professors never will be on the recruiters' tickboxes.

[Anonymous Coward]

"Those of us whose education pre-dates computer science as a supposedly-serious degree subject[1] started with all sorts of gaps in our knowledge."

What only became obvious in hindsight was that we were still pioneers. When we were given a task to do - it was often the case that no one else in our IT company had done that before. You had to have an enquiring mind, tenacity, and a capacity for lateral thinking.

My teenage hobby of amateur radio and electronics gave me a good grounding for working in the IT industry. My intended career had been in those fields - but serendipity put me into O/S support programming and technical system trouble-shooting.

[Mr Benny]

"What only became obvious in hindsight was that we were still pioneers. When we were given a task to do - it was often the case that no one else in our IT company had done that before. You had to have an enquiring mind, tenacity, and a capacity for lateral thinking."

Ditto those of us who were kids in the 8 bit era in the late 70s and 80s. If programming was a hobby and you wanted to get something done there was no google to turn to to find out how to do it and the hobbyist magazines were usually of limited use being more interested in software reviews and games. So you either figured it out yourself or gave up and did something else. I figured out a few things myself such as a selection sort without ever seeing it in a book, but some things eluded me such as recursion (but then I was using BASIC), and it was like a lightbulb appearing above my head when I eventually read about it.

[KittenHuffer]

To understand recursion you first need to understand recursion!

[Marshalltown]

Yes indeed

My own chosen caeer path included outdoors, rocks, dirt, a hammer and magnifying glass. I was forced by poverty to learn to build my own computer(s), and after my degree, was baffled to find myself installing coax for a 10BASETWO LAN, BNC connectors in the PCs, and terminators on the machnies at the end of the line, and later serving at the help desk. Between boughts of explaining the lesser mysteries of WordPerfect to my boss.

[jake]

"Noone explained to me heap vs stack, I just figured it out on the job sometime back in the 1980s."

You never wrote (or took apart) a compiler in school, just to see what made them tick?

[Nick Kew]

In school?

Computers existed, and were indeed beginning to appear in schools by the time I left. But not in the kind of school I went to, that relied on the state for its budget.

I had the impression you were of a comparable vintage yourself?

[jake]

Yes, in school.

Cubberly High School in Palo Alto had access to Stanford's computers. Pre-Internet. They extended that access to a few of us kids at the across-the-parking-lot Greendell Elementary. Wilbur, the Jr. High that most kids went to between Greendell and Cubberly, had an HP machine (I can't remember what it was) and access to Stanford. I started playing with compiler source in my early teens.

THAT said, when I referenced "school" earlier, I meant Uni ... which is a school, no matter how you look at it. (Yes, I know the difference in lingo across the pond. I usually make allowances for this, to avoid confusion.)

[AK565]

Re: Yes, in school.

I lucked into a course in Basic in the summer of 1979, while in middle school. The next time my education involved computers was grad school.

[Mr Benny]

Seriously?

Just how many school kids do you know who could understand the code to a compiler, much less write one? Where on earth did you go to school, Mensa Grammar?

[J.G.Harston]

Re: Seriously?

In my mid-teens in the 1980s I wrote a Forth system and started writing a Basic interpreter.

[BinkyTheMagicPaperclip]

Re: Seriously?

I have to admit I've never taken apart compiler code (although I might be doing soon, given that I'm not primarily a coder this fills me with a little fear), but whilst still at college I both wrote a bytecode interpreter in assembly, and tried to write another interpreter in BASIC (it was horribly slow, I gave up. At that point I didn't have access to a C compiler).

At the time I had more interest in using tools to create something, rather than creating the tools themselves.

[Marshalltown]

What computers

In college in the late seventies, where I did my degree, there were only punch card computers, and I personally never took a computer course. There were texts in the book store and a little later PCs and a bit after that Turbo Pascal. Bits of BASIC, Avenue (used in GIS but discarded by ArchGIS for Visual Basic). Bits of FORTRAN, PROLOG, ... Finally I got my hammer and hiking boots back, but, you can never really leave.

[Goldmember]

You never wrote (or took apart) a compiler in school, just to see what made them tick?"

Where I went to school (Lancashire, UK in the 1990s), we had a school assembly one day regarding 20 of our school PCs being "connected to the Internet", plus a permission slip home for parents to sign, stating we were allowed to use them.

The actual IT "lessons" consisted of "here's a page of text, printed out. Type it in Word" or performing a Yahoo search for your favourite PS1 game/ TV show (circa 1998). There were no IT lessons into GCSE time (after third year) as we simply did not have enough teachers to teach the courses.

So... No. I can honestly say we never took apart a compiler in school.

But to be fair, this total disregard for coding - and to be honest, general computing knowledge - has led me into a nice career as a coder... I've had to forge my own path and have not had to compete with many of my peers achieve it.

[Alan Brown]

" No. I can honestly say we never took apart a compiler in school."

Although I'd already been doing it for a while as a hobby, polytechnic level ELECTRICAL/ELECTRONICS engineering (3000 level - basically the same as 6th form) in 1984 had us creating "on paper" processors, writing assembly code and stepping through the programs to understand how they worked - along with doing some level of Boolean work and making gate arrays.

We then got to play with an (on paper) compiler to turn higher level code into assembler to see how it worked.

The lecturers (rightly) felt that it was critical that circuit and power monkies had a reasonable understanding of the guts of microprocessors because they were in the process of or about to turn the entire industry on its head (it depended which industry, but all aspects ended up being affected) . Most of us ended up spinning out in the computer direction anyway, via various paths.

"A hardware engineer blows PALs, a software engineer burns PROMs"

[gnasher729]

I started with Fortran 77. There was no heap, and there was no stack.

[jake]

"I started with Fortran 77. There was no heap, and there was no stack."

Not in the language. Perhaps[0]. The compiler is a whole 'nuther kettle o'worms. (Ask yourself "Why isn't Fortran self-compiling?")

[0] I prefer vi vs EMACS, more scope for argument :-)

[Anonymous Coward]

"I started with Fortran 77. There was no heap, and there was no stack."

Nor was there any syntax analysis whatsoever, leading to random code being compiled. Ok, I'm actually joking, this was F66 I'm speaking of, since 77 added this, as I remember.

Point is, it was already an obsolete language in the early 90s, despite some old beards telling us it was fundamentally faster than C, all benchmarks showing otherwise. There's no reason 1 out of 2 quite similar language would compile faster machine code than the other.

There were fanbois back then, already.

[MJB7]

"All benchmarks showing otherwise"

[citation needed]

If you are running large scale simulations, what you desperately, *desperately* need is speed - and for that purpose, Fortran does superbly well. It's not until `restrict` became widely available (and it certainly wasn't in the early 90's) that C had any hope of competing with Fortran - by the time it had, all the parallel processing extensions were available for Fortran (because that's what the code it would speed up most was written in).

[Doctor Syntax]

"why I'd have a need to thump a TV with a screwdriver"

Isn't it obvious? Providing you it it with the handle and not the blade it does less damage than a hammer.

[Anonymous Coward]

Round about 2000 I started seeing "programmers" that didn't know what the heap and the stack are on a fairly regular basis

First real job after graduating uni and I had to write a binary file processing program for an obscure system with no disk buffering and no OS level file buffering. Glad I took a "how operating systems work" course.

Source code for the main program was customized by a series of IBM consultants who didn't seem to have read the BASIC manual that clearly stated strings were passed on the stack. Not a reference, mind you, the string. Code was full of comments like "don't remove the next line, I don't know why, but leave it there."

[Alan Brown]

"In a similar vein, a few months ago I had a child CEO ask me why I'd have a need to thump a TV with a screwdriver"

Most people who did it never knew the whys either:

In the very old days (valve kit with components soldered to tags on sockets) it wasn't uncommon to find unsoldered tags (and dry joints)

In the less old days (before wave soldering rigs) the occasional unsoldered joint on a PCB would still get through QC (if there WAS a QC station). I once saw a report of one such joint that finally failed and went intermittent after 26 years in the field.

Even more occasionally it was a bad plug/socket, shitty cable connection. Less occasionally (on colour sets) it'd be fractured solder joints around the line transformer.

Percussive maintenance was never a good substitute for a proper fix.

[Old Used Programmer]

Take care what you argue over...

In a book my wife wrote, which I usually describe as "cyber, but not punk", one character refers a dispute as being big endian versus little endian. The copy editor at the publisher changed it to big and little Indian. In the proofs, by wife changed it back and put in a marginal note "See J. Swift". When she mentioned it to me, I noted that big vs. little endian is also a computer architecture issue. That was something she was unaware of when she wrote it.

[J. R. Hartley]

Re: Take care what you argue over...

Amiga users have been bombarded with big endian / little endian explanations for decades now.

[Nick Kew]

Re: Take care what you argue over...

@Old Used Programmer - your missus may be more authentic than us. Swift's usage (which I had forgotten until you mentioned it) pre-dates, and may be the inspiration for, ours.

[jake]

Re: Take care what you argue over...

http://www.catb.org/jargon/html/B/big-endian.html

[Rol]

You would be amazed how many IT people under 50 don't understand the concept of nibbles/BCD/2's compliment......

[Arthur the cat]

You would be amazed how many IT people under 50 don't understand the concept of nibbles/BCD/2's compliment......

I suspect the number of people understanding excess-3 representation is less than the population of a small village.

[A.P. Veening]

Re: Number of people

There are only 10 kinds of people, those who understand binary and those who don't.

[Arthur the cat]

Re: Number of people

There are only 10 kinds of people, those who understand binary and those who don't.

There are only 10 kinds of people, those who understand binary, those who don't and those who understand ternary.

T,FTFY.

[Dave 125]

Re: Number of people

And in general, there are 10 kinds of people: those that understand n-ary, those that understand (n-1)-ary, those that understand (n-2)-ary, ..., those that understand ternary, those that understand binary, and those that don't understand any of this.

[Anonymous Coward]

You will therefore, be pleased to know that Open University students on the IT degree programmes do study 2s complement arithmetic along with gaining an understanding of bits, bytes and nibbles. All covered in the first 2 compulsory modules TM111 and TM112. They also start off learning how to code by decomposing a problem into an algorithm.

[Doctor Syntax]

"You will therefore, be pleased to know that Open University students on the IT degree programmes do study 2s complement arithmetic along with gaining an understanding of bits, bytes and nibbles."

S100, the original OU science foundation course, provided students with a simple balance.

My problem students - a couple of teachers - couldn't understand why a foundation course would use this when they had digital scales at work. The magic lure of numbers! Could I get them to realise they were relying on all sorts of things they weren't able to see such as the linearity of transducers?

When we bought a digital balance in my lab the first thing I did was to check it out with some standard weights borrowed from the local Weights and Measures dept.

[Alan Brown]

"My problem students - a couple of teachers "

You didn't really need to explain much further than that. My parents are both teachers and I grew up around hundreds of them.

There are more than enough of the kind you're invoking to justify the trope "Those who can, do. Those who can't, teach". It was even worse when you'd find obvious errors in textbooks (vs multiple other sources) and teachers who'd go with the textbook anyway because what's printed is always correct isn't it? (even for things like basic algebra where a calculator showed the errors).

It was frustrating for me (as a 15yo) to have to explain issues of basic physics to some teachers who didn't understand what they were trying to teach 7-8yo pupils out of textbooks (math usually SCARED them), have them apparently "get it" and then realise they'd still managed to get things mostly bass-ackwards - but at least they weren't so "scared" of science and math.

There are _SOME_ brilliant teachers - quite a lot of them in fact. The problem is that there are also a large bunch of dunces who saw it as an "easy career option" and a collection of martinets who should have been weeded out as fundamentally psychologically unsuited to the job before they finished training. Guess which ones tend to rise to the top in the administrative side?

[A. N. Onymouse]

"S100, the original OU science foundation course, provided students with a simple balance."

That goes back a bit. I did S102 in 1989.

We did get a balance and weights in the home experiment kit.

[Vincent Ballard]

Why should IT people understand BCD? Only a very small percentage of them work with VAX nowadays.

[jake]

Why should IT people know how to type? Only a small percentage of them work on text-only interfaces nowadays.

[Trixr]

Yes, har har, but if you can get thru 5 minutes of a real IT task without typing, you're a character in Minority Report.

[Scene it all]

or IBM 1401

[phuzz]

It's not surprising that younger people have never learnt about binary/ocatal etc. as they've never needed to.

Pretty much the entire history of IT over the last thirty something years has been to abstract away the hardware level details, and to just present the user with a higher level view.

Or to put it another way, you can have an entire career in IT without ever needing to understand binary, and personally I don't see that as a bad thing.

[Anonymous Coward]

"[...] you can have an entire career in IT without ever needing to understand binary, [...]"

If you do anything that does a calculation then you should be aware of the constraints imposed by the hardware.

eg.

1) the effects of numbers exceeding the limits of their 8, 16, 32, 64, ... bit variables.

2) the different limits of signed and unsigned numbers

3) the rounding of floating point numbers depending on the machine's precision

4) doing calculations in the right order to avoid losing an intermediate value due to rounding

[Simon Reed]

One of the call loggers on the Helpdesk said "The call reference number has got a dash in front of it". Another pointed out "And it is going down".

"What are we up to?" I asked. One said "It is -32766 now".

I called over the annoying boy who had written the bespoke Helpdesk system. (He had recently completed his IT degree and wrote the system during his work placement year.) "Change the call reference number to unsigned, and, when you get the chance, make it a longint in the database and the code". He looked at me with his mouth open.

They all thought I'd done magic + was psychic + must have seen the code.

I had never seen the code and did not know what language it was written in. But I knew the call reference numbers were in the range 32000s and wondered if something interesting might happen.

That annoying child with his IT degree wrote horrifically slow and inefficient code. But hey, that doesn't matter these days, does it? Just keep putting in faster processors, more RAM and bigger disks...

... to do what we did in 32K at under 2MHz.

[jake]

"... to do what we did in 32K at under 2MHz."

My first PDP11 (LSI11, actually) ran a flight simulator in 64K. Kind of makes Steve Jobs' claim that "128K should be enough for home users"[0] sound somewhat more sensible than it sounds today,

[0] He was demoing the original Macintosh at the Homebrew Computer Club, in late 1983.

[Anonymous Coward]

"Just keep putting in faster processors, more RAM [...]"

In the late 1960s a series of 3rd generation mainframes started at 16KB - through to a rare beast with a massive 1MB of memory.

The 1MB mainframe gradually exposed errors in the coding as the size of parts of the O/S grew in new releases.

When parts of the O/S went above 32KB there were crashes because address manipulations had used 16 bit signed arithmetic instructions.

When they went above 64KB it crashed because it then required 32 bit arithmetic instructions to manipulate addresses.

[VikiAi]

About 10 years ago, our standard desktop deploy went from 2G RAM to 4G RAM, and suddenly a particular bit of software started refusing to run, reporting insufficient RAM available. Much gnashing of teeth until I suggested my manager ask the supplier if they were using signed 32-bit integers in their code that checked how much RAM the system had available! (I am not a trained programmer, but I know 6502 ASM and all the associated low-level stuff from my childhood and have kept myself casually up to date since).

[Alan Brown]

"suddenly a particular bit of software started refusing to run, reporting insufficient RAM available."

We had problems crop up with different answers code was giving between 32 and 64 bit OSes.

It turned out that BOTH were wrong, and it wasn't a case of one being less wrong than the other.

In that case it was because floating point answers were being rounded to the nearest 32/64 bits and then fed into the next iteration of the same loop.

After a while rounding errors add up and your proto-solar system flies apart after 10 million years or your red supergiant star blows apart when it shouldn't. Solving that issue answered a number of long-standing astronomy questions (but raised more, because CPU floating point precision STILL isn't good enough to run simulations for billions of years)

[IanRS]

Octal problem

When I used to interview graduates for a coding position I liked this test (although back then it was in C rather than C++).

What is the output of the following code:

#include <iostream>

int main()

{

std::cout << 001;

std::cout << 010;

std::cout << 100;

return 0;

}

Most candidates realise the leading zeros will not be printed, and write down 1 10 100 on separate lines. They are wrong. Some candidates realise that the output has no \n and write 110100. They are still wrong.

[Conall O]

Re: Octal problem

0 I presume ; call the method, get the return value. Ignore the shiny distractions.

[BinkyTheMagicPaperclip]

Re: Octal problem

Conall: No, << redirects the value to an output which in this case is stdout (std::cout)

A leading zero in C/C++ means 'Octal'. Therefore the first number (001) is in Octal, but it doesn't matter because under all number bases it still outputs 1. The second number is in Octal (010), and as base 8 runs with digits 0-7 per digit position, it evaluates to decimal 8. The third number (100) has no leading zero, so it's decimal and outputs as 100.

Therefore 18100

[Alan Brown]

Re: Octal problem

> leading zero in C/C++ means 'Octal....The second number is in Octal (010), and as base 8 runs with digits 0-7 per digit position, it evaluates to decimal 8

Which is all fun and games until someone writes 010 as part of an IP address and your code interprets it as octal, but the RFC states that IPv4 address dotted quads are ALWAYS decimal.

And then someone else (a certain Mr Vixie) says that the code is working correctly.

And someone else points out that Mr Vixie is the same person who wrote the RFC AND wrote the code

And when someone else asks that either the code or the RFC be adjusted to agree with each other, a religious war breaks out.

[BinkyTheMagicPaperclip]

Re: Octal problem

That really is an awful test. Yes, I can pass it because I remember the octal syntax, but unless you're in a specific niche that uses octal I've probably used it once or twice, because a specific function specified values in octal.

Surely a test should be to check reasoning and logic, not to trip people up with less commonly used language features.

Likewise I've moved beyond using crap such as ++i in compound statements. Spell it out, perhaps one day you'll be debugging your own code whilst suffering a heavy cold/hangover, and won't be in the mood for fancy syntax trickery.

[PerlyKing]

Re: Fancy syntax trickery

Maybe I'm showing my own age here, but in a C-based language I wouldn't consider ++i to be "fancy syntax trickery".

[BinkyTheMagicPaperclip]

Re: Fancy syntax trickery

True, I was more thinking of things like

result=foo_bar(++i,j++);

Apart from in loops, it's best to split the variable modifications out.

[TrumpSlurp the Troll]

Network tests

Not me but some colleagues were given a very large user test file to send over the demonstration network.

Much confusion when the transmission seemed to take a couple of seconds.

Turned out the file was the same pattern repeated many times and the comms kit had link compression enabled. So what was sent was the pattern and the number of instances.

[Anonymous Coward]

Re: Network tests

"Much confusion when the transmission seemed to take a couple of seconds."

Most people in IT have always been unaware of what happens beyond the interfaces of their particular area of concern. This has been exacerbated in recent years by the need for repeated renewal of qualification certificates. Job roles are usually predicated on these certificates or narrow training - so people usually don't get the chance of a wider experience.

We had similar problems when roles were outsourced to India. Thick client transactions that had been apparently instantaneous in the UK - now crawled due to the latency on the many RPC-type handshakes in each transaction.

There were demands from management to make the network go faster. To which we had to respond - that even if the transmission was at the speed of light - the transactions would still be slow. You can't change the Laws of Physics.

[Anonymous Coward]

Re: Network tests

> Most people in IT have always been unaware of what happens beyond the interfaces of their particular area of concern.

The relevance of this never ceases: https://www.joelonsoftware.com/2002/11/11/the-law-of-leaky-abstractions/

[Anonymous Coward]

Re: Network tests

"The relevance of this never ceases"

One problem is that learning about things outside your necessary area takes time and effort. If you want promotion then you invest that effort in focusing on a management approved skill. Good generalists often only get valued in instances when they use their knowledge and skills to pull a rabbit out of the hat.

In my 40 years as a generalist trouble-shooter there were many promising youngsters who could have taken over my mantle. They soon decided that my role was hard work and underpaid - definitely not a quick career progression path.

Management often regarded me as a necessary evil to solve the problems that had foxed the highly paid specialists. My refusal to fudge the results was not welcome with some people.

[Ozzard]

We need more generalists - join us, jooiiin ussss!

We need more generalists - increasingly difficult as the body of knowledge in IT grows by the day. Can I interest any readers in taking it up as a career? :-)

It's a really enjoyable job if you love learning, thinking, communicating, and having a pretty good idea of where organisational cock-ups tend to originate (hint: who would benefit from this situation?) and hence where to look for the root cause of the gotcha. Definitely not all technical, and really fulfilling to leave an organisation with people knowing more, communicating better, and (ideally) less siloed and more able to deal with future gotchas without external aid.

[Anonymous Coward]

Re: We need more generalists - join us, jooiiin ussss!

Never certain if I'm a general specialist or a special generalist... Neither seem to have a defined place in any org. chart my (soon to be ex-)employer could create.

[Justy]

Re: Network tests

The curse of the good generalist. Able to troubleshoot, fix stuff and save the day. At best receives a begrudging respect, usually just tolerated, sometimes seen as a troublemaker. Very presence reminds management of their bad decisions or mistakes. Kept around because in the final analysis indispensable.

[Alien8n]

Re: Network tests

I was always under the impression that if you learnt a little about a lot of different areas of IT you were only good as an IT manager.

Now I am the IT manager it seems I was right...

[Alan Brown]

Re: Network tests

"Management often regarded me as a necessary evil to solve the problems that had foxed the highly paid specialists"

On the other hand, when I was in such a role I was valued by management (less so by the specialists who would get called out to explain shortcuts by management when I'd finished working out what went wrong - in an organisation fixated on eliminating repeats in implementation problems the pushback may not come from management). A troubleshooter is a valuable employee when used correctly.

One of the more telling organisational problems is that "specialists" often start out as being extremely knowledgeable "specialist generalists" but over time the job moves to be inhabited by inexperienced "box ticking exam-passers" who eventually become a liability to the organisation and see anyone who knows what they do in any detail as a threat to their existence because all they know how to do is tick boxes and not address issues holistically.

You can see this in the increasingly manic singleminded focus on speeding across several countries where the last 15-20 years of more rigorous enforcement have resulted in virtually no (in come cases negative) road safety improvements. It's gone past simple self-congratulatory back patting about getting speeds down in areas where it wasn't making any difference to casualties, and into outright statistical manipulation of figures to make things look as though they've changed when they haven't and going into full attack mode when called out on casualty figures remaining unchanged (or increasing).

[Mike Pellatt]

Re: Network tests

That "always" bit is so true.

Sometime around 1984, I was at seminars or training (I forget which) on the new stuff in Unix System V Release 3. This included AT&T's "competitor" to NFS, RFS (Remote File System) together with, of course, a whole new layer to support it - FSS (File System Switch - which Unix couldn't do before that). I remember asking the RFS guys a question over the FSS functionality which would clearly have significant impact on RFS (again, I forget exactly what - but it was critical). The guys said they knew absolutely nothing about the FSS implementation, just the interface they were developing against.

Doomed, Capt Mainwaring, we're all doomed.

[Anonymous Coward]

Wrong City

This happened twice to me, I had a meeting in Newcastle (before the days of Sat Nav's) anyway - upon arrival in the outskirts of Newcastle after a l-o-o-o-n-g drive North, I went an bought a local A-Z in a petrol station and spent a fruitless 15 mins in the car trying to find my meeting address. In desperation I went back to kiosk to ask and they pointed out that the address was for Newcastle-Under-Lyme and this was Newcastle-Upon-Tyne. Needless to say, I didnt make the meeting. Who knew there were 2 Newcastles...

On another occasion, I was booked to do some performance tuning work for a large drugs company in St. Petersburg. Anyway, the flights were booked for me and I was told that a limo/taxi would meet at the airport. Needless to say, no-one was waiting for me at the airport. Who knew there was a St Petersburg in Florida!!! That one took a bit more sorting out.

[jake]

Re: Wrong City

About a billion years ago (Internet time, call it 1985ish), I was booked on an "emergency" flight to LAX to fix some computers for Disney. I got the call at noon, was in the air by 1PM. Unfortunately for me, the Disney offices were in Glendale, so I should have been flown into BUR ... which might as well be on another continent at 4PM on a Wednesday if you're using wheeled transportation.

Fortunately, helicopters exist and I was only 2 hours late. My fault, naturally.

Did the job, staff came in Thursday morning & were happy with the change, customer signed off on it at noon, and I was home in time for supper. Job well done, right? Maybe not ...

The PM (or necropsy, as I prefer to call such things) showed the temporary secretary assumed that everything in the LA area had to go through LAX ... but it was my fault anyway, as I should have known better & flagged the bogus destination before I boarded! This despite the fact that my instructions were oral "Take this briefcase full of hardware, fly to LA and install it. You will be met at the airport and taken to the site. Further instructions by telephone will follow." I didn't even know the name of the destination company until I was on the ground in LA. But my fault, so no bonus for the emergency call. It was right about then that I started thinking about going freelance ...

[Anonymous Coward]

Re: Wrong City

A few years ago I had to send someone in Scotland to pick up some kit and decided to be helpful and give an ETA. Went to Google Maps and entered Perth and Stirling and it said 'no route found'... huh?... they are only a few miles apart?!? It turns out that Google UK thought I meant Perth in Australia and for some strange reason couldn't find a suitable land route

(and at one point they also confused Wood Green in North London with Woodford Green in Essex)

[A. N. Onymouse]

Re: Wrong City

"On another occasion, I was booked to do some performance tuning work for a large drugs company in St. Petersburg. Anyway, the flights were booked for me and I was told that a limo/taxi would meet at the airport. Needless to say, no-one was waiting for me at the airport. Who knew there was a St Petersburg in Florida!!! That one took a bit more sorting out."

When travel departments foul up they do it spectacularly.

I was booked to do a training session in Hong Kong with a telecoms company. Travel dept sorted it all out. I was leaving in the morning, arriving about 12 hours later and the training was the next morning.

Great.

Except that they forgot that Hong Kong is 8 hours ahead so instead of landing at 8pm, we arrived at about 5am. With the training session starting at 8am we barely made it out of the airport and into the office in time. We barely shambled through the training looking like unmade beds on zero sleep and far too much free in-flight alcohol.

Jet lag is one issue but jet lag, sleep deprivation and sobering up while awake, talking and being watched by a room full of polite but slightly bewildered people is not a great experience.

The post script is that we just barely managed to get through the day and then they insisted in taking us out for dinner and drinks afterwards.

Oh and they also managed to get a hire car delivered to my colleague's door in Blackburn. He lives in the one in west Lothian. The car ended up outside a house with the same street name in Lancashire.

[Anonymous Coward]

Vindicated

My workplace names computers after landmarks in the country that one partner is based in. It's a life hack to make sure different departments have different naming schemes and avoid overlap.

We recently got the first of a new type of server and I stamped down hard on the suggestion to name it after a city, anticipating that the Vienna situation would result if we did.

[Anonymous Coward]

Re: Vindicated

"Vienna" is also a confusing name because it means nothing to some people.

[jonathan keith]

Re: Vindicated

Ah, shaddap you face.

[Anonymous Coward]

Re: Vindicated

That may have been my slowest upvote ever. Well done.

[chivo243]

Re: Vindicated

Mmmmm Vienna sausages arrgggh!

[Pascal Monett]

Re: Vindicated

And what exactly is "the Vienna situation" ?

I'm just a programmer. Please enlighten me.

[ibmalone]

Re: Vindicated

The "Mistaken identity" story from the article.

[Admiral Grace Hopper]

Re: Vindicated

“Why did you call her Vienna?”

“Leave the door open and whoosh! It’s goodnight Vienna”.

[\Rigsby]

[Keith Oborn]

Re: Vindicated

Throughout my working life (I recently retired) I harboured a secret Unix hostname convention. I never dated use it:

1: Name your hosts after common Unix commands

2: Write complicated and undocumented shell scripts to do mysterious things between them

3: Resign

[OopsSorryMyBad...]

Re: Vindicated

That's evil. I love it.

[Anonymous Coward]

Re: Vindicated

I would have been tempted to use this when I left Serco, but I think it would have been an improvement on where I came into the role.

[Arthur the cat]

Re: Vindicated

Throughout my working life (I recently retired) I harboured a secret Unix hostname convention.

In one company I named the machines after the elements(*) because they have obvious numbers and convenient 1 or 2 letter abbreviations. I also added "me" as the 2 letter version of localhost. This all worked fine until I had complaints that people couldn't remotely access mendelevium. It turned out everyone thought the 2 letter version was "me" rather than "md" and were baffled when they kept finding themselves on their own machines rather than the server.

(*) until we got too big and the names got a little random. "tupperware" was one "element".

[Ozzard]

Naming conventions

I know one person who named his hosts using common pronouns such as it and they, with job runners such as something and nobody. Conversation about these systems rapidly became impossible.

[Nick Kew]

Re: Naming conventions

Not entirely original. Homer got there first, where Odysseus outwits the Cyclops. But still a jolly wheeze.

[Public Citizen]

Re: Naming conventions

I see much potential here for a digital version of an Abbott and Costello Routine.

[J. R. Hartley]

Re: Vindicated

"Throughout my working life (I recently retired) I harboured a secret Unix hostname convention. I never dated use it:

1: Name your hosts after common Unix commands

2: Write complicated and undocumented shell scripts to do mysterious things between them

3: Resign"

And make life hell for the PFY. Never change.

[jake]

Re: Vindicated

I ran across a new University teaching network where the UNIX servers were named after the ASCII control characters (NUL, SOH, STX, ETX &etc.). It was unveiled during the summer session, to great fanfair but not much by the way of testing. Much hilarity ensued.

[PerlyKing]

Re: Naming conventions

Back in the good old days when computers had names, my team had to give all of our desktop machines (SPARCstations of various flavours) two names each, one for each network to which they were attached. Being nearly all young men, one side was named for missiles and the other for guns. I think mine was rapier/colt or some such. Our boss was away at the time so we had to choose for him. Being a good South Londoner we settled on halfbrick/sawnoff. Which he loved :-)

[Anonymous Coward]

Re: Vindicated

Some will know to whom this refers.

The sysadmin got the first sun workstation and named it 'Ariel' intending that the later ones for the team, would also be named after Shakespearian characters from the 'Tempest'

Much to his chagrin, he was on holiday when the next four machines - promptly named 'surf' 'bold' 'persil' and 'tide' - were delivered....

[Anonymous Coward]

Re: Vindicated

Sort of expected Little Mermaid characters.

[J.G.Harston]

Re: Vindicated

I once set up some machines named after local towns, alloa, sterling, dollar... guess what happened next.

[Anonymous Coward]

Re: Vindicated

Euro?

[DiViDeD]

Re: Vindicated

Back in the day, my employer named all their unix servers after fossilised pop groups (beatles, everlies and such). My colleague, a big Babylon 5 fan, was excessively pleased with himself when he persuaded them to call a new box "Shadows".

[Alien8n]

Re: Vindicated

My home PCs have always been named after various afterlifes.

The main PC has always been called Hades (this dates back to the early Usenet days and Demon Internet, where you could edit the header information in Demon's DOS email software. I think it returned "[e-mail]" from memory).

Laptops have varied between Pheonix, Valhalla, and Nirvana.

[Alan Brown]

Re: Vindicated

"My home PCs have always been named after various afterlifes."

Probably the most famous set of machines on the Internet were named after Bloom Country characters.

I've seen a set named after Rocky Horror characters. Apparently that started after a system was built and named Frank(enstein). Eddy ended up as the core of a group named after HHGTTG characters.

[W4YBO]

"ran like a three-legged dog"

Being from the southern US, I never really gave this phrase much thought until a few weeks ago. That's when I saw my neighbor's three-legged Malinois (missing left front leg) run down a 30-35 mph bicyclist in front of our house.

Yes, I'm that guy that counts seconds while speeders traverse two utility poles. If I lived in town, I'd be yelling "Get off my lawn!"

[Anonymous Coward]

Re: "ran like a three-legged dog"

A neighbour's cat had lost a rear leg. It was interesting to see it stalking birds. When it launched itself to pounce the unbalanced thrust made its body rotate in the air.

[John 110]

Re: "ran like a three-legged dog"

I once had the gob-smacking experience of watching a three legged Jack Russell lift it's only back leg to pee. I'm convinced it manipulated gravity to do it.

[Criggie]

Re: "ran like a three-legged dog"

Laws of Rocketry - just think of supporting the weignt on the thrust power alone....

[rcxb]

Re: "ran like a three-legged dog"

Had a 3-legged dog, myself. Missing a rear leg. Ran circles around everybody. Walking along slowly was another matter, entirely. He had two speeds: 1. sit down 2. run like mad.

[Mpeler]

Re: "ran like a three-legged dog"

A three-legged dog sidles into a bar.

Bartender asks, "What can I get for you?"

Dog replies, "I'm-a here to find the one that shot my paw"...

[ma1010]

The Force is strong with this one

@"Clyde"

While I was working away, he suddenly interjected, 'OK, well now it definitely needs to be replaced, it's on fire. I'll call you back'."

Well done! A result that would gladden the BOFH or PFY, and you did it remotely on your FIRST EVER HELLDESK CALL! You, sir, are a true BOFH, and I salute you. Have a virtual pint on me.

[Stevie]

Re: The Force is strong with this one

Seconded.

Aside: Are there any funnier words in the English language than "Flames shot out ..."? They always crack me up, even if I was the one beating out the carpet fire or staring at the remains of the PC.

[NateGee]

RE: Getting one over on the boss

Sounds like the IT management where I currently work. Everything is set to old, outdated procedures (which also aren't kept up to date) and heaven forbid anyone other than them should come up with an idea to streamline these processes.

Only been there 2 months. Wonder if I'll make it to Easter before jumping ship?!?!?

[Adrian 4]

Re: RE: Getting one over on the boss

The best ones are where there are old outdated procedures in the documentation, but everyone follows different ones (that aren't written down) because the old ones don't work.

[Swarthy]

Re: RE: Getting one over on the boss

The worst ones are where you are writing software to streamline some of the old processes, and show the initial design to be told "No, do it <stupid way> so it's more like our current procedure."

It's a requirement, so you do it their way, and when the application gets to User Acceptance Testing, ALL of the users make a comment like "Why did you do it that way, we don't do it that way. It would be better if it did it <originally designed way>."

[Anonymous Coward]

Re: RE: Getting one over on the boss

I find that if I state the better way of doing it once and don't push it, then the customer will often come back to me 3-5 working days later with a new idea that is suspiciously like what I told them.

[Yes Me]

Re: RE: Getting one over on the boss

Yes. This is mainly why IPv6 has proved hard to deploy despite working better than IPv4...

[Anonymous Coward]

Re: Re: RE: Getting one over on the boss

Yes Me,

You say '... better...' I say 'different' !!! :)

[Dimmer]

Re: RE: Getting one over on the boss

Just a question, how many IT guys are securing IP v6 or turning it off? And how many are ignoring it?

[Conall O]

Re: RE: Getting one over on the boss

if IPv6 was more human readable, i'd agree, but for now I much prefer have ipv4 on LAN and only using ipv6 for gateways

[Nick Kew]

Re: RE: Getting one over on the boss

Once upon a less-than-happy time, I was tasked with setting up a company mailserver to keep copies of everything (company afraid of lawyers).

On paper.

With all attachments. OK, I made that one up: they didn't think of attachments, but I did, and pointed out some issues with them.

[Keith Oborn]

Re: When things could be worse

Years ago, in the early days of Thinkpads, IBM had a wonderful network of local service centres. I used them a couple of times, and their fix times tended to be quicker than the 30 minutes it took to get back to the office.

One day, a sales guy called me: "I'm in Glasgow, got to do a presentation in a couple of hours, and the laptop power supply has died".

So I call IBM, who said "There's a service centre about 20 minutes from him. Give us his details and get him to drop in".

I call him back. He'd switched his phone off ;-(

20 minutes later, IBM rang: "We've spoken to your colleague" (ME: "wow, how did you manage that?"). "

"We can swap his power supply, but he has to bring the dead one in."

ME: "??"

"He says he left it in the office" (that was in Byfleet, Surrey)

ME: "Thank you so much for such superb service. I will KILL the idiot next time I see him".

IBM: "<giggle>"

(the power supply was, needless to say, on his desk and fully functional).

[Prst. V.Jeltz]

Re: When things could be worse

(the power supply was, needless to say, on his desk and fully functional).

Saw that coming. I've dealt with many calls , usually from meeting rooms , where people have plugged laptop into a switched off socket , and then , surprise , "it just went dead"

[StuntMisanthrope]

Just a Nibble, it'll come in useful;

Since it's twixtmas (ruined) I might add, there's the one from an old message bard, about the Vaxen '87 debacle, the long-legged original schmetterling (for hire) and would you adam and eve it, the artisan french one inbetweet BBS and and t'Internet, I had to down an armag.nac to swallow that one. #penfriends #_PFY3.14.sylvester

[Criggie]

Re: Just a Nibble, it'll come in useful;

https://media1.tenor.com/images/a5fe45988180d25ac3efa32f53bca743/tenor.gif?itemid=9682982

Que?

[Chairman of the Bored]

Need a little personal help here

The Marx bank is a type of high voltage generator, sometimes used to drive high power RF sources such as cavity magnetrons - frequently called "maggies"

When a Marx is fired one says it "erected"; google a schematic and you will see why.

So faced with a maggie that failed to thrive, a radar that wouldn't work, and a deadline ... I sent out an email to damn near everyone asking something like: "The e2v maggie is still unhappy. I don't think the pulser is getting an erection. help?"

Best answer: "take matters into your own hand"

[Suricou Raven]

Re: Need a little personal help here

I've heard a story in circulation regarding a stock exchange installing a email/web filtering system to keep out anything which might leave them exposed to legal troubles for harassment or a hostile work environment. The story goes that it was turned on, and five minutes later someone from the trading floor bursts through the door screaming "Turn it off, we were trading rape futures!'

[Chairman of the Bored]

Re: Need a little personal help here

Trading rape futures? That's a hell of an edge case for the filter! Sometimes I will do a quick check to see if anyone is actually reading statements of work or work packages by including a Software Hardware Integration Task (SHIT) Job somewhere in the hundreds of lines of a WBS. Occasionally a filter will catch it. Far more rarely, a human will.

[jake]

Email is ever so helpful ...

I once sent a rather steamy love letter to my Boss, and a system status report to my girlfriend (now Wife). Thankfully, they both found it hysterical. No harm, no foul. An extra pint that evening helped :-)

[Anne Hunny Mouse]

Re: Email is ever so helpful ...

I once sent an email to the wife ending Regards.

Hell hath no fury...

[steviebuk]

I remember a fix...

..that I'm still proud off regarding explorer freezing on the network where I worked. You'd load explorer and pick a network drive and thats it, you now had to sit there waiting for at least 5mins before it would populate or respond. Was only happening on certain devices, namely laptops.

Anyway. I wasn't allowed to troubleshoot "Just rebuild the laptops and move on". I explained that's not fixing the issue. Surely we should be given the time to work out what is causing it. Then we don't have to waste time reimaging. The fix might be quicker. Nope. Still told to reimage or give a prebuilt one off shelf but even that took time. I believe we were still on XP and this wasn't that long ago, probably 5 years ago.

I'd discovered Sysinternals tools. Watched loads of older videos on it that Mark Russinovich would do at TechNet. This was a perfect issue to try out with the tools, to help me learn how to use them. I carried on bugging my manager. I FINALLY got permission to just use her laptop then. She had finally been hit with the issue and was off on leave for 2 weeks. So I sat down and lucky the issue popped up pretty quick. Within about 20mins I'd found the issue and fixed it (well technically it was a work around).

Watch explorer with task manager and you could see it running at 50% but thats it and the reason task manager is bollocks. Using process explorer allowed me to look at the threads of explorer at the time and there it was and .dll file from the encryption software we used back then running at 50%. Doing a seach online to see what this .dll actually did, I discovered all it did was search the network for any files that were encrypted so it could change the display icon to show the file was encrypted. Well we never encrypted actual files, just the whole HDD and only on laptops. Reading the guide on the .dll it was fine to disable it.

.dll disabled in a few seconds and the problem went away. Nice. A few second fix to replace an hour reimaging a laptop. Even taking a users laptop away and giving them a prebuilt replacement took up time. So I was proud of my fix and have loved the Sysinternals tools ever since.

[Antron Argaiv]

Re: I remember a fix...

Latest Win10 update to my work laptop has Cortana chewing up processor cycles. Fan coming on intermittently (never used to do that). MS doesnt want you to turn Cortana off. Bloody idiots.

Do fhey really expect office workers in open plan offices to be talking to Cortana?

Nothing would get me to change jobs quicker.

Googled and found that disabling search is the only way fo disable Cortana.

Happy New Year, and thanks to all at El Reg for the interesting and wildly humorous tales.

[Conall O]

Re: I remember a fix...

Win10 LTSB has it completely off I believe.

[damworker]

Oracle Uk have their HQ just outside Reading, Unfortunately the main shopping centre in Reading is called The Oracle. Many years since I worked for them but it was a regular occurrence to get a confused (usually foreign) person call asking for the correct address or turning up an hour late having gone to the shopping centre.

You get in a Taxi at Heathrow and ask to be taken to Oracle, Reading and the taxi driver nods.

We did warn people but it got forgotten.

[werdsmith]

I would have preferred the shopping centre.

[CFtheNonPartisan]

The Hot Shot Database Team

3 fine high priced consultants from the sub-continent were doing some database development. They would 'put data in the database' and then demonstrate 'retrieving data from the database'. Management was impressed by the speed of their database implementation, until it was rebooted and for some reason every time the system was rebooted the previous session's data disappeared. A mere hack of a programmer on-staff realised the hot shots did not know the difference between putting data on the stack as compared to storing it in the database. The hot shots were not sacked and the staffer got no recognition.

[Anonymous Coward]

Re: The Hot Shot Database Team

Databases are great fun, despite the perception that they're all just data black boxes and can all be treated the same they are most certainly not and they all have their own little foibles which you only learn as you crank up the demands.

1. SQL Server developers not used to Oracle. SQL server will auto-commit non-transactional DML, Oracle will not by default. 3 hours of loading data only to find no data has been pushed through.

2. Disabling auto-commit on Oracle processes to speed up the processing, then either blowing the trx undo segments or forgetting the final commit and losing all the inflight data.

3. Developers committing after every single DML operation on Oracle on millions of DML ops and wondering why the process runs like a 2 legged dog against the wind! ( If anyone dosn't know why then I suggest you learn about Oracle's redo log writer mechanism very quickly! )

4. Tripping over their own feet during looped DML code and wondering why something as simple as single update inside a loop with commit will keep failing no matter how huge the database memory, CPU and rollback/undo segsments are made. This one is bloody good fun with developers, they scream and rant about it, demand more resources, more money be spent and all you can do is keep calm and keeping pointing them to the Oracle docs about 1555's as there's absolutely nothing that will fix it other then fixing the code!

[werdsmith]

Re: The Hot Shot Database Team

Databases are great fun, despite the perception that they're all just data black boxes and can all be treated the same they are most certainly not and they all have their own little foibles which you only learn as you crank up the demands.

Oh yes. People will throw compute power and RAM at slow databases when 9 times out of 10 .......

Developers are often developers in their favoured or project language and treat the SQL going to the DB as something that should just work.

[MacroRodent]

Re: The Hot Shot Database Team

Reminds me of the about only SQL application I made, calling MySQL from a command-line program. Fairly simple, just two or three tables. One operation was entering a lot of record from a file into the tables. Worked fine and quickly locally for even a quite large file. Released it and soon people from a China office complained the enter from file operation took hours. Turned out the latency killed it, the db was in Finland, so each MySQL command paused for the round trip, waiting for response, and several were needed per each entry! But fortunately I found a fix that made the speed tolerable: pack the per-entry operations into a stored procedure, and call that with the data to be entered. It was also possible to send several (around hundred) of these calls in one MySQL api call.

[Anonymous Coward]

Re: The Hot Shot Database Team

"Turned out the latency killed it, [...]"

A customer had a department who insisted on writing their own application in Visual Basic. The local network was blamed for the client taking 30 seconds to return one line of results.

I took a network trace and showed them the latency was 1ms. The Visual Basic application on the client was issuing 30,000 requests to the server objects for that one operation - rather than caching some objects locally.

[Herring`]

Re: The Hot Shot Database Team

Turned out the latency killed it

I remember going on a (rare) on-site visit with a user of our software. Watching them operate it, there was one bit of processing where they copied everything to the local drive, ran the processing, then copied it back to the network. Intrigued, I asked why. They said that on the network drive, the processing took half an hour instead of a few seconds.

We'd not seen this but I did spot one thing: they were on token ring and we were on ethernet. Getting back to the office, I investigated and found that one of my colleagues had decided that the best way to write a bunch of binary data to disk was looping through the buffer writing one byte at a time - instead of just one write operation for the whole thing. The effect on ethernet wasn't really noticeable but on token ring, well it had to wait for a token for each byte. 30 second fix once found. The guy who wrote that had a particular talent for coding obscure bugs, hiding a stupid decision behind many layers.

[Anonymous Coward]

Re: The Hot Shot Database Team

> Oh yes. People will throw compute power and RAM at slow databases when 9 times out of 10

I had a developer team no less[1] ask for a test VMware server to be created with 10GB of RAM because they thought all the data needed to be loaded into memory before they could loop through it doing what ever it was (some trivial summation type calculation).

[1] I say team - it was actually just two people but I was shocked that neither had the first clue about relational databases.

I offered to write the code for them but being a mere "infrastructure architect" at the time I wasn't allowed to.

[ShortLegs]

Re: The Hot Shot Database Team

>I had a developer team no less[1] ask for a test VMware server to be created with 10GB of RAM

> because they thought all the data needed to be loaded into memory before they could loop through it

> doing what ever it was (some trivial summation type calculation).

>[1] I say team - it was actually just two people but I was shocked that neither had the first clue about

> relational databases.

>I offered to write the code for them but being a mere "infrastructure architect" at the time I wasn't

> allowed to.

Um, you didn't work at an organisation called, um, "steel hill", or "ferrous geographical feature" or "metal molehill"....

[steviebuk]

Re: The Hot Shot Database Team

Sounds like the culture in a place i used to work

[OzBob]

Someone else fixed my fix

I wrote a C program (which I am really pants at) to update the personnel table in a helldesk systems by using the OS-level API. It kept crashing intermittently where I forgot to close something off inside the code but damned if I could find where it was. My replacement had a brilliant idea, split each input file into a whole bunch of singe line files, and run the program once per line. Simplicity itself, but damned if I could see that when I was looking at it.

[Anonymous Coward]

Re: Someone else fixed my fix

> My replacement had a brilliant idea, split each input file into a whole bunch of singe line files, and run the program once per line.

That's not a brilliant idea, that's a hideous kludge.

:-)

[Anonymous Coward]

Re: Someone else fixed my fix

One person's hideous kludge is another person's brexit agreement.

[Martin]

Is it just me, or does that picture look like Mesut Ozil crouching behind the desk?

[acomav]

Just you .

[swm]

I once designed a board to drive a laser printer. The commands were sent via a UART and status returned via the same UART. The UART was driven by microcode on a Xerox ALTO computer. The test group claimed I was sending an extra garbage byte at the end of printing. So I added microcode to dump into memory a trace of all characters sent or received. No garbage character?! I watched their test setup and noticed that the garbage character was being sent when my program terminated. Turns out that resetting the UART would send a garbage character. Sending a character without a start bit before resetting the UART solved the problem.

I was in research and the testers were in development but I'd learned to listen carefully to developers and people on the production line - they might know something you don't.

[Anonymous Coward]

Another consultant take of woe.

A number of years ago a company I worked for hired a new ops manager. He was one of those who didn’t trust his staff’s technical skills and preferred spending company cash on external consultants who told him the same the internal staff would have told him.

The company ran a number of critical databases on IBM Power boxes and a consultant was brought in to “health check” them. The guy who built the systems had built a redundant solution with multiple VIO servers managing the SAN and LAN traffic. He had gone on holiday when the consultant rocked up.

The manager had told the support team to give the consultant access to the VIO servers and all the LPARs on the P series boxes. After a poke round, the consultant started running a check script on the VIO servers. The first server kicked him out of his SSH session once the script started so he went on to the next VIO server. The SSH session was kicked out here as well. In fact, every time he ran his script, he was kicked out of his SSH session. More worryingly he couldn’t get back in. At that point he said the scripts would take a while to run and this was expected behaviour and he then left for the night.

Soon after that, operations started getting alerts, the application servers were struggling to get to databases. None of the database monitors were returning any information and DBAs couldn’t log into the systems. To all intents, all the databases were down and nobody could get into the system.

Frantic calls were made to the guy on holiday who built the system and despite his best efforts on the phone, nobody could get the boxes back online. He actually drove 4 hours back to the office and managed to sort the problem in about 30 minutes. The script the consultant ran caused a kernel panic on the VIO server, this was a documented issue the consultant should have known about. The issue was compounded by him causing a panic on every VIO server and not checking why his SSH session was terminated.

At the post incident review, there was a concerted effort from both the manager and consultant to shift the blame for the issue. The manager lasted a few weeks longer and the consultant was left in no doubt as to how his invoice would be dealt with.

[Anonymous Coward]

Expensive Oracle consultant

A while back I was testing a software program to count and sort personnel by job and work title. There were something like 90K personnel involved. The database was held in Oracle, but the report had to be in Excel. For a week the project was stymied because somehow the transfer to Excel wasn't receiving all the data. When someone mentioned to me that 60K-odd persons were missing from the Excel spreadsheet, I twigged and said, "You do know that Excel has limits on the number of rows in a sheet." The Oracle consultant (making £600 a day) wasn't aware of that and poor me, making a good deal less than that, had to point it out.

[Doctor Syntax]

Re: Expensive Oracle consultant

"The Oracle consultant (making £600 a day) wasn't aware of that and poor me, making a good deal less than that, had to point it out."

The fundamental issue here is the attitude that "Excel (or any other spreadsheet) is the solution, what's the problem?".

[jake]

Re: Expensive Oracle consultant

"the attitude that "Excel (or any other spreadsheet) is the solution, what's the problem?"

I know a guy who builds sheds, pump houses, small cabins and the like for a living. He is quite good at it, and is the go-to guy for those of us in the know around these here parts. He does all his drawing in Excel ... I've tried to sell him on a CAD program, but he wants nothing to do with it. Claims CAD just makes life difficult.

[Alan Brown]

Re: Expensive Oracle consultant

"He does all his drawing in Excel ... "

I know of entire Regional Health Authorities with 9 figure budgets and tens of thousands of staff whose financial systems are run on Excel

Then again my company decided to use the same Excel systems and consultants (my objections were overridden by other directors) and by the time I managed to get a proper accounting system in place (guerilla computing) we discovered about 6 figures in discrepencies and it wasn't down to anyone fiddling the books.

[Anonymous Coward]

Re: Expensive Oracle consultant

All hardware and software in IT has its constraints. Unfortunately many of the people using it don't know about the constraints.

On a proposed bid a supplier wanted to have their networked terminals continuously running as they said it would take too long to reload the software over the WAN every day. The terminals were basically PCs.

When asked what memory error detection was used - they said none was needed as hardware faults were rare.

I wasn't so sure about cosmic ray FITs for DRAM. We delved into the company depths to find our tame expert on such matters. He estimated that the terminals would have a memory error due to cosmic particles, probably undetected, every two weeks.

It reminded me of the prototype EELM KDF9 mainframe circa 1960. It had no memory parity checking "because ferrite core memory is now so reliable". It soon proved necessary to add parity checking. They also had to add another "no longer necessary" feature - an engineers' panel.

On a customer's 3rd generation mainframe there was a problem with random memory corruptions. The local support person insisted it couldn't be a hardware problem - "because the manual says that the memory is parity checked".

I had cut my teeth on trouble-shooting the factory prototype - I knew that the internal bus didn't have parity. Sure enough it was a noise problem due to a clean earth fault. Supposedly unchanging bytes in a word were corrupted in the CPU - then the word was written back to memory with recalculated parity.

[[e-mail]]

Images on PC screen

12 years aho, when my University mates and I where having our first jobs there was a lot of funny things that happened to us. One of them started working on a call center providing remote IT support for Broadband users on Telmex (the mexican 3play giant) here in Bogotá. One night he received a call from a troubled(and complicated) lady complaining about everything and not helping to find why "she cannot access the expensive Internet service provided by you idiots!!". When my friend asked her about what was on the screen, she answered " a couple of birds". She resulted to have one of those new ViewSonic LCD screens having a two Toucans' image on an upper corner. Once she understood she needed to focus on what was inside of the screen, all went smoothly and solution was found by adjusting something on the browser.

[avakum.zahov]

AI may be OK, the Human Inteligence, no so much ...

This is a real one from 2018. I very good friend of mine works as a software development manager. Couple of months ago she was in a meeting about implementing OCR with AI in order to read reports being printed by their systems on daily basis. The success rate of the pilot was below 40%, which according to the upper management was due to the printers producing copies not good enough for the OCR and the AI being not advanced enough. As the reports in question were being generated by systems connected to the network, my friend suggested that her group could simply interface those systems with their main DB thus removing any need for printouts, manual feeds and OCR with AI. Her managers became very frustrated, told her that she does not understand, they have always done it this way, and that she must support the Corporate decision.

[Prst. V.Jeltz]

Re: AI may be OK, the Human Inteligence, no so much ...

that is very upsetting ... must have lie down now ....

[bbqq]

Server Fail

About 20 years ago I was doing Unix admin in the same room as the help desk for a government department in Canberra. One Monday the guys at the help desk got a call from their Brisbane office saying that no-one could contract the server in that office. The helpdesk guy asked them to go and see what lights were lit on the server. The caller came back after a couple of minutes and said "Never mind. We've had a break in over the weekend and the server has been stolen". Problem solved.

[Unicornpiss]

Length..

I once had one of our young interns come in and ask me for an Ethernet cable. I was distracted and asked "How long do you need it?" He sheepishly looked at me and said "Forever?"