@Kusalananda Thank you. And thank you for your help and advice.

@StephenKitt looks like one for you:

@terdon indeed...

@StephenKitt Do you understand why the OP is using a chroot environment there?

@terdon not from the question, but it could be for a number of reasons

such as setting up a system using another, or just for testing purposes

OK. Just curious if it were obvious from the question.

I don't see why a chroot would make a difference.

I don't think it does. I was just curious.

@Kusalananda Well, I wrote about that SSD Amazon India listing to Amazon Customer Service. I got a predictably clueless response back. (At least it was fast.)

So I called up Customer Service, and got someone who at least seemed to understand what I was saying.

I'm probably making a mountain of a molehill, but it makes me nervous when a product I'm thinking of buying has nonsense or wrong information on the product page.

@FaheemMitha Now deleted it seems.

@Kusalananda Huh?

"[Deleted by Amazon 8 hours ago]"

@FaheemMitha Ah, you wrote elsewhere to them as well?

Well you got some kind of reply that was dolotod by Amazon on that form post as well.

"dolotod"? My Dvorak keyboard is dyslexic.

@Kusalananda Sorry, I don't follow. Are you referring to the question I posted to that product page? I got a fairly useless reply, which said "exact spec available on samsung site". Maybe someone deleted that.

I hadn't checked the page directly.

@Kusalananda Dvorak? How advanced.

That's the one I was referring to. amazon.in/description-hardware-interface-dimensions-version-/…

Yes, I see it's been deleted.

@Kusalananda Yes, that's the one.

The reply/answer was deleted. It said: "exact spec available on samsung site".

@FaheemMitha Yes, I'm very "l33t". I'm also permanently crippled an a QWERTY-keyboard thanks to it.

But actually it wasn't so useless, because I went and looked at the Samsung page for that product, which had the spec. Which was useful.

Good

But the answer itself didn't answer the question. But that's quite common in India. The reply from the Amazon rep didn't address my question at all, either.

In India, it's quite common to call up customer service, ask something, and have them start talking about something completely different.

Often they're trying to sell you things.

Monty Python would be wasting their time in India - they'd never get an audience. The whole freaking country is one big Flying Circus.

Heh. I've heard that if you ask for diretions in India, you will always get them. They might take you in the oppsoite direction of where you're supposed to be going, but at least you got some "help" ;-)

@Kusalananda Actually, that's one of the less dysfunctional things in India. Having said that, it's prudent to ask the direction from multiple people. Then again, that's true everywhere, I think.

@FaheemMitha Yes, you may be right there.

And Indians don't know roads. They navigate by landmarks. And in Bombay, at least, there are no signposts, as far as I can tell. At least, I hardly ever see them. Maybe they were taken down and sold as scrap. Who knows.

Anyway, as I said, I called Amazon Customer Service, and talked to someone who at least seemed to get the point. At least he repeated it back to me multiple times.

When I lived in the UK, people seemed to often use pubs as landmarks when giving me directions. I'm not a fan of pubs though, so I would be as bewildered after as I was before.

One thing you can say about Indians, at least they are friendly, especially if they think you are a foreigner. Even though they will probably still try to rob you.

@Kusalananda Really? That's an odd choice of landmark.

Then again, most people would probably be familiar with them.

@FaheemMitha Not really, there's one around every corner.

@Kusalananda Yes, I suppose that's true.

And about half of them are called "The Red Lion".

@Kusalananda Hopefully that isn't true.

I tried to eat in British pubs occasionally. Terrible food. And I don't really drink, so...

No, that was'nt quite true, but there's one in Ickleton and another one in Hinxton, and when going for lunch at the "Red Lion" from work at the Wellcome Trust, you never quite knew which one they meant.

@Kusalananda I see.

Usually good veggie-burgers though. At least the ones I remember.

I remember microwaved pies. Which are probably fine if you are really, really hungry.

So, buying the SSD or not?

@FaheemMitha :-) Had one yesterday.

@Kusalananda I will. I'm kind of waiting for that issue with the product page to be sorted out.

Maybe I'm being super cautious. But I get increasingly paranoid as I get older, I guess.

Is this a big investment for you?

@Kusalananda What kind of pie? I'm partial to chicken pot pies.

@Kusalananda Not specially. It's not nothing either. It's more expensive here than in the US, at least.

I made a half hearted attempt to get my cook to make chicken pot pies. But I couldn't find a chicken pot pie recipe in Hindi. Which is my lazy go to method of instruction.

@FaheemMitha Being vegetarian, it was a cheese pie with mushrooms. Quite ok for a microwave pie. This one in fact :-) felix.se/produkter/vasterbottenspaj-210g

I think I do need to get new hard drives. The ones I have are around 5 years old, and no aging particularly well.

@Kusalananda Looks reasonable. But tv dinners aren't the healthiest things around. Lots of additives.

Do you cook much?

Yes. I knooow...

No. I don't have the energy for it, or the time.

@Kusalananda Sure, I can relate.

Do you know how to? I don't really.

I tried cooking once. It's been a while.

I'm actually quite good at cooking, and I enjoy it. It's just not happening at the moment :-/

@Kusalananda Ah, that's nice.

Being able to cook, I mean. I can't really, like I said.

Regarding:

Does Debian now handle apt authentication automatically? Is there anything the user needs to do? My impression is that it's automatic.

@Gilles the interactive flag of bash causes it to source ~/.bashrc , /etc/profile and also run aliases. Therefore, if you wrote #!/bin/bash -i instead of #!/bin/bash as the shebang then you could use your aliases throughout the bash script.

@CedaEI context?

@CedaEI that's pretty dangerous

if your script uses your aliases then it won't work for other people

and IIRC bash -i requires a terminal

well that is for personal testing as i mentioned in the comment

@CedaEI Oh, right, build scripts. You absolutely don't want to rely on the user's aliases in build scripts! Build scripts have to work for anybody who compiles the program.

@CedaEI but the context of the question has nothing to do with personal testing

you obviously would have to change that build script either ways because you do not expect other people to have multiple gcc's installed at different places

@CedaEI that is what the $CC environment variable is for

@CedaEI Aliases are really only used for making the command line a bit snappier to work with. If you have more complex operations in that script of yours, you could consider using shell functions (defined in the script).

@CedaEI No, the point is to use existing build scripts

@casey CC is only the C compiler, selecting a toolchain typically requires at least LD as well, often CFLAGS and LFLAGS, and often other tools like as, strip, etc.

To select a gcc-like toolchain, often the easiest way is PATH=/my/alternate/gcc/bin:$PATH make

depends how the build scripts are written of course

@Gilles true

Or set sane defaults and let the user override them.

In linux, which is the directory in Linux file system hierarchy

to create dummy directory structure for testing

Is it /tmp

?

Depends on what you're testing... /tmp is typically shared system-wide. So e.g., if you create /tmp/myprog predictably, other users on the system could interfere.

/tmp is also often automatically cleaned up, especially at boot

Of course, you can use mktemp -d to create a directory for you

If you fill in the various details, seems like an appropriate question to ask on the site

Folks, my hard drives have become remarkably slow recently. I was just wondering - is there some simple and robust test to measure I/O, and is there some simple measure of normalcy I could apply? These are elderly HDDs, not SSDs.

I should clarify, slow, as in I type okular filename.pdf, and the computer sits there a while thinking about it before okular starts up. Which didn't use to happen.

Of course it could be something else, not I/O.

@FaheemMitha start by checking whether you're swapping

how's free?

look at iotop, but it's not always easy to read usefully

root@orwell:/home/faheem# free
             total       used       free     shared    buffers     cached
Mem:      16367424   15703760     663664     340232     949460    2744664
-/+ buffers/cache:   12009636    4357788
Swap:      8388604    4607340    3781264

Swap does seem kind of high, but I don't hear much swapping. I can usually tell when swap is happening.

@FaheemMitha iostat -kyx 2, or whatever number of seconds instead of 2, while okular is starting would show if its I/O—and also if its unexpectedly slow I/O

%util is even useful on boring magnetic drives (some drives can process multiple request simultaneously—think hardware RAID—it's not so useful on those)

r/s and w/r will give you the I/Os per second, rkB/wkB give the throughput

magnetic drives do 100 (random) IOPS max, even for good fast ones. Slightly less for slower ones. Maybe 60 or 80.

Thanks for the tips @derobert. I'll try iostat.

I'd also try vmstat (if it's all you have) or dstat...

Because I'd guess you're swapping.

@derobert If it's all I have?

@FaheemMitha Every machine has vmstat installed. dstat is much rarer. Of course, you can apt-get install it.

(But when you're logging in to a server and wondering why the performance is terrible—apt-get install isn't high on the list of things to try!)

dstat 10 for example is nice—it'll show you second-by-second values, and every 10s leave a line around

@derobert Well, it's my personal work machine - so apt-get install is not a problem.

These is what dstat 10 output while I was waiting for okular to start up.

faheem@orwell:~/personal$ dstat 10
You did not select any stats, using -cdngy by default.
----total-cpu-usage---- -dsk/total- -net/total- ---paging-- ---system--
usr sys idl wai hiq siq| read  writ| recv  send|  in   out | int   csw
  6   2  81  11   0   0|1630k  558k|   0     0 |9407B   12k|2851  6843
  7   3  60  30   0   0| 977k 2325k|8353B 5440B|3277B    0 |3112  6723
  6   2  62  30   0   0| 942k  108k| 945B 1064B|   0     0 |2315  5381
  6   2  61  31   0   0| 932k  364k| 690B 1015B|   0     0 |2585  6071

gahhh! hit the code format button

so, no paging, but there is some disk I/O. Try iostat.

Of course, I imagine the second launch is fast if it's I/O...

@derobert Nope, still slow.

@derobert I did. It produces reams of stuff.

Not easy to summarize.

@FaheemMitha It should produce one line per block device, and repeat that every X seconds. You can limit it to only a few block devices by giving them as extra arguments (e.g., iostat -kx 10 sda

The fact that launch #2 is also slow is weird, though. If it's IO, it shouldn't be: everything should be cached. Here, first Okular launch takes a second or two. Second is almost instant.

@derobert Mmm. Should I stop it once okular has started up?

@derobert Maybe not I/O then.

@FaheemMitha It's just displaying statistics... you can stop it whenever... But its the best basic command I know of to see I/O stats quickly

I'd suggest adding a new user, and logging in as that new user... then see if Okular continues to be slow. To see if its something with the machine or with your particular KDE profile.

Device:         rrqm/s   wrqm/s     r/s     w/s    rkB/s    wkB/s avgrq-sz avgqu-sz   await r_await w_await  svctm  %util
sda               0.00     8.50    0.00   45.00     0.00   258.00    11.47     0.25    5.64    0.00    5.64   2.67  12.00
sdb               0.00     8.50    1.50   45.00     6.00   258.00    11.35     0.25    5.33    9.33    5.20   2.49  11.60

This is from the first set.

@derobert Did you mean r/s and w/s?

You wrote r/s and w/r.

@FaheemMitha yeah

Is that with or without -y? Without it, the first line is since boot—that seems like a fairly high w/s on a dekstop for since boot

@derobert I just did iostat -kyx 2 like you said.

ok, so that's with -y

so that doesn't look I/O bound, then

if it were I/O bound, %util would be much closer to 100%

@derobert Well; it's just the first bit.

is it the bit from when Okular was starting?

@derobert Well, I started it after starting okular. So, in theory, yes. Though I think okular might take like 5-10 seconds to start.

How could I time it?

@derobert Where is %util? I don't see it.

xtoolwait would let you, but that got removed from Jessie :-(

@FaheemMitha last column

Oh, I missed that.

Your terminal isn't wide enough, it's wrapping

Yes, the %util thing stays fairly low most of the time.

$ time xtoolwait okular
okular(1586)/kdecore (KConfigSkeleton) KCoreConfigSkeleton::writeConfig:
okular(1586)/kdecore (KConfigSkeleton) KCoreConfigSkeleton::writeConfig:
okular(1586)/kdecore (KConfigSkeleton) KCoreConfigSkeleton::writeConfig:
okular(1586)/kdecore (KConfigSkeleton) KCoreConfigSkeleton::writeConfig:
okular(1586)/kdecore (KConfigSkeleton) KCoreConfigSkeleton::writeConfig:

real    0m0.281s
user    0m0.000s
sys     0m0.000s

... that's how long it takes here to show the window

@FaheemMitha throw some I/O load on the system, and it should go up to near 100%

@derobert Sounds like a handy command. Where did you get it from?

@FaheemMitha errr... let me check...

grumble packages.debian.net still down grumble

@derobert It's down??!!

.net not .org

Ah

@FaheemMitha AFAIK, you get it from archive.debian.org/debian/pool/main/x/xtoolwait ... I have 1.3-6.2, which I think came from Squeeze.

@derobert Hmm, will I have to rebuild it?

no, it just installs

@derobert ok

Well, I haven't installed it recently, but it probably still does

was removed back in 2011 bugs.debian.org/cgi-bin/bugreport.cgi?bug=619706

Got it. So time xtoolwait okular filename.pdf?

Yeah, that should tell you how long it took to map a window

aheem@orwell:~/personal/business/smyt$ time xtoolwait okular daryanagar_bmc_tax_issue.pdf
xtoolwait: warning: timeout launching okular
okular(23960)/kdecore (KConfigSkeleton) KCoreConfigSkeleton::writeConfig:
okular(23960)/kdecore (KConfigSkeleton) KCoreConfigSkeleton::writeConfig:
okular(23960)/kdecore (KConfigSkeleton) KCoreConfigSkeleton::writeConfig:
okular(23960)/kdecore (KConfigSkeleton) KCoreConfigSkeleton::writeConfig:
okular(23960)/kdecore (KConfigSkeleton) KCoreConfigSkeleton::writeConfig:

wow, it took longer than it was willing to wait?

Apparently I was off about the time.

@derobert Yes, I don't understand the timeout part.

Why does it have a timeout?

it was intended to be used in e.g. an xinitrc to start up a bunch of stuff on login—if one of those programs hangs for whatever reason, you still want to be able to log in

@derobert oh

Hmm, the time listing only happens after existing okular. so it isn't accurate. Basically `xtoolwait fails here.

Too bad :-(

That may just be a bug in it when it takes more than 15s (the timeout)

$ time xtoolwait sleep 20
xtoolwait: warning: timeout launching sleep
xtoolwait: warning: child (sleep) died prematurely

real    0m20.006s
user    0m0.000s
sys     0m0.000s

... yep, looks like a bug in it. Yeah for unmaintained software!

@derobert Hmm.

Why was it removed?

for being unmaintained — orphaned and dead upstream as well

@derobert Oh, that's a shame. And with no substitute?

Not that I know of. I'm sure you could script some poll loop with wmctrl

Another command I don't know. :-)

On a random off-topic, if you like it when people lay into the British royal family (and I do), you might enjoy Russel Brand's videos in the subject. On Youtube and possibly other places. There are a bunch of them. Apparently it's a bit of a theme with him.

@derobert That last comment wasn't directly specifically at you, to be clear.