/dev/chat

RogUE

10:17 AM

Anyone here knows how to bind both my Super keys to the Whisker menu pop-up?

Faheem Mitha

12:00 PM

One more time - thoughts about:

@FaheemMitha similar to @Barry I suggest using pgrep -a apt instead. — Pablo Bianchi yesterday

?

Fabby

@FaheemMitha reading...

The only time I've had /var/lib/dpkg/lock locked when I was using mum's rate-limited Internet connection and update-manager was downloading packages in the background

I already upvoted your answer. Want me to give it an edit and a bounty as well?

Faheem Mitha

12:23 PM

@Fabby Heh. Whatever you want. :-)

My question was just - should I alter the answer as suggested? Or not? Feedback appreciated, but till now, not received.

And I thought you were on holiday.

And one more time, how bad is:

The following warning/error was logged by the smartd daemon:

Device: /dev/sdg [SAT], unable to open device

? Should I be concerned? Or should I ask on SuperUser?

Fabby

@FaheemMitha Looks like an SD card that is not inserted.

Faheem Mitha

@Fabby It's an external USB drive.

Fabby

@FaheemMitha Close...

try recreating a partition table on it.

if that works: recreate a FS

Faheem Mitha

I'm just wondering it it's a sign of impending failure.

Fabby

if that works, format

Faheem Mitha

12:37 PM

@Fabby It works most of the time. But it periodically throws this error.

Fabby

if any do not work, throw it in the bin

Faheem Mitha

@Fabby Pardon?

Fabby

@FaheemMitha yup, what I said.

Faheem Mitha

@Fabby I didn't follow it.

Throw my USB external drive in the bin?

Fabby

it's a USB stick...

Faheem Mitha

12:39 PM

No, it's a Passport drive. 1 TB.

Fabby

Holy shit!

Faheem Mitha

I'm not sure there is anything actually wrong with it.

Fabby

smartctl --all /dev/sdg

Faheem Mitha

But these things to die of old age.

Fabby

post output to that command here.

(preferably with fixed font)

Faheem Mitha

12:40 PM

@Fabby ok

root@orwell:/home/faheem# smartctl --all /dev/sdg
smartctl 6.6 2016-05-31 r4324 [x86_64-linux-4.9.0-6-amd64] (local build)
Copyright (C) 2002-16, Bruce Allen, Christian Franke, www.smartmontools.org

=== START OF INFORMATION SECTION ===
Model Family:     Western Digital Elements / My Passport (USB, AF)
Device Model:     WDC WD10JMVW-11AJGS2
Serial Number:    WD-WX51AC3W9747
LU WWN Device Id: 5 0014ee 60445ba31
Firmware Version: 01.01A01
User Capacity:    1,000,171,332,096 bytes [1.00 TB]
Sector Sizes:     512 bytes logical, 4096 bytes physical

Fabby

@StephenKitt What is a Paris Second?

It's the smallest amount of time you can measure in physics:

It's the amount of time between the light turning green and the person behind you honking their horn...

@FaheemMitha Reading

@FaheemMitha: 140 of 200 spare sectors are in use:

5 Reallocated_Sector_Ct 0x0033 200 200 140 Pre-fail Always - 0

when that reaches 200, the disk will be dead.

Faheem Mitha

@Fabby You seem well informed about such things.

Fabby

Meh. It's still OK

Faheem Mitha

Is this like a SSD? Stuff gets used up?

Fabby

it's worse: it's magnetic media with moving parts.

the disk still looks OK though, so try another cable.

Faheem Mitha

12:50 PM

@Fabby You think it might be the cable? Ok.

Fabby

@FaheemMitha I used to be a DC Manager.

@FaheemMitha disk looks OK.

only been power cycled 96 times and started and stopped 363 times.

is this a backup disk?

@Fabby Yes

Hence my concern.

Like I said: I used to be a DC Manager and we had SMART collection on all HDDs in the DC and used to be able to predict when disks were about to fail.

Faheem Mitha

12:52 PM

My previous Passport actually died. So it is definitely quite possible.

I've been lucky with hard disks. They have rarely died on me.

Fabby

Everything dies; nothing lives forever.

Entropy of the universe! /shrug

I've had 2 die on me in 30 years, but one of those 2 I dropped from the table onto a stone floor.

/shrug

Fetching something to drink. BRB

Faheem Mitha

Internal hard disks, I mean.

My experience with external ones is rather limited.

Fabby

Meh. same technology.

OK, beck to regexp...

Stephen Kitt

@Fabby excellent, I’d forgotten about the Paris second

Faheem Mitha

1:10 PM

@Fabby What does DC stand for here?

Fabby

@FaheemMitha Data Center

Faheem Mitha

@Fabby ok

I suppose SSDs are more reliable. I switched to that awhile back.

Fabby

@FaheemMitha Read this answer:

73

A: Is Ubuntu destroying my internal HDD?

HDDs die a slow, painful death like cancer, whereas SSDs just suddenly stop working like a heart attack : you can diagnose both by using SMART technology. Why you heard the noises under Ubuntu and not under Windows is because Windows was installed first, so it resides on the inner side of your h...

SSDs have their own problems.

Faheem Mitha

@Fabby I'm sure they do. But by and large they seem more reliable.

Fabby

@FaheemMitha They're shock-resistant, consume less power, ...

They're definitely not more reliable.

Faheem Mitha

1:24 PM

Actually, while we are on the subject, even with my SSDs, I observe the following peculiar behavior.

When I type stuff at a prompt, it often just hangs there before returning.

I don't understand why it would do that. But it's very annoying.

Fabby

post same info as above from your / drive.

Stephen Kitt

@FaheemMitha do you have a fancy prompt, or just the basic kind?

Faheem Mitha

@Fabby Ok. One sec.

@StephenKitt AFAIK, it's pretty basic.

I'm about to go for dinner. Will you guys still be here in 1/2 hr or so?

Fabby

@FaheemMitha Please don't ping me to tell me to wait...

:P ;-) :p

Faheem Mitha

root@orwell:/home/faheem# smartctl --all /dev/sde
smartctl 6.6 2016-05-31 r4324 [x86_64-linux-4.9.0-6-amd64] (local build)
Copyright (C) 2002-16, Bruce Allen, Christian Franke, www.smartmontools.org

=== START OF INFORMATION SECTION ===
Model Family:     Samsung based SSDs
Device Model:     Samsung SSD 850 EVO 500GB
Serial Number:    S2R9NX0J504679W
LU WWN Device Id: 5 002538 d41f8a8e8
Firmware Version: EMT02B6Q
User Capacity:    500,107,862,016 bytes [500 GB]
Sector Size:      512 bytes logical/physical

@Fabby What? I don't follow.

Fabby

1:27 PM

Don't reply as that sounds a bell

@FaheemMitha bell

Faheem Mitha

@StephenKitt Ok, this seems to be the prompt, but I also think it's the Debian default

Fabby

especially to tell me "I'll start doing that now

Faheem Mitha

#set a fancy prompt (non-color, unless we know we "want" color)
case "$TERM" in
xterm-color)
    PS1='${debian_chroot:+($debian_chroot)}\[\033[01;32m\]\u@\h\[\033[00m\]:\[\033[01;34m\]\w\[\033[00m\]\$ '
    ;;
*)
    PS1='${debian_chroot:+($debian_chroot)}\u@\h:\w\$ '
    ;;
esac

I'm not sure I follow, Fabby. You don't want me to reply to your messages?

Stephen Kitt

@FaheemMitha by fancy prompt I meant a prompt which runs programs to calculate what it’s going to display

like those which show git status etc.

Faheem Mitha

@StephenKitt Oh. No, I had those once. I took them out. They were a nightmare.

Fabby

1:29 PM

@FaheemMitha Disk looks good must be something else...

Faheem Mitha

Actually, I see I still have that commented out in my ~/.bashrc.

@Fabby I wonder what. Debugging suggestions welcome.

Stephen Kitt

@FaheemMitha hmm you’ve never run any self-tests, so some of the SMART data won’t be up-to-date

Faheem Mitha

Any way of asking bash what the hell it is doing when it is pausing?

@StephenKitt Should I run self-tests?

Stephen Kitt

@FaheemMitha set -x

Fabby

Stephen already asked you about second Idea I had: your prompt

Stephen Kitt

1:30 PM

@FaheemMitha yes, smartd can take care of it for you

Fabby

@FaheemMitha Can you just type 10 times [enter]

Faheem Mitha

@StephenKitt In .bashrc?

Stephen Kitt

@FaheemMitha on the command line

Fabby

Is that slow?

Stephen Kitt

just to see what’s happening

Faheem Mitha

1:31 PM

@Fabby Type enter 10 times?

Fabby

@FaheemMitha yup. Is that slow?

Faheem Mitha

++ history -a
++ history -c
++ history -r
++ echo -ne '\033]0;faheem@orwell: /home/faheem/personal/medical\007'

@Fabby No, not really. It's more of an issue if I type actual commands.

The above is what I get if I just type enter with -x on.

Fabby

@FaheemMitha echo $PS1

Faheem Mitha

I'm off to dinner, back in a bit.

${debian_chroot:+($debian_chroot)}\u@\h:\w\$
++ history -a
++ history -c
++ history -r
++ echo -ne '\033]0;faheem@orwell: /home/faheem/personal/medical\007'

Fabby

I hope I'm having a nap by the time you're back.

Faheem Mitha

1:33 PM

Does that look harmless? That's echo $PS1.

Stephen Kitt

@FaheemMitha it’s updating the history on disk, clearing in memory and reloading it for every prompt, which could introduce a pause if there’s I/O contention

Fabby

----^I agree.

remove those 3 lines.

terdon

2:12 PM

I can understand why you'd want the -a and -r, but why -c in the middle?

You've probably set it in your PROMT_COMMAND, by the way.

Stephen Kitt

@terdon you need -c before r because -r appends

Faheem Mitha

2:29 PM

@StephenKitt So, do you also recommend removing those 3 lines?

Is this different from regular history updating?

I think bash saves history regardless.

terdon

@StephenKitt -a appends

Ah, so does -r

@FaheemMitha It does, this sort of thing is only useful when you want the history to always be in sync between different bash sessions (e.g. different terminals)

Faheem Mitha

@terdon Do you also think it might be causing hanging problems?

BTW, I think there is a lot of history too.

I mean, a lot of history saved.

terdon

2:51 PM

@FaheemMitha Could, if you've set HISTSIZE and HISTFILESIZE to something big.

What do you get for history | wc -l?

Faheem Mitha

6054

echo $HISTSIZE
100000000

Hmm, I wonder if I really meant to write that.

That's 10^8 - a hundred million.

echo $HISTFILESIZE
100000000

I see a theme here...

Faheem Mitha

3:36 PM

Wow, the amount of output when you hit tab in bash with -x turned on, is terrifying.

No wonder it's so slow.

terdon

3:47 PM

@FaheemMitha That is probably it. I would just remove the history commands from your PROMPT_COMMAND. The rest should be fine now that you don't re-read it all on every showing of the prompt

Faheem Mitha

@terdon You mean the -a, -c, and -r lines?

terdon

yep

I used to have those but removed them for the same reason.

And they're not really worth the overhead. Just close the terminal and open a new one and the history is there.

Faheem Mitha

@terdon Because they would hang the prompt?

terdon

Slow it down, yes

Not hang though. If you're actually hung, as in nothing happens, then you have another issue.

Faheem Mitha

This is about instantaneously updating history across all open terminals?

terdon

3:50 PM

Yes

Faheem Mitha

@terdon Well, the prompt returns eventually.

But it can take a few seconds. Which can seem like an eternity when you're trying to do something.

This doesn't happen all the time, but when it happens, it's quite annoying.

terdon

The -a writes the current history to $HISTFILE, the -c clears the current history and the -r reads from the file again. If you have those set as the PROMPT_COMMAND, that means ~6k lines being written, and then read every time you see a prompt.

Faheem Mitha

@terdon Yikes.

terdon

If it isn't super important for you to have the exact same history shared by multiple open terminals, just remove them

I'd remove them anyway, if only to check that they're causing the problem

Faheem Mitha

@terdon Ok, will do. Thank you.

terdon

3:52 PM

np

Faheem Mitha

@terdon No, it's not super important.

Actually, sometimes it can be confusing. Like when you do a command recall, and you get a command you didn't type at that terminal.

Not exactly intuitive behavior.

Of course, sometimes it can be useful too...

Jesse_b

4:47 PM

I think this question is not actually (exactly anyway) a duplicate but I mistakenly marked it :( unix.stackexchange.com/q/440598/237982

terdon

@Jesse_b It's shown as closed by you and "community". That means you cast the 1st vote, but the OP agreed and voted to close.

So it's all good

Jesse_b

Yeah, I don't like answering questions that I voted to close though :-P

@terdon: Have you seen this user yet: unix.stackexchange.com/users/288391/…

They seem to be going around posting some spam links in answers this morning

Stephen Kitt

@Jesse_b oh dear what have I done

I mean the cow thing

BTW @Jesse_b you should just flag those answers as spam, without editing them

Jesse_b

@StephenKitt :p sorry, I initially tried to just write "Spam deleted" etc and it kept telling me too short

Yeah I didn't know if the link is malicious or not

seems sketchy

Evan Carroll

5:06 PM

@Fabby ? Depends on how you measure them, I would assume they're more dependable generally.

terdon

5:17 PM

@Jesse_b Yeah, please don't edit spam. That makes it harder to detect, harder for mods to deal with spam and just makes the whole thing more complicated to deal with.

Just flag as spam.

Jesse_b

Will do

terdon

thanks

Evan Carroll

5:55 PM

No longer relevant: unix.stackexchange.com/a/440550/3285

Faheem Mitha

6:40 PM

So here is what I have for that history reload thing:

# Save and reload the history after each command finishes
export PROMPT_COMMAND="history -a; history -c; history -r; $PROMPT_COMMAND"

At some point I must have thought this was a good idea...

Hmm, the prompt response after commenting that out, even on return, is noticeably faster. Maybe that was the problem.

terdon

yay!

Faheem Mitha

@terdon Enthusiastic!

Come to think of it, I think I read it here first.

Yes, let's blame U&L.

I was chatting to some people on IRC who swear by ZFS. I suppose the Debian installer still does not support it, right?

Judging by unix.stackexchange.com/q/383566/4671 using ZFS on Debian isn't that easy.

Faheem Mitha

7:22 PM

Hmm, I'm tempted to experiment, but I don't really have the time.

Could I do this ZFS installation thingy in a VM?

Eliah Kagan

@FaheemMitha Why not?

Faheem Mitha

@EliahKagan Dunno. So I could?

Are the current answers satisfactory? There seems to be a certain amount of confusion.

Eliah Kagan

I don't think I know enough about zfs to evaluate the answers properly. It's something I've been meaning to get around to trying, too.

But yeah, you should be able to use a VM for that. A system installed in a VM has its own kernel and drivers and everything.

You can use a VM to virtualize a completely different OS, after all. So it should work for this.

Do you mean use a VM to format and write a zfs filesystem to a physical disk? You should also be able to do that, though I think it might be easier with some hypervisors than others.

Faheem Mitha

Does anyone feel like reviewing these minor edits? There is just one place I'm doubtful about.

unix.stackexchange.com/posts/396160/revisions

@EliahKagan Not necessarily. Depends on the VM.

@EliahKagan No, I just meant test ZFS inside a VM, running Debian.

The answer says:

> it's a sort of saving resources mechanism ( but even in case this will still throw errors inside journalctl -xe )

Eliah Kagan

@FaheemMitha Oh, yeah. Totally. That should be no problem. It may not have the same performance characteristics, of course. If you're writing to a disk image that is itself stored on another filesystem on the host (as you usually would be, with a VM).

Faheem Mitha

7:34 PM

"even case" is clearly not correct. I changed it to "even in case", but that's just a guess, because I'm not sure what that sentence means.

@EliahKagan ok

See my question just above...

Eliah Kagan

Do you mean about reviewing the edits? Or something else?

I've looked at revision #7. It looks good.

Faheem Mitha

@EliahKagan Yes, my edit. See my question about that sentence.

Eliah Kagan

@FaheemMitha You have it as "even in that case" now, right? That looks reasonable.

Faheem Mitha

What did he mean by "it's a sort of saving resources mechanism ( but even in case this will still throw errors inside journalctl -xe )"?

@EliahKagan Yes, "even in that case".

Eliah Kagan

I think they're trying to say something about what zpools are for.

Faheem Mitha

7:38 PM

But what case is he talking about exactly. It's not clear.

Feel free to edit for clarity if you know what he's on about.

Well, you hardly need my permission. Or anyone else's provided you're over 3k.

(If I recall correctly.)

Eliah Kagan

I think they're talking about the case where you do create the zpools. I think the post is saying that you get scary messages in your log even if you do it right. I am not 100% sure.

Faheem Mitha

@EliahKagan Yes, that sounds likely. But it could be more explicit. It's partly confusing because it uses the word "still" about scary messages, but no scary was mentioned before that.

It would be nice if Debian would support zfs properly so one didn't have to screw around.

But I suppose the license issue isn't Debian's fault.

Eliah Kagan

Yeah.

Faheem Mitha

@EliahKagan To which bit? :-)

Eliah Kagan

That the license issue isn't Debian's fault. (But I agree that it would be nice if Debian had better zfs support if it could, too.)

Faheem Mitha

7:46 PM

@EliahKagan ok

Evan Carroll

8:46 PM

XRandR doesn't keep any state, or set any stateful features in videocards or the bios/uefi. There is no way booting in and doing anything will preserve a change to another boot up. Which gets back to the point: it worked for you, which is awesome for you but it's not all that useful for others, and it may even confuse them or discourage them. — Evan Carroll 1 min ago

That whole question should be closed

Transcript for