@CodeMed - virsh destroy vmname to immediately shutdown a VM (i.e. unclean shutdown, the VM equivalent to yanking out the power cord). see virsh help destroy for info.

any mod here ?

see unix.stackexchange.com/questions/235992/… and unix.stackexchange.com/questions/235607/…

@Archemar You can flag the questions.

What is the issue, exactly?

issue is he won't take no for an answer. today's is IIRC the third minor variant of the same question.

Yes @cas say it all, second question in a row on same subject.

@Archemar Ok, So flag it and say as much.

@terdon ^^

Seems to have been dealt with. Both Qs are closed.

@terdon Ok, sorry for bothering you.

No bother at all. I was just pleased to see the community dealt with it already.

I was wondering (maybe hopelessly) for a long term solution ...

@Archemar I wish one existed... but there is no way to implant understanding in another person when they are not willing to accept it.

@JennyD, a flying drone army of clue injectors

@cas they will be shot down by the bogon emitters

a BIGGER army then

@JennyD Well, things are mostly relatively quiet here, so one cannot complain too much.

@cas There will always be more bogons than cluons...

@FaheemMitha True... there is a lot more work in keeping SF reasonably clear of crap than in doing the same for U/L.

Anyone know if there is a lower age limit for SE? Someone just said 13. I wasn't aware of one.

@JennyD Right, you spend more of your time there.

@FaheemMitha US companies will usually have an age limit of 13 due to there being laws against storing personal information about children younger than 13.

@JennyD Oh. I don't think I've seen it anywhere on SE, though.

@FaheemMitha SF suffers greatly from communitybuilding.stackexchange.com/questions/1113/…

@FaheemMitha It's at stackexchange.com/legal - *Subscriber certifies to Stack Exchange that Subscriber is an individual (i.e., not a corporate entity) at least 13 years of age. *

@JennyD I would have thought that was one site that would not have that problem, seeing as it is... exclusionary.

@JennyD Huh. Never seen that.

@FaheemMitha There's a huge Dunning-Kruger problem - people who know only a little don't realize how much there is left to know, so they think they are experts even though they're not.

Also there's an influx of people who haven't read the help center, or who've come from other SE sites where the rules are different.

@JennyD I can imagine. But why would SF be more of a problem than U&L? They're quite similar sites.

lots of questions about apache from php web devs.

PHP, ugh. Sorry, had to say that.

@FaheemMitha They're not, really... there's an overlap, but not as big as you'd think. SF also takes Windows admins, and a lot of people come there from SO because they've gone to SO for programming their web stuff and then they go to SF for deployment questions - without reading the help center and without the least bit of understanding of what professional systems administration is.

I don't mean to be whiny here, there's no reason for you guys to be subjected to my rants...

U&L naturally excludes lots of dumb question because the command line scares off people.

@JennyD, yep. that's exactly right.

@JennyD No problem. Rant away.

@cas Well, free software implies some degree of good taste. So maybe it also acts as some kind of filter.

Yes, yes, I'm a snob.

i used to be active on SF but hardly spend any time at all there now. spend most of my SE time on U&L

@cas Didn't like it there, then?

I've never asked or answered a question on SF. And I hardly ever read stuff there, either.

Lunch time. Se you!

@faheem, no...it's not a matter of not liking it. more that fewer and fewer of the questions were interesting to me.

@cas I see.

in U&L you can focus on specific answers to specific questions. in SF, you often have to take about ten steps back to explain the fundamentals of how whatever it is they want to do actually works and what processes are involved and even how basic networking works. "my apache doesn't work" is so much more open ended than "help me with my awk or perl script".

In SF that takes a long time.....and often you end up feeling like you've thrown pearls before swine. or perl before $pig{canonical}.

@FaheemMitha Yes, there is. No users under 13 allowed.

@terdon Yes, @JennyD kindly posted me the relevant link.

@FaheemMitha No it doesn't. I wish it did but it simply doesn't. It might have back in the '90s but not in this modern day of computers shipping with Linux.

@FaheemMitha Ah, true, I missed that.

@terdon Hmm, I'd like to think it is still true to some extent. But it's hard to prove, one way or the other. It's true, 15-20 years ago, our little club seems more exclusive.

A lot of users use Linux/OSS just because it works, or it looks cool or that's what shipped with their computers. They neither know or give a damn about any kind of politics behind it, they just want their machine to let them type a document, browse the internet and watch a movie.

And that's a good thing, as far as I'm concerned. We can't expect everyone to care about this sort of thing. There's nothing wrong with just being a user.

15-20 years ago, the internet was better, had a higher clue-quotient. but the world is what it is and even the time of eternal september has its benefits. (not having to explain to non-geek friends what the internet is, for starters).

Or Linux, for that matter.

True, the world changes. But at least the people who come here have made a conscious decision not to use MS, or Apple. Which is something, I think.

Though maybe they just wanted something that is free.

Both MS and Apple are quite expensive to run.

Wouldn't know. I've never paid for Windows (but don't tell anyone)

@terdon Well, I bought a laptop with Windows 95 pre-installed once.

I don't know if that counts.

Oh, yes, me too. I have decided that doesn't count since it came with the machine.

lots of people here use OSX

Do I need to reboot for a new locale to become available? I want to run a command with the locale I just generated and it doesn't seem to work. I ran dpkg-reconfigure locales, chose the new one and it was generated. It is now shown in locales -a but LC_ALL=pt_PT.utf8 COMMAND still prints in English.

An older locale does work. Do I need to reboot or something? ssh localhost also doesn't get the new one.

@terdon, my first ever windows machine was a win7 box i built from spare parts about 3 or 4 years ago just to be a steam games console (i had several games i'd bought in steam that didn't work properly or didn't work at all in wine). i went from MS-DOS -> OS/2 (very briefly) to SLS Linux (and thence to slackware and on to debian).

@terdon I think it should take effect immediately, certainly if you explicitly change the setting in the shell. But I'm no expert on the subject.

@cas Heh, that's what I call holding out!

@terdon: Maybe that's netstat problem

@FaheemMitha That's what I thought too but neither of my two freshly-generated ones do while the older ones (which were generated in the same way) work as expected.

does . <(locale -v) work?

@cuonglm Yes, it's for that question but it works fine with another locale.

@terdon, not holding out...just had no use or liking for microsoft software.

@terdon Um, maybe you need to set a default? How does your machine know which one to use?

@cas I'm not setting it as default, so that shouldn't make a difference, I guess.

@FaheemMitha I have a default. Hang on, let me give you guys the details

$ locale -v
LANG=en_US.utf8
LANGUAGE=
LC_CTYPE="en_US.utf8"
LC_NUMERIC="en_US.utf8"
LC_TIME="en_US.utf8"
LC_COLLATE="en_US.utf8"
LC_MONETARY="en_US.utf8"
LC_MESSAGES="en_US.utf8"
LC_PAPER="en_US.utf8"
LC_NAME="en_US.utf8"
LC_ADDRESS="en_US.utf8"
LC_TELEPHONE="en_US.utf8"
LC_MEASUREMENT="en_US.utf8"
LC_IDENTIFICATION="en_US.utf8"
LC_ALL=

At least in that one instance, you did set it explicitly.

$ locale -a
C
C.UTF-8
el_GR.utf8
en_US.utf8
fr_FR.utf8
POSIX
pt_PT.utf8

$ LANG=pt_PT.utf8 netstat -apn | head -n2
(Not all processes could be identified, non-owned process info
 will not be shown, you would have to be root to see it all.)
Active Internet connections (servers and established)
Proto Recv-Q Send-Q Local Address           Foreign Address         State       PID/Program name

But:

$ LC_ALL=fr_FR.utf8 netstat -apn | head -n2
(Tous les processus ne peuvent être identifiés, les infos sur les processus
non possédés ne seront pas affichées, vous devez être root pour les voir toutes.)
Connexions Internet actives (serveurs et établies)
Proto Recv-Q Send-Q Adresse locale          Adresse distante        Etat        PID/Program name

The fr one was generated a few months and several reboots ago, the PT one was generated 5 minutes ago.

@terdon Well, the strings do have to exist for that locale.

Try a different program. Say apt/aptitude.

try egrep -v '#|^$' /etc/locale.gen. if it looks OK then run locale-gen

@FaheemMitha Nah, I figured it out. It works if I log in through a tty. Not sure why ssh localhost didn't work though.

@terdon Huh. Probably the env variable is not set, then.

That's the obvious possibility, at any rate.

@cas dpkg-reconfigure locales does that.

@FaheemMitha There's no env variable here, I'm setting it explicitly when calling. That's what's surprising.

it should do. but maybe there's something broken in your /etc/locale.gen and dpkg-reconfigure is suppressing the error output. maybe.

@terdon Well, there are multiple variables that can be set. Did you check the environment?

I don't know locales. If say Gilles was here he could probably diagnose it for you.

Actually, @FaheemMitha you might be right. It works as expected with ls but not with netstat. Perhaps the strings are indeed missing.

OK, so it works for some commands and not for others. Odd:

$ LC_ALL=pt_PT.utf8 date
Qua Out 14 12:29:42 EEST 2015
$ LC_ALL=el_EL.utf8 date
Wed Oct 14 12:29:49 EEST 2015

PT works for date and ls but not netstat and EL doesn't seem to work for anything so far.

Yes, someone has got to write the stuff. And there are many programs.

@cas OK, tried but no difference. It seems to be command-dependent.

I think mercurial/git probably have reasonably good coverage. You might try them too.

@FaheemMitha Better than GNU date or ls?

what's el_EL? my system only has el_CY and el_GR

@terdon Hmm, maybe not. :-)

@cas Should be the same as el_GR I guess. Let me see what else I have, perhaps one of those will work better.

@cas Turns out my system also only has _GR and _CY, I was using the wrong name. Idiot. OK, still no love for netstat though.

Hello everyone.

Hi

Oh My !! I just fed a troll .. unix.stackexchange.com/questions/235831/…

@Archemar Why? Is that wrong?

Yes, it's pedantic but you may as well edit it.

And why is he trolling? That's a pretty harsh word. As far as I can tell he's pointing out a (minor) mistake.

@terdon Agreed, trolling seems a bit strong.

troll or pedant?

Petroll!

Well, answer edited

I post things like that myself. I'd hate it if people called me a troll.

pedroll

I don't roll no peds.

very droll

I just asked this on the Frying Pan, so...

Random poll: what is the most emotive word in the English language?

@terdon: I figured it out, see unix.stackexchange.com/a/236058/38906

@Archemar - BTW he's right. an int isn't an ssize_t, and discarding or ignoring a return value isn't the same as using the wrong type. In this case, it's relatively unimportant because ssize_t is close enough to an int on most systems to make no difference. in other cases, it could be the cause of a core dump...so it's not just trivial pedantry for the sake of trivial pedantry.

@cuonglm You're on Ubuntu, right? I don't seem to have that on my Debian:

$ sudo apt-get install language-pack-fr
[sudo] password for terdon:
Reading package lists... Done
Building dependency tree
Reading state information... Done
E: Unable to locate package language-pack-fr

Also, apt-cache search language-pack returns nothing.

Damn.

@terdon: No, I'm on Debian, my test machine is Ubuntu

@cuonglm WTF? How come I don't have it.

@cuonglm Ah, so the output you showed was on Ubuntu?

Yes

OK, I can find it on Ubuntu. However, the list of files doesn't make sense:

packages.ubuntu.com/trusty/all/language-pack-fr/filelist

@terdon That's not in Debian. What are you trying to do?

All it seems to provide is:

/usr/share/doc/language-pack-fr/changelog.gz
/usr/share/doc/language-pack-fr/copyright

And I know it's not in Debian, that's what we're discussing.

@terdon Probably a metapackage. See the dependencies.

@FaheemMitha Of what?

Ah, right, I see.

Of language-pack-fr.

Yes, got it in the end. You're right, it's language-pack-fr-base that does the work and that's a dependency.

But how come French works on my system if I haven't installed that? There must be some Debian equivalent.

@terdon: The language pack seems to be include when installing locale in Debian

@cuonglm Which one, the French one specifically?

And how did you check?

In my Debian, there is net-tools.mo in /usr/share/locale/fr/LC_MESSAGE

@cuonglm YES! OK, so my el locale doesn't have net-tools.mo but the fr one does! That's very probably it, thanks.

$ ls /usr/share/locale/{fr,pt,el}/LC_MESSAGES/net*
ls: cannot access /usr/share/locale/pt/LC_MESSAGES/net*: No such file or directory
ls: cannot access /usr/share/locale/el/LC_MESSAGES/net*: No such file or directory
/usr/share/locale/fr/LC_MESSAGES/net-tools.mo

And that was installed by the net-tools package:

$ apt-file  search /usr/share/locale/fr/LC_MESSAGES/net-tools.mo
net-tools: /usr/share/locale/fr/LC_MESSAGES/net-tools.mo

So, yes, it depends on what the specific package ships with.

@cuonglm OK. I wonder if I can simply copy the Ubuntu one over. Could you send me /usr/share/locale/pt/LC_MESSAGES/net-tools.mo please?

Ah, no, never mind. I'll get it from my Ubuntu VM.

@terdon Why do you need to copy an Ubuntu thing?

@FaheemMitha Because I can't find the Debian thing.

@terdon Oh. Well, you could post a question. :-)

I was going to but i) the OP I was trying to help got it to work on his Debian and ii) I got it to work on mine by copying the file from Ubuntu.

I can never remember when i need it - how do i invita a user to a chat?

@cas Go to their chat user profile and there's a button.

Or just create a chat room and leave a comment under the user's post asking them to join you there,

do they need to have previously logged in to chat.stackexchange.com? i.e. how do i find their chat profile?

@cas Ah, yes, they do.

They also need at least 20rep on any SE site to chat.

Who is this about?

frostware. he wants to edit my answer and replace my crappy python script with his (much better and presumably working and tested) script. i want to tell him that i don't mind him adding the script but replacing mine with it gives the impression that i wrote it when i didn't. he should add it instead.

ah, doesn't matter. i just "improved" his edit to make it clear that it was his code.

@terdon What was the missing file, and did you run apt-file on it?

@FaheemMitha Just scroll up.

@terdon Ok, I see.

(i can never get apt-file to work...maybe coz it's not designed to work with local mirrors)

@cas Don't see why. Maybe file a bug report.

@cas Hmm. I'm actually using LMDE but I've never seen any real difference between it and vanilla Debian apart from certain packages coming preinstalled. Perhaps this is one.

@faheem, no need to use apt-file when i have my own local mirror with Contents-* to grep.

ah, okay. duh. rtfm. run apt-file update first.

@FaheemMitha the club used to be more exclusive, when the only way you'd even know about Linux was from newsgroups, dialup BBS, obscure software boxes if your store happened to carry a boxed version of redhat, and CDs bundled with non-mainstream linux-specific and sometimes programming-related magazines and you were lucky enough to own a CD reader. These days its no special club. Most of the smartphone-using public worldwide carries Linux in their pocket and most are none-the-wiser.

^^

and the day linux will absolutely not be a club is the day Satya Nadella decides he loves linux so much that windows will be re-implemented as a full featured DE on top of Linux

@casey Technically, that's android. Which is a fork.

@FaheemMitha you can run a mainstream kernel on a fair number of devices (minus proprietary driver blobs). Thats Linux to me.

Huh, the CEO of MS is Indian-born. I didn't know that.

@casey Some of them. Possibly.

Not just Indian-born. Went to college there.

Take a look at recent kernels, there is a staggering amount of ARM support in there these days for all sorts of SoC systems.

You can argue Android isn't GNU (because it isn't), but it is certainly Linux

Unless of course, you want to argue Debian isn't Linux, its a fork too because they patch the kernel

@casey I thought there was a lot more patching to get to the Android fork.

@FaheemMitha most of that patching is to bring up new hardware support, and that stuff does flow upstream.

@casey Ok.

Well, I don't really care whether Android is Linux. It's mostly semantics, anyway.

e.g. I have a utilite with a freescale imx6 ARM SoC. When I got it you were stuck running a version of the kernel they provided patches for, but these days I can probably (haven't checked 100%) run a mainstream kernel on it and get everything except hardware accelerated video (propietary blob...)

@casey Do you actually run a mainstream Linux distribution on any of your phones?

I run stock android on my android devices

which for my devices is "almost pure" and "pure" android

Which is Linux 3.4.42-gfff168c + Motorolla's nearly-pure Android and 3.4.0-g8aa6344 + AOSP pure android.

Is anyone willing to help me figure out how to add a second bridge on a server? The bridge would be between two virtual machines. I am guessing it just involves a few nmcli commands such as nmcli con add type bridge... But I am new to this, and want to learn how to do it properly:

I read the documentation at this link: access.redhat.com/documentation/en-US/Red_Hat_Enterprise_Linux/…

My guess is that I should create the bridge from INSIDE the datavm GUEST virtual machine, and then also create something within the webvm GUEST virtual machine. But I would like to have more than a guess to go on.

I just read this entire link: happyassassin.net/2014/07/23/…

My guess is that, from inside the datavm, I should create a bridge connection and a slave interface connection, with the webvm being the slave. But this is just a guess.

@CodeMed I don't have time to do a detailed answer (or go through it step by step...) but basically you'd add a second bridge to the host. Then you'd add a second network interface to both VMs, connected to that host bridge. Then finally you'd configure the new network interfaces inside each guest.

But I have to run.

Thank you very much. I updated my OP in case another person has time to look into this. And I am continuing to read.

In the host, I might imagine adding the second bridge as follows:

nmcli con add type bridge ifname br1

But my google research is not giving clear instructions for adding a second network interface to each VM. Does anyone else have any suggestions?

maybe something like virsh attach-interface ? (I don't know anything about virsh - just curious -> googled it)

@JeffSchaller Thank you. Where would I type that?

Inside datavm, I have:

I wonder if I should add an /etc/sysconfig/network-scripts/ifcfg-eth1 file to configure a new virtual ethernet connection, to attach to the new bridge.

I would guess that the virsh config would happen outside of the VM (whether you think of that as 'up' or 'down' a level)

@JeffSchaller This link seems to imply that running virsh attach-interface at the host, would cause nmcli con show in the guest to show a new ethernet interface. But before I run the command, it would be nice to make sure that it will produce private ip addresses in a new network that is only shared by the two vms, so that we know data will not exit the physical box: kashyapc.fedorapeople.org/virt/add-network-card-in-guest.txt

@CodeMed you are now -- well, you started -- outside of my experience with virsh :)

@JeffSchaller no worries. you got me to read man virsh, and to find in it:

my guess is, after running nmcli con add type bridge ifname br1 on the host, I might type:

virsh attach-interface datavm bridge br1

But I do not know what to use for the other arguments.

@CodeMed me neither, sorry!

I tried the commands I imagined based on suggestions from @JeffShaller and @derobert, and got the following results:

well, you'd have to configure IP addresses on the interfaces. Use some RFC1918 space.

data vm could use 10.1.1.1, and web vm could use 10.1.1.2, but the man virsh excerpt above did not seem to have an argument for ip.

@CodeMed as long as that's a different subnet as your other IPs, yeah that'd work. But you configure IPs on the guests, not on the host. Using nmcli or whatever on the guest.

the local area network uses 10.0.0.x, so is it correct to expect that using 10.1.1.x for the two virtual machines in the new bridge would not cause conflict with the lan?

if the local network is 10.0.0.x/24

and you're using 10.1.1.x/16 or smaller subnet

I'm heading home. Good night everyone.

Thank you.

@casey Well, Linux or not, my current phone is currently running Android, but has a proprietary feel. Lots of crappy Google apps which I'm apparently not allowed to remove, but can "disable". Yay. Occasional popups. Very different from Debian.

I guess this is proof, if any was needed, that something can run on top of the Linux kernel (or something like it) and still look like a proprietary system.

Hi @derobert.

Would the correct syntax to add the ip address within each vm be:

nmcli con modify 'Wired connection 1` ip4.addresses "10.1.1.0/16 10.1.1.2"

Typing that gives a > on the next line:

@CodeMed you are mixing forward and back quotes. That works for typography, but not your shell.

@casey How do I use this information?

By fixing the mismatch. 'Good', 'Bad`

I thought that there was just one key for ' and another key for ". But in any event, I retyped it, and a downstream error was produced:

@CodeMed can't help there, I don't use nmcli for networking.

But I'd guess you mean to use "10.1.1.2/16"

@FaheemMitha was there ever any doubt? I don't think GPL extends its reach through system calls. You can build any userland on top of that.

And no surprise the dalvik VM isn't reminiscent of Debian with a GNU userland.

@casey dalvik?

@casey Well, not surprising, no. I suppose my comment amounts to says that running the Linux kernel is not equivalent to the free software "experience".

Never mind, dalvik is part of Android.

@FaheemMitha dalvik is the java VM that everything you interact with ob your phone runs on top of.

@casey Yes, I see. I'm not really familiar with Android.

Sounds like you are, though.

For anyone who is interested, i updated my OP with the results of trying to set the ip addresses. The nmcli con mod command did not throw an error per se, but yet the two virtual machines are still not able to ping each other. Does anyone have any ideas?

I posted the contents of /etc/sysconfig/network-scripts/ifcfg-Wired_connection_1 in both virtual machines, but I do not know if the problem is in /etc/sysconfig/network-scripts/ifcfg-Wired_connection_1 on the virtual machines, or if something else needs to be added to the host. or if it is a firewall issue.