Add –initrd before kernel_image if you need an initrd.

dude, we know

kernel_image modules_image

let me relook 1 sec

binary-arch. makes everything you usually need (image and headers)

can noone tell me a tool to recover files on a ext3?

!undelete

hmm… undelete is great with Midnight Commander (mc). Just unmount the fs and select “Undelete files” in mc, or doesn’t work for journalled filesystems.

wols your right i never had to add kernel-image before i dont know whats going on now

s/, ever/now

I’m not seeing any target in that make-kpkg command though and the docs don’t list a default target that I can see,

it doent have a default target

you’re right

if you dont give it a target, it doesnt make one

it has been all this time

BS

that exact mantra result is exactly what i build with everytime less the version stuff

ffs

please stop that.

why not simply test the advice given?

although it can be configured through ~/.kernel-pkg.conf or /etc/kernel-pkg.conf, I’m guessing that’s unlikely.

well i have no doubt that it will work im just saying why all of a sudden has it changed

if you don’t know how to adopt the kp mantra to your needs, you shouldn’t build kernels

!vlc

from memory, vlc is the VideoLAN Client. It’s a MPEG2 client/server program for broadcasting. It can also be used standalone as an normal MPEG2/DVD player. See http://www.videolan.org, or apt-gettable.

it hasn’t, and discussing this is off-topic here

azeem havent you read…i made prob at least 50 kernels now all of sudden it stopped working you all know i have been building kernels here

haven’t you read, you’re mistaken so drop it now or you’ll be silenced

bullshit, you fucked up when you pasted whatever command you did into that file

i pasted the mantra command (complete) then made a new command in there less the –version stuff

I want to play games like quake and alien arena

i might have accidently edited it ill look 1 sec

is there apt-get packages for quake and alien arena?

which part of “drop it” did you not understand?

i am new to debian and don’t know how to install these games - quake and alien arena, please help

/msg dpkg search

does anyone know about getting madwifi working on usb, or do I still need to use ndiswrapper? (woot.com features the product)

!info openarena

(A fast-paced 3D first-person shooter), section games, is optional. Version: 0.7.0-1 (sid), Packaged size: 691 kB, Installed size: 1580 kB

WHAT’s the difference modprobe via-agp; dpkg-reconfigure xserver-xorg and dpkg-reconfigure xserver-xorg ?? please

madwifi doesnt support usb devices

ok, I knew it didnt, but the last i knew was about a year ago

# !info openarena !info: event not found

some chips are supported by the zd1211rw driver though

is that driver in the madwifi one or is it a different driver?

read the basic Debian documentation on how to install packages, or run “synaptic” as root

completely different. its hosted on sourceforge

azeem, i have already found quake2 and installing it thanks

ok, will check it out thanks a ton

http://www.reghardware.co.uk/2007/09/04/nopopo/ - cant w8 until they run windows based laptops with this… i allways wanted to piss on one to make it work better XD

np, hope it works for ya

loading oss sound output driver, ok .. /dev/dsp: Device or resource busy

so… how do i get vlc in lenny?

Could not open /dev/dsp.

i can’t seem to find any lenny vlc

what may be the problem with /dev/dsp? what should i do?

close other applications that are using sound

but i don’t think anything else is using sound well gmail chat does a ding ding when a message arives arrives* can that be a problem?

yes

awww!! debian can not let multiple software share the sound. awww!!!

what other program is it? and why are you using OSS? use alsa.

that has nothing to do with Debian

thats not a debian problem

how can i switch to alsa? dpkg-reconfigure alsa?

eh it’s an application-level setting

from where can i do that? i tried running /usr/games/quake2 when this problem arised but my other applications requiring sound run well except quake2 mode changed to +l meaning?

azeem do i need to goto another channel to ask help with deb make kpkg? as i need to compile this kernel as its not set.

we keep giving you advice, you keep ignoring it.

i can only ask azeem until he lets me know

anyone have advice getting vlc in lenny?

its been missing for some time !why is vlc not in testing

vlc is not in testing for the reasons listed in http://bjorn.haxx.se/debian/testing.pl?package=vlc

^^

I said you should stop argueing that your broken command-line has worked in the past

thanks

I didn’t say “stop using kp mantra”

why does the mips compile affect the x86 compile?

i wish you did

what’s the best tool for a newb to create deb packages?

ok and thnx for the people that are telling me the work around but im just trying to get back to the way it was working…let me do some reading and see if i can get this to work the way it was

because versions in testing must be the same across different arches

ah

btw, why are you trying to recompile your kernel? You told us about 6 hours ago that it was all fixed and you were working ok.

the current kp mantra doesn’t even include the initrd, why don’t you just use that?

so then… i can just compile it on x86, right?

yes

azeem im trying to get the mantra to come up right now…let me see if i got the command to make it list it to me

k

eh?

i want to check whether a particular package ‘foo’ is installed on my debian.

got it

/msg dpkg kp mantra

what is the command i should give?

dpkg -l foo

thnx

if i call my helloworld system call in my program and i run strace -c ./prog should i see helloworld or ipc_subcall?

ask in a kernel-related channel

hi, why when somebody is downloading a file for more than 1 hour its connection expires? (i use apache2)

why are debian package commands revolve around two different commands, apt-get and dpkg?

guys, is there a way to tell alien to generate packages for i386 in an amd64 box? i get dpkg-gencontrol: error: current build architecture amd64 does not appear in package’s list (i386)

if I install something using apt-get, why can’t I check its status and remove it using apt-get itself?

you can remove it, and you can check its status with apt-cache show foo

rabbit, see apt-get as a comfortable way to use dpkg

rabbit, to remove apt-get remove pkg

ok

wen I do a apt-get install, it seems to hang on connectting to ftp.us.debian.org, is the ftp site getting slammer or down? I got thru once, just really slow!

how to get a list of softwares installeds in machine??

try again, there is more than one server behind ftp.us.debian.org, maybe one of them is down dpkg -l | grep ^ii

dpkg –get-selections

in the output of dpkg-l there are other beginnings like rc, ri, etc. what do they mean?

it denotes if it’s been installed or not, removed or never installed, etc

man dpkg

have a look at the first 3 lines of output

look up the top of the list for a guide rabbit

for both, program and config files

yes, got it, thanks

How to have a list of installed packages, to install in another computer?

if you used dpkg –get-selections then dpkg –set-selections !tell S4nD3r about debian clone

hi guys I had a tremendous amount of help from here the other day and we nearly configured my wireless card but Im still having trouble. here is my post explaining my problem. id really appreciate help solving this problem. http://www.linuxquestions.org/questions/showpost.php?p=2880927&postcount=5

!xchat-gnome

what do you mean with “join”? is this a WPA-secured network?

associate?

azeem in my network icon top right of desktop i can select join network.

off

azeem not secured

at the cafe?

yeah

MAC address filtered?

not the same one a wifi zone cafe

i saw your pings, they were good

it’s an ipw3945 isn’t it?

Any good tutorials on software RAID under Etch ?

do you have control over the access point?

themill yes

I think we’re up to the wireless now the wired is working and the ipw3945 modules are loaded.

liable so i was close with ethernet but we change tac

ah

azeem i dont think so .

do you have control over the access point?

with ipw3945 I’d be tempted just to use network-manager-kde. It’s a doddle to get working then

what happens if you run “dhclient eth2″ as root?

i would be more tempted to kill all the gui things

after you’re connected to the WLAN of course btw, you got an ipv6 address

very true. but getting wep keys right is a pita I gave up on wpa_supplicant etc when the config file format and location changed again on me.

ok i can do these things as you tell me as i dont have any other computers to work from. im in my xp dual boot at the moment.

is there a chance of forcing re-read of the partition table without reboot?

How do I remove a package from the list of packages to be installed? I’m attempting to remove libc6 from this list because it now requires linux 2.6, but linux 2.6 doesnt support my touchscreen

so to get this right: your access point is an internet cafe which just lets everybody use it?

and you’re right, kwifimanager is horrible and should die.

I agree

your touchscreen doesn’t support linux 2.6? i’d complain to them

^

you can’t

Re-reading the partition table failed with error 16: Device or resource busy. The kernel still uses the old table.

ill try dhclent eth2 and netwrok manager kde. brb

wtf

any hope for that? i can’t afford a reboot now

ha

which version of Debian are you running?

unstable

oh, i dont think that so much, i just wouldnt be using these things untill you are sure the cards is working well

I guess I have to reinstall eh?

unstable doesn’t support 2.4

this sucks hehe

what touchscreen?

got a recommendation like oldstable perhaps?

azeem yes

or install the etch 2.6 kernel

Fujitsu B142

and then upgrade libc6 and then upgrade to the unstable 2.6 kernel

the problem is, 2.6 doesnt provide a direct psaux layer

azeem an internet cafe that lets everybody use it

so my TS driver cant set the TS to absolute mode

i see you finally did manage to reinstall linux-image

network-manager-kde is slick; kwifimanager is horrible. The gui stuff is pretty hard to debug when it’s not working, but ipw* generally just works and it’s easy.

then it’s either your touchscreen, or never upgrading again

in 2.6 it behaves like a 10″ touchpad :X

yeah wyzard took me through it veRY patiently.

Never upgrade it is thanks does oldstable like 2.4?

i know, i have a 3945 card

oldstable? stable still supports 2.4, if you know how to do it

nod, there’s stable, oldstable, testing, unstable

then it could likely be supported

okies

yes, I know

xserver-xorg-input-evtouch

does that work with the B142? the problem is, the driver needs to send commands out the psaux port the device is in relative mode on poweron (much like a touchpad)

and here’s the configuration on this page somewhere: http://homepage.sunrise.ch/mysunrise/joerg.hau/linux/b142.htm a href=”http://homepage.sunrise.ch/mysunrise/joerg.hau/linux/b142.htm”http://homepage.sunrise.ch/mysunrise/joerg.hau/linux/b142.htm/a although that uses a different driver, oops

so does iwconfig show it as associated?

if it works with my TS, I’ll be happy

anyway, evtouch should work lbtouch looks outdated and unmaintained

how can i do a relpy-all with mutt?

check /usr/share/doc/xserver-xorg-input-evtouch/*

read the manual, “group reply”

associated but no wep key

it should give you some hints for configuring it

most cafe’s do open access but make you login via web browser

Hm I’ll have to try this cool links now to unbreak my 2.6 image install

everytime I install a pkg, it tried to upate webmin, could I disable this somehow, anyone know?

I’d really like to use 2.6

adept manager ( kde frontend for apt-get ) cannot handle deb files with user interaction: for example, i want to install sun-jdk-6 and the eula must accepted. to solve this problem, i need to change the default window class for apt-get interaction. from dialog to qt maybe. any ideas? badept manager ( kde frontend for apt-get ) cannot handle deb files with user interaction: for example, i want to install sun-jdk-6 and the eula must accepted. to solve this problem, i need to change the default window class for apt-get interaction. from dialog to qt maybe. any ideas?/b

an awful lot have the password of the day written up on a chalk board next to the menu too

heh, never seen that myself anyway, he decided to bail

i know, that there is a programme, but i dont know how it works

debconf

if this is the same cafe, the ethernet didnt give out leases, i guess the wireless doesnt either

dpkg-reconfigure debconf

themill, yes thats it, thanks

how do you tell fdisk to use MB or GB instead of cylinders when creating a partition? i can’t figure out how to make exactly 15GB one

use cfdisk

hm..

ive setup a chroot jail for sftp, but when i try and log into it via sftp it doesnt work and my auth.log says ‘PAM-env[3536]: Unable to open config file: No such file or directory’ but i cannot figure out what config file it cannot open

liable oh wow thank you

yw

can’t be easier now

hello, I’m trying to install madwifi on etch, I entered m-a prepare, but have trouble with kernel headers and 404: http://www.pastebin.ca/681165 I tried the us mirror also, and ran apt-get update. is there something I could do to fix the url?

2.6.18-5 is the new kernel

hark hmm, so I just need to update to that?

you need to aptitude update && aptitude upgrade first

ok then

yes, it’s a security update

was about to do upgrade anyway but forgot btw, is it better to run aptitude or apt-get, is there any difference?

someone had that problem yesterday as well, but i can still get the -4 headers

!aptitude

aptitude-fu, http://tinyurl.com/2ebd4m http://tinyurl.com/2hw8nx http://tinyurl.com/2y783t

guess I should use apt-get then since I haven’t used aptitude on this one

!factinfo why not aptitude

7 2007 (148 days); it has been requested 76 times, last by hark, 20s

you can switch to aptitude just fine. but switching backj is not good

thank you for all the information

i have copied over /etc/passwd + /etc/groups into my chroot but the permissions still look like this 2 0 0 ? users + groups i mean

means that your nss isn’t working. those just mean uid and gid of 0, i.e., root

what is smallest install I can create with debian ?

about 248MB

need to create ‘virtual appliance’ appliance with a copy of our software Failed to partition the selected disk i’m using 316Mb disk

uh, wait

hm there is no reiserfs type in cfdisk

using business card

but for some reason it shows my partition as reiserfs… strange

DustyBin, try to copy nsswitch.conf as well

i’m sure i gave it 83…

http://www.us.debian.org/releases/stable/i386/apds02.html.en

!deutsch

from memory, de is deutschsprachige Hilfe bekommt ihr in #debian.de (auf irc.oftc.net und irc.freenode.net) - German speaking users please go to #debian.de (on irc.oftc.net and freenode)

!list

one warez list being sent to asd

lmao !0day

is there any way to make dhcp3 only listen on a given interface?

/etc/default/dhcpd

dhclient iface

3

thanks

oh, the server?

ks, maybe you could install the virtual machine on a bigger disk, then resize it?

liable thanks, i see now joga i just figured out it’s qemu qcow, so it ‘ll occupy only this much

ah yes, in that case out of curiosity, what sort of virtual appliance is it?

liable do you have an idea why would cfdisk call type 83 as reiserfs? while fdisk says my partition is Linux… a bug or is there a problem?

while declaring 1Gb it’ll have 300Mb

83 is linux in general

liable i know

joga an nbd app

ks, at least it can grow now

liable but cfdisk 2.12r says it’s reiser

ks, nbd?

no idea

liable while i have no choice of reiser in there at all and never tried to set reiser for it

joga network block device

kind of strange, huh?

yep

ks, ah ok

joga is business card a good distro to build a headless install ? or it doesn’t matter ?

good afternoon I’m here on old debian (3.1) with logrotate 3.7 and he says to me “error: unable to open /export/log/apache/feipecus/error.log.1 for compression” aparrentily all logs that I compress have this problem

azeem i tried the dhlclient eth2 command and this is the output. I am going to move to another cafe and try my car card there

unable to open /export/log/apache/feipecus/error.log.1 for compression

http://pastebin.ca/681160

my dhlclient output can be seen here http://www.linuxquestions.org/questions/showpost.php?p=2880927&postcount=5

thats good! you actually got a lease..

Ok but i cant load web pages

ks, I don’t know, really

anyone know how to insert a “tab” insertion into vi’s :%s?

dns borked?

your net should have worked. put 72.14.207.99 in you web browser and see if it works

Ronin im really new to this so im not sure on those terms

anyone successfully run Squirrelmail on Etch? I’m having lots of trouble…the link cannot be found when I use the browser.

check your /etc/resolv.conf it should have 192.168.0.1 in it

ok is there anything elese as i have to go offline again

someone can help me? in web not have much that doubts

I wish to back-up the data on my system (~3GB total for /) would tar be a good solution?

bbl thanks for helping i need to change internet cafes

is the file there?

the files in pastebin is only one of various that I have problem

tar is fine for bulk backups use the compression switch just make sure you have enough free space

tar -czvf backup.tar /; after doing cd /

hi, i installed tomcat5.5, now when i try to restart it, it return: error writing to /var/lib/tomcat5.5/logs/catalina_2007-09-04.log ? someone knows why this happens ?

I am just not sure if it will get recursive on me

hello folks, is it a good idea to replace NIS with LDAP for authentication?

so try to back up the back up file being written

specify an exception list

I have free space

proc, dev, your backup tarball is the file there

I guess the -X option is what I want

what is the fastest, while still relatively secure (encrypted network traffic and such), way to mount a directory on a remote machine and allow a bunch of users to access it at once? performance being the n goal.

this file have 500M, in partition have 900M

read-only btw

the problem is really extrange …

but I would rather pass them on the command line

this is my logrotate.conf and apache conf for logrotate http://pastebin.ca/681208

a file list is easier to type, but I’m not typing

Hello everbody, I’m trying to install a wifi adapter with an atml chipset and Debian sid. When I type m-a a-i at76c503a I’v got an error during the compilation

is the file there?

is there any know problem with the kernel 2.6.22 and this driver ?

hi - i`m having problems with viewing pdfs on a standard debian install, it works but there is something wrong with the fonts. any suggestions?

does anyone know a way of PAM giving more detail on log files, PAM-env[3843]: Unable to open config file: No such file or directory means nothing to me

you mean .1 file or original file? the original file yes, the .1 file not, apparentlly he isn’t created or created but deleted

if the file isn’t present, then compress can’t compress it find out why logrotate isn’t creating it permissions? no source to rotate?

have source and permission is ok, 775 for directory and 664 for files bhave source and permission is ok, 775 for directory and 664 for files/b

ahh this brings back memories of my slackware days

delaycompress

arghhhh

as well as notifempty

when I run with -d he only debug but not move file ……………..

I would put missingok in the there too

thx for atention but in debug mode in really he not move and compress file thx

Is my particular copy of cpp for gcc-4.1 borked? file reports /usr/bin/cpp-4.1 as a Posix TAR archive. while file on /usr/bin/cpp-3.4 reports back as: ELF 32-bit LSB executable

i ran memtest86 for 2 hours - 8 passes and return no error. should i stop the test now or should i let it run longer?

Run it for 4 hours to be sure.

I’d say its ok but tyou can try longer if you wish

CharminTheMoose, borked!

ok

I dont bother unless I have areal problem

run it overnight to be sure

basix, damn

can you tar –list the file?

Turn on the heater in the room and run memtest

pickcoder, nothing seems to be coming up just hangs

has anyone here used rssh? whoops shouldnt of said that

!anyone

polls

any ideas how to determine with dimm is bad base on the mcelog?

run memtest86

can memtest86 shows which memory dimm is bad?

no

it only show if there are memory problem. i need a way to id which dimm is bad

take one out, run the test rinse, lather, repeat

So, correct me if I’m wrong, but if you don’t set a root password a la Ubuntu and you boot the computer into single-user mode, then you’re immediately dumped in a root shell?

you cant do that easily

i can do that but i will take a very long time for system with 16 dimms. i need a faster way to id the problem dimms

Because someone just opened a root shell, and this kinda freaks me out.

the install doesnt let you not set one iirc

It does in expert mode.

testing a new freeNX setup and amazed that a fluxbox seasons is way slower then a KDE or Gnome one, any debugging comments apretiated

memtest86 cant know the on physical ( on die ) mapping.. consult your bios/chipset manual

2.6.11 apparently has the driver in the kernel tree, but debian doesnt enable it :X (off to build a kernel!)

oh really?

It looks like it needs more polishing, though. F’rinstance, modifying the default gksu and kdesu config files to use sudo instead of su if you don’t set a root password.

cpu 1 ADDR 5e4f8040 error

you can figure it out empirically though

Anyone using lazarus ?

btw, thanks for the help

And it seems, it also needs to NOT dump you in a root shell without a password if you boot into single user mode.

that error does not mean that you have a physically broken anything..

echo $((0×5e4f8040)) ~ 1582268480 ~ 1.5g range?

i guess expert mode expects you know what you are doing when you dont set a root password

there is problem with dimm on cpu 1 (first cpu for a dual cores processor), right?

are you still running fedora ?

i dont know this error

nope

Hi!

most commonly such errors are psu or heat, less commonly hardware breakage

where can I get software that is easy to install on debian?..

man apt-get

Yeah, but it could automate a few things. This thing about dumping you in a root shell in single user mode, well, I just thought you wouldn’t be able to do anything at all if there was no password, not that you’d be dumped in a root shell.

yes and no. I do have some systems running centos, fc, rhel and debian.

like which software?

Debian is a distro. It serves its own software. That’s what being a distro is all about.

and it doesnt get any easier to install software then aptitude install foo

if you set a root password, it does ask you, so i guess thats the point. anyway, if you have physical access its a moot point anyway

unix gives you the gun and ammo but doesn’t explain where to aim it or stop you from shooting at your feet

it explains, it just doesn’t force you to follow what it says and it’s targeted at more than a pre-school level of reading comprehension

Well, there’s physical access, but there are BIOS passwords and someone in the lab making sure nobody with a screwdriver gets near to the machines.

I wasn’t able to install any software that wasn’t on the cd(s)

!tell Jo_R about sources.list

Then don’t point it at your foot!

!tell Jo_R about apt howto

the niceties of pam_unix’s nullok and the history behind why it’s a default takes some digging up.

the point is, if they can see the grub screen (if grub doesnt have a password) then they can get single mode without a password.

that’s exactly my point !give cookie to EvilGuru

grub has a password.

so, you have bios passwords, grub passwords, yet you didnt set a root password?

single mode doesn’t ask for one..

the default grub line does

!I forgot root’s password

1) ‘linux init=/bin/sh’ at a lilo boot prompt (hold Shift while booting to make one appear), 2) ‘fsck’ your root file system, 3) ‘mount -o remount,rw /’, 4) ‘passwd root’ 5) ‘mount -o remount,ro /’ 6) reboot -d -f (exec /sbin/init should work); For GRUB (e.g. etch, sarge): in step 1, press ‘e’ to edit the kernel setting (add init=/bin/sh to the end of it), ‘b’ to boot. Steps 2-6 are the same.

I didn’t think it was possible to login to root without a password! There’s a difference between a blank password and no password, y’know. Accounts without a password are supposed to be disabled.

er what? take a look at pam_unix(8), nullok option.

No, I didn’t forget the root password. I just didn’t set one a la Ubuntu, but I’ve discovered that this means that if you boot into single user mode, you get dumped into a shell without a password prompt.

I see he just cant log in

nobody doesn’t login. That doesn’t mean that nobody doesn’t do things.

but it doesn’t have a password and it’s not disabled

you sure about the no password bit? mine has :*: in shadow.

semantics g

… matter.

hi all

Hm, any character besides an ‘x’ is treated as an encrypted password, from passwd(5). So when you set it to some character, does this mean that if you found a key that hashed to that single character, then you could login?

yes, and given that the first character is *salt*, you’d be wanting something which had a salt of * and produced a null cyphertext

theoretically

good luck on that quest.

Oh, crypt(5) says such a salt is simply never an output.

unless there’s a defect in crypt(3) hence, possible, just improbable.

not a show stopper.. but still, epiphany freezes when signing out of gmail after an embedded html chat session. This was happening on lenny, now on Sid also. I upgraded to Sid last week how to get it fixed?

file a bug

!sid upgrade

Sid is NOT AN UPGRADE! Sid is meant for developers and testers, so stay away from it if you’re neither. Use stable or testing instead.

this sucks happy white 0wled xmas

MOTHERFUCKING SID IS NOT A MOTHERFUCKING UPGRADE!

motherfucker!

heheh

Why don’t we ever say “fatherfucker”?

same reason we say motherfucker and not fatherfucker

don’t you get enough of that with the pr0n you watch ?

heh heh heh

does anyone have recommendations on a really lightweight graphical browser? the system is 233mmx/96MB

w3m

dillo is pretty … ehm

http://www.prevedgame.ru/in.php?id=764725

spammers must die!

simonrvn, w3m is graphical?

Emacs!

heh I use emacs for slime

Emacs is lightweight enough for 96 megs of ram. And it displays images. I’m not sure if that’s all it takes to count as “graphical”.

i’d just like for some web stuff to not bug out, otherwise elinks would be fine and to see pics

pics, eh?

do you need some extension to emacs to browse with pictures?

w3

then what was that ? .. seems like spam to me

!info w3m-el

hello there

 

Cheap Mysql5 Web Space!

 

w3m-el: (a simple Emacs interface of w3m), section web, is optional. Version: 1.4.4-3 (sid), Packaged size: 622 kB, Installed size: 2404 kB

I guess it’s just w3m after all.

how can I install DSL connection ?

pppoe

sera a tutti.

shows images, “graphical” enough for me

!it

per aiuto in lingua italiana, provi il canale #debian-it sulla rete irc.freenode.net (non irc.debian.org). C’è anche il canale #linux-it là.

so, like, you know, how about posting the english version instead?

of anti spam filter webmail host ?

not all of us speak russian here

does anyone ?

How many people have actually ever gone to a language-specific Debian channel when told to do so instead of just switching to English?

i want him to prove it isn’t spam

I do, a little.

bah .. let’s just call ops!

i almost did

how can I install dsl connection ?? pppoe. pppoeconf doesn’t work: command not found! anyone?

Hello all, I’m running Debian etch on a Dell PE2950 with gigabit NICs… my network performance is bad, though (looks like it maxes out at about 5-10 megabytes/sec). The processor is not loaded, ethtool reports correct line settings, and I’ve double-checked that there is nothing between this machine and my test machine which is not cat 5e or higher.

it is spam

run as root

Is there any tuning that I generally have to do to get gigabit ethernet to work properly on modern Linuxes?

ethtool maybe or mii-tool

I’ve tryed

jumbo frames, use tcp.

are all devices capable and reporting a 1Gbps link?

Yep, spam. Cheap online games.

are u sure that there is only this way to install dsl connection with pppoe ?

is it a freshly installed system?

!ops Vitaly spamming URLs

Hydroxide, bob2, caphuso, dondelelcaro, doogie, eeyore-, ElectricElf, ):, helix, ljlane, LoRez, RichiH, mentor, Netsnipe, TML, walters, xk, gravity, azeem, Maulkin, stew, peterS, Alife, Myon, Ganneff, Maulkin, weasel, zobel: simonrvn complains about: Vitaly spamming URLs

yep..

who is ):?

My test setting has only two hosts plus cabling, and the Debian end reports auto-negotiated 1gbps via ethtool. The other end is Win XP, though… maybe I should try both ends Debian+gigE and see if I still have problems

I don’t know if it helps, but I often install machines from the netinst cd, but have no network, so I exit, install the needed packages and off you go. similar situation with you?

hi guys… where did “network-manager-gnome” go ??

directly connected to each other?

!info network-manager

what application/protocol are you using as your benchmark?

network-manager: (network management framework daemon), section net, is optional. Version: 0.6.5-1 (sid), Packaged size: 369 kB, Installed size: 1516 kB

http download from apache

or connected through a switch/hub/router?

change your card mtu to 4096 or so (jumbo frames) repeat the test case; look at the debian end in a recent top at si%, hi%

I tried netcat earlier and that seemed to show good performance (70 megabytes/sec), but that was weeks ago. Now I’m trying to hand the problem off to our network people

you can leave gracefully, or get kickbanned. take your pick.

Nothing seems particularly badly loaded on the Debian end

and also run vmstat 1 and log the “in” column paste some actualy numbers please

take it easy

k. I’ll see if I can put together a test setup that doesn’t involve disconnecting from the network. My client machine’s only got a single NIC

bye

are there any ethernet errors logged per ifconfig ?

In previous tests, enabling jumbo frames _decreased_ the performance in the real application (it’s a perforce server). Right now I’m ignoring that fact until I figure out what’s going on in a much simpler setup than the production setup. Sec. Lemme try to get a test setup so I can debug and give you numbers without disconnecting from IRC. I’ll also try reenabling jumbo frames for the interface I’m using for my testing.

n/p

if it’s going at 10MB/s, something is happening other than jumbo frames that’s roughly what you’d expect from 100Mbps

nb jumbo frames gets to be enabled at source and sink - and in your production sense, also on everything else on that (broadcast) segment yes; hence the desire to see interrupt rate.

zinx, 0.o

arh… this is driving me crazy! I’ve installed debian in a LAN so I had static IP. But now I moved computer and I wanna install dsl connection. And when I do as root: pppoe, pppoeconf it says command not found what can I do to install dsl ??

No errors logged in ifconfig, BTW. I’ll be able to get another machine set up as the test client (still WinXP), but I won’t be able to cut and paste. I can retype interrupt rates and whatnot from ‘top’, though.

50-100MB/s with a tiny mss will kick his arse to kingdom come; interrupt rate is a reasonable proxy for “holy small packets batman”

install the pppoeconf package?

how If I don’t have connection?

how do i add etch/devel packages to my sources.list ?

!installit

*** NEWS FLASH! *** Your box does not come with every app, tool and utility known to debiankind installed already. If you find that the program you’ve been told to use isn’t there, install it. Also ask me about search. if someone suggests an application to you, it’s highly likely that it’s available via apt-get

well, it should be on the debian cd under /cdrom/pools/…

pool/ fyi

s/pools/pool/

so I just put in cd and instsall as normal: apt-get install ppppeconf and debian will know where to search ?

that’ll do, open up a nice big fat long flow and wait for the si/hi% and vmstat in numbers to settle - then give us approximation

no, because apt-get relies on package sources

why don’t you try?

if s/he’s got the cd in sources.list, yes it will work

true. so either add the cd to the apt sources or go old-fashioned dpkg –install

Okay, this is from downloading a file from apache over gigabit at around 4 megabytes/sec, with _no_ jumbo frames: http://paste.debian.net/36225 That’s the top output

ok barely touching metal; vmstat 1?

http://paste.debian.net/36227

not too obscene; 2s.

This server is lightly loaded in production, BTW, sorry… if you see load, it might not necessarily be test-related I’m trying jumbo frames now

yeah that’s perfect

what’s it serving, cheezwhiz?

You say I have to enable it on both sides, right?

hi

yep; it’s link-mtu (framelen) before we get that far - “ip route list table all | grep ta.rg.et.ad” - see if there’s a cached entry? interested in mtu/rtt/advmss etc bits there’s a shock

no surprise there

“sudo ip route list table all | grep 10.150.150.20″ did not produce any output. Did I do it right?

im having problems installing debian with a linksys usb ethernet USB200M. i’ve heard people using it, but it’s not detecting my hardware. . I guess.

10.150.150.150 is my server’s test interface, 10.150.150.20 is the client machine

start the flow up again first; then try but yes; should be correct; eyeball without the grep to see what I’m on about. linux route cache is a funky funky thing as far as host implementations go

greetings

quick question

quick answer

my initial suspicion is that your WinXP box is using a too-small tcp window size; you’d confirm this with packet capture; I don’t mind eyeballing that if you want to record one with tcpdump -w somefile and put it up for download.

When i create a partition that uses the entire space of the disk howcome its not fully avaliable to me its like a little of it is saved for something

muaddib, a bit is always used by the filesystem

root retains 5%

by the way, is xine able to play mp4 format ?

in what way

or rather, which app is able to play mp4 format ?

bandwidth delay product on 1Gbps link is big enough that 100base hardcoded defaults for window size (ie, no window scaling) will hurt

Your funny hahahaha NOT!

muaddib, the filesystem needs place to safe what files there are, what directories there are and all

inodes?

muaddib, some afaik

check out this http://xinehq.de/index.php/faq

i wasn’t joking

thanks for the help but i was looking for a quick yes or no answer!

witf

why would they think i’d lie …

what else

I’m not having any luck enabling jumbo frames on the Win XP side. I will grab a packet capture as you recommend.

muaddib, ahm, sizes of the files and all kinds of information you see if you do “stat filename”

i already gave you the answer

muaddib, how much of the space is it you can’t use?

well, you are in the wrong channel then, we prefer people who are able to read

when you create an ext2/3 file system .. 5%(by default) is reserved for root

I believe that our switch is not jumbo-frame capable anyway (see my earlier comment about enabling jumbo frames resulting in reduced performance), FWIW

abrotman, interesting to know

Even though my current testing bypasses the switch

yeah .. too bad he’s listening to your jibberish

(joined 2 minutes ago and already learned something.)

Then your talking about the inode count then

tune2fs -l /dev/sda1 .. for example .. see all the Reserved this/that oh ffs ..

man tune2fs /reserved-blocks-percentage

…

wtf is wrong with that guy

muaddib, and abrotman, i’m not too well-informed i better be quiet

i give the *correct* answer .. and i get laughed at

*nod* fair enough; what jumbo frames does *once the other gremlins are cured* is give you higher throughput at the same interrupt rate b*nod* fair enough; what jumbo frames does *once the other gremlins are cured* is give you higher throughput at the same interrupt rate/b

#bash

you have plenty of headroom as far as interrupt rate is concerned; if you were nearer 8k/s than 1k/s I’d be worried.

*nod*

Thanks

BTW, enabling jumbo frames on the server side but not the client side results in… well, rather poor performance BTW, enabling jumbo frames on the server side but not the client side results in… well, rather poor performance

yes you can

abrotman, sorry about me talking “jibberish” before, i was trying to help

i know .. it’s okay .. it just hurts to be laughed at …

heheh

!comfort abrotman

There, there, abrotman. It’s OK. I’m here for you.

aaw

I’d imagine it would.

apt-get moo

framelen must be constant within a given datalink domain

i didn’t know about it either .. but i don’t usually come close to filling up / which is where it usually becomes a huge issue (didn’t know at one time)

otherwise the upper layer protocols can’t negotiate segment size

abrotman, well, i had my own “root filesys full, go delete some stuff” warnings to learn from XD though i tend to place too many things on / anyway…

there’s not knowing, and not knowing and being a dumbass / cluebie about it

it’s what happens after you crash a filesystem where someone has thoughtfully set that down to 0 to free up space, that’s hilarious

i thought you couldn’t tune that after you made the FS

You can

rm -rf /* should fix all root filesystem is full issues

xdr, rotfl

DO NOT DO THAT AGAIN

you know what i did recently

i believe you can

almost as funny in fact as watching the fully loaded jbod macarena its way out of the rack once you hit 90%+ full and fragmentation gives you io access patterns from hell

that can get you a k-line

!lart xdr

oh, that was really bad, be glad no OP didnt see that…

bah, lame LART

i did *DO NOT DO IT* dpkg –remove –force-depends libc6

!lart xdr

http://qix.demiurgestudios.com/~moise/slow-gigabit.pcap

!lart dante_2core

heh

XD look

seriously?

there’s things you don’t do

digestifying, gimme a few

thx

yes .. people get k-lined for taht .. don’t do it again

and things you never do again once you where dumb enough to do them

people wouldn’t say it if it weren’t serious

seriously you get a +q 24 hours

some people may not know better .. that’s a very dangerous thing to even jokingly suggest

I’m looking at it too, I’m just not all that familiar with grotty details of TCP tuning

I won’t

FFS, aptitude install grey-matter

hm so you didnt learn…

I thought it was so stupid so people wouldn’t do that

yeah your window size is crap mate; http://www.speedguide.net/files/TCPOptimizer.exe - grab this guy and clicky clicky; docs are http://www.speedguide.net/tcpoptimizer.php

thats even more stupid then

some people are that stupid …

even people who have not that experiences with *nix

just let it auto-everything you; set for 1Gbps link rate, retest; if it’s still crap pop up another packet capture.

hmm … well I supposed it is like einstein said

Okay — are you saying the server settings seem to be okay (I haven’t tuned them at all) and that I need to retune the clients?

sure, but he didnt give advice HOW to be stupid

hi Ive been trying to get my wirless card to access the internet. I can access a cafes wireless zone but not acces the web . my prblem and command output are here http://www.linuxquestions.org/questions/showpost.php?p=2880927&postcount=5

oh good .. xdr broke freenode

yep; the flow is starving

DUDE, we are in a Help Channel, people trust us, so why the fuck do you think what kind of advice you gave?

so can anyone shed some more light on my wireless internet access problem

Hello everybody

hi

I have some questions regarding Debian distro

still here?

Okay, thanks very much. I’m rebooting the test machine right now, I’ll retest from there and if everything goes okay I’ll retest with my machine over the production network.

Currently I use Fedora, and I would like to know how is Debian better than Fedora Currently I use Fedora, and I would like to know how is Debian better than Fedora

ok; there’s a possible explanation…

since I disabled my on-board sound so I’d only have 1 sound card neither work

what does “cat /proc/version” say?

!why debian

http://www.debian.org/intro/why_debian or http://people.debian.org/~srivasta/talks/why_debian/

alsaconf finds both (even though one is disabled)

in the ~2.6.19 timeframe linux started dynamically sizing tcp sockets and the associated scaling options based on the amount of physical memory

but then gives me an error when I try to set up the one that isn’t

^^^ see what dpkg said

I was perusing the TCP optimizer docs out of general internet suspicion (TM). “Make your conncetion faster! Just download and run this magic .exe! Also there are kittens! Click now!”

this can create sub-par conditions on old stacks which do not do window scaling by default. or on current ones which do, and are buggy.

2 UTC

from memory, XP qualifies as the former; the acid test is - can you download many many Nx flows at 4MB/s in parallel

its up-to-date testing

I’d expect the answer to be yes.

Hm

lspci -v | grep -i audio

but in general, let that tool turn on window scaling and other goodies on the hosts on your LAN

JoeBlakesley, blacklist the alsamodule for the onbaord to prevent it from getting an entry in alsas list

is there anyway I can get it tp wipe *all* my soundcard settings and search for soundcards again as it worked orginally

that’ll get you the best result - if this theory is correct.

In my previous testing it seems that I’m capped at about 10 megabytes/second aggregate flow. With many flows going simultaneously it seems that each one is limited to about 1-2 megabytes/second. That really sucks

did you aggregate across hosts just out of interest?

Letting that tool retune my client didn’t help a lot. The transfer initially happened at 8-9 megabytes/second, but quickly dropped back down to 4 megabytes/sec. No, that’s what I’ve been seeing in production.

hi, i’ve got debian running in a vmware environment.. but somehow it doesnt remember its route to the internet… ip-address and such are set correctly, the gateway is also setup in the /etc/network/interfaces, but as soon as it’s rebooted, the route to the internet is gone… any way to solve this?

where do I do that?

ok in that case we probably do have a server-side problem that needs isolation - which means a quick trip down into /proc/sys/net/ipv4

it gives me: f.5 Multimedia audio controller: Intel Corporation 82801EB/ER (ICH5/ICH5R) AC’97 Audio Controller (rev Intel Corporation SoundMax Intergrated Digital Audio 2.0 Multimedia audio controller: Creative Labs SB Audigy

4 megabytes/second to each host would actually be semi-acceptable, 10 megabytes/second would be ideal (since many are not gigabit-capable), 1-2 starts to get painful Yeah, I’m looking at some server-side tuning docs as well http://www-didc.lbl.gov/TCP-tuning/linux.html

intel is on-board obviously

what’s tcp_wmem?

well, so you didnt disalbe the onboard chip in BIOS, most likely

JoeBlakesley, look into /etc/modprobe.d/alsa-base-blacklist

well, so you didnt disalbe the onboard chip in BIOS, most likely

thanks

nb, those are exactly what is sized dynamically post about 2.6.19 or so.

tcp_wmem is “4096 16384 4194304″ This is stock Debian etch, 2.6.18

and tcp_window_scaling is ?

pretty sure I did

tcp_window_scaling is 1

tcp_adv_win_scale ?

both sound cards were working fine together before I disabled the on-board one in the bios anyway

tcp_adv_win_scale is 2

that’s when they both stopped working

JoeBlakesley, why did you disable it?

heh

hmm, all good so far

? you pinged?

sorry, nickcompletition failed

Hm, the tuning guide I just sent you the URL to claims that bigger than those is better

tab completion misfire

ok

to save me having to set programs up to use the right one

no problem

That seems to be targeted more at long paths though

JoeBlakesley, http://alsa.opensrc.org/index.php/FAQ see 3.14 - to make then one you want the default

yep; but some of the same math applies at short high clock rate paths too.

0000/ ie NO “S” in https, I get the message “Error - Bad Request, This web server is running in SSL mode. Try the URL …instead.” any toughts? thanks

–purge webmin now

ta

is there a way to change the settings of my audio card’s onboard equalizer from linux? it’s some onboard VIA crap

Would you be offended if I started playing with these values, following the directions in http://www-didc.lbl.gov/TCP-tuning/linux.html ?

Onboard audio isn’t as shoddy as everyone makes out that it is

I know it’s frustrating when you’re helping someone debug and they go off on their own little tangent

give the tcp_wmem one a shot then restart apache

#alsa will be able to help you more than we can

the software mixers don’t work ? like alsamixer or aumix ?

you serious about that? and ifso, why? (the –purge webmin)

yes i am ask the bot

the other line of inquiry I have is whether *apache* is somehow set to use relatively small windows

!tell SpeedyG -about webmin

!tell SpeedyG about huge gaping security issue

That actually sounds somewhat plausible; I’ve never gotten anything repeatable, but I’ve had a few anecdotal experiences where it seems that different apps display different behavior

ok, clear

I’ve seen apache and perforce go slow, and netcat and scp go fast, under what I believe to be similar conditions, for example “similar”

*nod*; ok; try the tcp_wmem setting, restart httpd, repeat test

No real change

Did anyone here tested this game http://packages.qa.debian.org/b/btanks.html, new in unstable? It always crashes on my system around 3s after launch. (Debian64-testing with fglrx-driver)

Again, it was going fast at the start, then slowed down to about 4 megabytes/sec. Is Firefox’s progress bar still broken enough to think that the time when it’s downloading while the dialog box is open doesn’t count, but the bytes it downloads do count? That would be sad.

I think so yes; it might be saner to use ‘doze ftp as a benchmark if you have an ftpd

What a bunch of baloney Yes, I probably should be using netcat or something

maybe strace it ? check the BTS ?

against the httpd while serving the file?

I had hoped to hand this off to the network guys, though… I thought that http transfers would display identical characteristics to netcat transfers.

hah, welcome to my day job

Either they don’t, though, or something has changed since I ran that test. I would feel fine about running strace against httpd.

http://video.google.com/videoplay?docid=-7675982915220014267&hl=en kaffeeklatsch preview

mostly what interests me is whether apache is using sendfile or write(2) to push data to the client if it’s sendfile, good, forget apache; if it’s write(2), then what matters is the size of the write, and how long it takes to complete; large write is generally better.

the quality is pretty horrible though

Hm, it looks like it’s calling sendfile() on 512k at a time

I tried gdb, but the output I got is close to nothing.

I’ll pastebin

ouch!

Well, maybe I just don’t know what I’m seeing

I think I’ve blacklisted it but doesn’t seem to help

try a strace or ltrace it’s like stripped

http://paste.debian.net/36244

Mixer attach default error: No such device” when I try to install one

JoeBlakesley, run dpkg-reconfigure linux-sound-base

ok thanks

That’s at a pretty rapid pace; the speed with which those lines show up seems pretty consistent with 512k per call.

hi, are the commands to find out the version number of an .so file ?

anyone know if there’s a repo for kde4 packages?

of a command to get general header information

that sounds like what I need

JoeBlakesley, any error?

nope

ask #debian-kde

JoeBlakesley, cat /proc/asound/cards - one or two?

Trying netcat. BTW, I ignored my own advice and wandered off on my own and did all of the sysctl tuning it recommended on http://www-didc.lbl.gov/TCP-tuning/linux.html (the four sysctl lines at the top). It didn’t help.

~madwifi

hmm… madwifi is atheros drivers at http://madwifi.sf.net

bt from gdb gives only that: http://pastebin.com/d1e6e5fd5 I am sure that is useless for a bugreport.

JoeBlakesley, or none?

no directory asound in proc

JoeBlakesley, no alsa loaded at all then - you know the name of the module for the audigy?

ok here’s something really screwey; that pcap you pasted, has the server issuing 11k+ *frames* what’s your current interface mtu on the linux end?

no; I’m sure I can find it out

1500

maybe just maybe you need to bounce that interface down/ back up? ie ifconfig down/ ifconfig up

Hm, will do

i said to try strace or ltrace

I think no alsa loaded at all situation is a result of me trying to run alsaconf rather than the bios thing

those are commands

yes, but I need to install them first,…

uhm ok

Also, my testing with netcat seems to indicate much better performance (28 megabytes/sec)… but that’s from watching the ifconfig byte counts rather than watching how much actually gets written to the disk on the client end, so it may not be 100% reliable

hi, i installed tomcat5.5, now when i try to restart it, it return: error writing to /var/lib/tomcat5.5/logs/catalina_2007-09-04.log ? someone knows why this happens ?

I’ll bounce the interface, do another http transfer, post the pcap, and go back to playing with netcat

the module name is what is shown from the first menu in alsaconf ye?

JoeBlakesley, audigy LS? seems it’s using snd_ca0106 - modprobe that

(ca0106) ok thanks for all you are help

that’s utterly bizarre, I’ve never seen anything like it

You mean that strace?

the tcpdump

JoeBlakesley, cat /proc/asound/cards - any card listed?

yep it’s listed

Hm, another funny thing: tcpdump: listening on eth0, link-type EN10MB (Ethernet), capture size 96 bytes

JoeBlakesley, ok now run invoke-rc.d alsa-utils reset 0

EN10MB certainly does not sound correct

ye

JoeBlakesley, then test alsa using aplay some wav

Uh-oh. mii-tool reports the interface being in 100baseTx-FD mode.

the card has mutiple outputs and I don’t know which is which

ethtool reports it as 1000Mb/s. How is that possible?

trying them now

that (the en10mb from tcpdump) is fine the mii-tool is also fine I think; the mii literally is the 100base transceiver trust ethtool, in this case mii-tool is misleading.

hi..what is the command for mounting an external cdr?what is its file system?

the thing that’s throwing me is watching the Tx end 10.150.150.150 throw out a *19.2k* sized frame and the Rx end then ACK it in 4380byte segments.

working now thanks a lot

Generally it can be auto-detected, but I believe it’s iso9660

thanx

the only thing I can think is that this is an artifact of the card doing segmentation offload

This is a packet capture on the sending end.

usually these are emulatied as scsi-devices. So mount /dev/sda /mnt/cd …

in fact that must be it

JoeBlakesley, problem still getting things to work at reboot

oh right ok

so the Tx host stack is throwing it whole windows and the card is segmenting it down into ether sized frames, that start to make sense.

I’ll try to reboot some time this eve to test that

but still evidently the Rx end is saying that it’s recieving 4380byte frames == jumbo.

…or maybe I shouldn’t

JoeBlakesley, ok good luck

thanks a lot at least I sort of know what to do in future

JoeBlakesley, loading (modprobe) snd_ca0106 was the essential

Okay, with tuning on both ends, still transferring at 4 megabytes/sec, after bouncing the interface: http://qix.demiurgestudios.com/~moise/slow-gigabit-1.pcap It dropped some packets this time

right

I mean tcpdump reported some packets dropped by the kernel during the capture; that didn’t happen during the first capture

linux itself is one area of linux I have no idea about

http://www.chrishowie.com/2007/09/04/kaffeeklatsch-teaser/

ok so this is after tweaking wmem?

It is

as an aside, is it normal for some apps to ignore what I’ve set as the default soundcard in sound prefs? that was kinda why I decided to try to dsiable one of the cards

re

what is the network-monitor for console called again ? I remember bandwidth, bytes so far …

JoeBlakesley, http://alsa.opensrc.org/index.php/FAQ see 3.14 - to make then one you want the default

right sorry ye you linked to that b4 thanks

JoeBlakesley, so really no need disabling the onboard

you’ve been extremely helpful

ntop ?

ok try “ifconfig eth0 txqueuelen 1000″ and also setting /proc/sys/net/core/netdev_max_backlog to 30000 or so.

im a hacker

i can hack all machine connect here

My eth0 txqueuelen seems to already be set at 1000. The netdev_max_backlog is only set to 1000 right now, though.

apart from that I’m not sure what to suggest beyond trying a different server NIC; linux is certainly pushing the traffic in large gobs down to the hardware, if it’s not turning up quickly enough at the far end it smells and quacks like maybe a too-tricky-but-broken hw problem.

!ops THEHACKER want to leave us…

Hydroxide, bob2, caphuso, dondelelcaro, doogie, eeyore-, ElectricElf, ):, helix, ljlane, LoRez, RichiH, mentor, Netsnipe, TML, walters, xk, gravity, azeem, Maulkin, stew, peterS, Alife, Myon, Ganneff, Maulkin, weasel, zobel: streuner complains about: THEHACKER want to leave us…

maybe turn off tso with ethtool or something

i will hack streuner STFU

I’m installing cygwin netcat on the client machine right now, so it’ll be a second before I can test. I at least have some things to work on (netcat vs. apache, that retune you recommended, turning off clever offloading options). Thanks very much for all your help.

shut the fuck off

I’ll let you know if I find anything else of interest.

wtf

yeah please do, just /msg

thanks LoRez

Hi!

could you actually ban him?

I Am an German User…

I’m pretty sure this is going to turn out to be related to the NIC on the Tx end; if you can back to back two of your XP boxes that’ll give you another useful datapoint

!de

from memory, de is deutschsprachige Hilfe bekommt ihr in #debian.de (auf irc.oftc.net und irc.freenode.net) - German speaking users please go to #debian.de (on irc.oftc.net and freenode)

!de ?!?

i guess de is deutschsprachige Hilfe bekommt ihr in #debian.de (auf irc.oftc.net und irc.freenode.net) - German speaking users please go to #debian.de (on irc.oftc.net and freenode)

dont ban me

Ah, good idea. I’ll check that out as well.

!beer LoRez

nb you’re still getting 3k+ ACK’s from the Rx end at least from the viewpoint of this side of that NIC

does anyone know where does evolution stores its preference files I removed .evolution but I still have all accounts and preferences in evolution the same way it was before

wat is the German Debian room?

the bot (dpkg) already told you

jean, try and look with gconf-editor

sorry for my Bat english

#debian.de

thanks

OK! Thank you!

Yeah, I understand enough now to see that. It would be interesting to run wireshark on the client end. I still don’t know enough to understand a TCP ACK packet, but like I said I have plenty to work on.

jean, (i^m not sure, just guessing)

Streuner? You are german or nothing? *not

german, lot of germans are here

if you want to speak german .. try #debian.de

anyone can tell about those authentication processes installed in public wifi places ? when you get a webpage to authenticate before getting in … anyone can tell me how it is done ?

okay ciao!

loca|host, i think automated redirection with php or something (guessing, again)

gotta run - good luck.

dante_2core, not about that redirection, it seems there’s a proxy underwear

Thanks. And thanks again for all your help.

you are proably right about ntop, but it just kills itself with “Hangup”… thanks for the pointer, i will dig into that.

loca|host, maybe a server between the network and the router which acts as block for non-authenticated users even i could do that if i took the time loca|host, i used to wonder how that is done too. but it has to be something like that. it’s simple and efficient

i like jnettop too

is it light on RAM and CPU ?

i guess

router with iptables, anything thats not recognised doesnt get through but gets an authentication page.. once authenticated, you can use the network

SpeedyG, that’s gotta be one hell of a router then, doesn’t it?

only one way to find out. thanks.

can someone explain this to me?i get this error all the time..http://pastebin.com/m56d6eb41

linux box acting as a router not much to it

SpeedyG, i see

one of my friends once set up such a system for a small hotel

for some reason during the burning process my external cd-r gets disconnected or something

waspius, disconnected in what way? does the burning stop?

yes the burning stops and i get that error..and if i look in my devices it is unmounted and lost

what’s the exact error?

i need to switch it off and on again so it can be remounted

The Best Php Web Space