IRC chat logs for #ltsp on irc.libera.chat (webchat)


Channel log from 18 May 2007   (all times are UTC)

00:26nickel2 has quit IRC
00:35freet15 has joined #ltsp
00:35privet has quit IRC
01:04jsgotangco has joined #ltsp
01:11MoL0ToV has joined #ltsp
01:25savetheWorld has quit IRC
01:55jsgotangco has quit IRC
02:13Egyptian[Home] has joined #ltsp
02:27vagrantc has joined #ltsp
02:52Burgundavia has joined #ltsp
03:02Daggett has joined #ltsp
03:02
<Daggett>
hello
03:02
I have troubles with sound and flash player 9
03:02
in firefox
03:03
already touched /tmp/.esd/socket and .esoud/socket
03:09fernando1 has quit IRC
03:14bobby_C has joined #ltsp
03:15fernando1 has joined #ltsp
03:46Q-FUNK has joined #ltsp
04:08ogra has joined #ltsp
04:19MoL0ToV has quit IRC
04:20vagrantc has quit IRC
04:28envite has joined #ltsp
04:28
<envite>
Hello all from Spain
04:28
Hola a todos desde España
04:30muh2000 has quit IRC
04:33muh2000 has joined #ltsp
04:39muh2000 has quit IRC
04:39Burgundavia has quit IRC
04:40MoL0ToV has joined #ltsp
04:57Avatara has joined #ltsp
05:03jsgotangco has joined #ltsp
05:10radoeka has joined #ltsp
05:11radoeka has left #ltsp
05:14chupa has quit IRC
05:26uniconformist has joined #ltsp
05:29jsgotangco has quit IRC
05:35nf1 has quit IRC
05:35ogra has quit IRC
05:35ogra has joined #ltsp
05:47avlis has quit IRC
05:48muh2000 has joined #ltsp
05:49
<envite>
does somebody used any kind of snmp systen with ltsp clients?
05:52bobby_C has quit IRC
05:53
<ogra>
should be trivial with ltsp5
05:53
just install an snmpd in the chroot and set up mrtg or some such to query the clients
05:54
<envite>
I'm in LTSP 4.2
05:54
stil...
05:54
still...
05:55
company politics :(
06:16a5benwillis has quit IRC
06:33Q-FUNK has quit IRC
06:42Q-FUNK has joined #ltsp
06:43MoL0ToV has quit IRC
06:48McDutchie has joined #ltsp
07:02cliebow_ has joined #ltsp
07:13a5benwillis has joined #ltsp
07:14jsgotangco has joined #ltsp
07:25jernst has joined #ltsp
07:29davidj has quit IRC
07:31mhterres has joined #ltsp
07:32Guaraldo has joined #ltsp
07:32jernst has left #ltsp
07:33plamengr has joined #ltsp
07:36
<cliebow_>
ogra: you be around for a while?
07:44meduxa has joined #ltsp
07:46
<mhterres>
morning
07:46ogra has quit IRC
07:47
<cliebow_>
ho!
07:48mhterres has left #ltsp
07:48mhterres has joined #ltsp
07:52
<envite>
I found it
07:55Blinny has joined #ltsp
07:57plamengr has left #ltsp
08:03
<cliebow_>
anyone using blade servers?
08:06J45p3r_ has joined #ltsp
08:15ogra has joined #ltsp
08:16
<Blinny>
j00?
08:16
Oops. Wrong channel.
08:19McDutchie has quit IRC
08:22sbalneav has joined #ltsp
08:22
<sbalneav>
Morning all
08:22
<ogra>
!s
08:22
<ltspbot>
ogra: "s" is Scotty!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
08:22
<sbalneav>
Hey ogra!
08:22
Was doing some fooling around last night, autoconfing the ldm-greeter.
08:23
<ogra>
oh207, cool, did you get it working ?
08:23
<sbalneav>
Should have it done today. I'll start up a branch on LP when I get it done.
08:23
Mostly, couple of gnome wrinkes I ran into.
08:23
<ogra>
take your time, i'm dragged into classmate stuff for the next days
08:24
<sbalneav>
Hey, quick question, if you know....
08:24
<ogra>
shoot
08:25
<sbalneav>
On this machine, it used to be a ide drive (hda), and I upgraded the disk to sata (sda), transfererred everything over with dd, etc. Worked fine. However, whenever I install a new kernel, it keeps setting root=/dev/hda1 on the linux command line in grub
08:25
know where the "default" is set?
08:25
<ogra>
i think in menu.lst
08:25
there should bea kopts line
08:25
*be a
08:25
<sbalneav>
There is, but if I run an update-grub, it gets RESET back to hda
08:26
so update-grub must pick it up from somewhere
08:26
<ogra>
show me your menu-lst
08:26
<sbalneav>
1 sec
08:26
<ogra>
s/\-/\./
08:26
<sep>
sbalneav, look for a line saying # kopts= (yes it should be commented)
08:27
<ogra>
shouldbe something like: # kopt=root=UUID=81fd944e-273f-4b5b-94c7-098ef0c983ad ro
08:27
<sbalneav>
Really? So it picks up the COMMENTED line?
08:27
<ogra>
likely yours has /dev/hda
08:27
yep
08:27
dont change the comments
08:27
<sbalneav>
yeah, there is a commented /dev/hda line
08:28
<Guaraldo>
good morning all...
08:28
<sep>
sbalneav, edit that, and run update-grub
08:28
<ogra>
sbalneav, sudo vol_id /dev/hda1
08:28
add the proper UUID value :)
08:30
<sbalneav>
Wow, it's, uhhh, non intuitive that a COMMENTED line would be where it pics the default from :)
08:30
<ogra>
yeah, grub is great, isnt it ?
08:30
<sbalneav>
I just blew past all the comments without reading them
08:30
thx
08:30
<ogra>
i think double ## are ignored
08:31
<sbalneav>
yeah
08:31
Man, that's pretty crackful.
08:31
<cliebow_>
sbalneav:pingo
08:31
<sep>
sbalneav, but convinient tho. if someone had actauly read the comments they'd know'n what to do right away :)
08:32
<sbalneav>
cliebow_: pongo
08:32
<cliebow_>
heh..i had a call from someone today re blade servers..must be bidding on instasll for 100 schools in Atlanta..any exp with blade servere?
08:32
<sbalneav>
sep: How's it convenient? In EVERY OTHER CONFIG FILE ON THE PLANET, you ignore comments. It runs completely counter to everything I've ever seen.
08:33
<ogra>
sbalneav, i'm sure there are ons of bugs (or duplicates) about that :)
08:33tiagovaz has joined #ltsp
08:34
<sbalneav>
typically, when I'm looking at configfiles, I do a grep -v "^#" confffile, so I can easily see what's "important" :) Ah well, just another fact that needs another neuron to store it :)
08:35
cliebow_: No, sorry, blade = expensive. Me = cheap :)
08:35
<ogra>
right, its just to keep your brain trained
08:35
<sbalneav>
lol
08:35
<cliebow_>
heh..jim kronebush must be the guy to talk to ..
08:35
<sbalneav>
I like the uuid one better, so I'll go with that.
08:35
<ogra>
sounds exciting
08:36dean_za_ has quit IRC
08:37
<sep>
sbalneav, should be fixed in grub2 btw http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=383282
08:38
<ogra>
if grub2 every gets ready :)
08:38
*ever
08:38
<sep>
:)
08:38
<sbalneav>
heh, that worked.
08:39
thx, ogra, sep
08:39
brb, I'm gonna reboot just to test.
08:40sbalneav has quit IRC
08:44sbalneav has joined #ltsp
08:44
<sbalneav>
That worked!
08:44
Perfect
08:52nicoAMG has joined #ltsp
08:52petre has joined #ltsp
08:53
<petre>
morning all
09:00jsgotangco has quit IRC
09:01dean_za_ has joined #ltsp
09:03J45p3r_ has quit IRC
09:03J45p3r_ has joined #ltsp
09:04J45p3r__ has joined #ltsp
09:07jeremydyoung has quit IRC
09:09uniconformist has quit IRC
09:11muh2000_ has joined #ltsp
09:12joebake1 has joined #ltsp
09:12jeremydyoung has joined #ltsp
09:13jeremydyoung has left #ltsp
09:15uniconformist has joined #ltsp
09:16cliebow_ has quit IRC
09:18freet15 has quit IRC
09:20uniconformist has left #ltsp
09:23J45p3r_ has quit IRC
09:24tiagovaz has quit IRC
09:24muh2000 has quit IRC
09:28irule has joined #ltsp
09:29plameng1 has joined #ltsp
09:32vagrantc has joined #ltsp
09:34cliebow_ has joined #ltsp
09:35Burgundavia has joined #ltsp
09:37ogra has quit IRC
09:38twinprism has quit IRC
09:39arie-rsa has joined #ltsp
09:40
<arie-rsa>
ok i wish to install a ltsp server an start with 1 thin client
09:40
iam thinking about setting the server box up with debian 3.1 with the ltsp server stuff on it, that would be the first step correct?
09:41
<vagrantc>
arie-rsa: if you're installing a new server, i'd strongly recomemnd using the current stable release, debian etch/4.0
09:42
arie-rsa: there's not much there yet, but see http://wiki.debian.org/LTSP
09:42
<arie-rsa>
ok i will install debian sarge 3.1 then dist up..
09:42
<vagrantc>
why install sarge?
09:42
<arie-rsa>
i have a burned iso ^^ and i hate burning
09:43
<vagrantc>
if you must ...
09:43plameng1 has quit IRC
09:43
<arie-rsa>
i also have loads of ubuntu CDs if that would b recommended?
09:43
however the server doesnt have to run a xserver right?
09:43
<vagrantc>
depends on if you prefer debian or ubuntu
09:43
<arie-rsa>
ubuntu and debian are the same to me really ;P
09:44
<vagrantc>
ubuntu 7.04 feisty is probably a little more feature complete LTSP than debian etch at this point.
09:47
you need to backport the local device support for it to work in debian etch
09:47
local sound support might be a little more clean with ubuntu as well
09:47ogra has joined #ltsp
09:48
<sutula>
vagrantc: do you have any details on sound support problems with Etch...I still haven't gotten it working
09:48
<arie-rsa>
vagrantc : witch os do you suggest than?
09:48
<vagrantc>
sutula: what have you done?
09:49twinprism has joined #ltsp
09:49
<vagrantc>
arie-rsa: i told you everything i know.
09:49
<arie-rsa>
k
09:49
<vagrantc>
i've mostly worked on the debian side of things, but work pretty closely with the ubuntu folks
09:49
<sutula>
vagrantc: The client kernel has /proc/asound entries, but udev doesn't create e.g. /dev/dsp, so esd can't open the device
09:50
<vagrantc>
sutula: do you have the alsa to oss compatibility module loaded? i forget the exact name ...
09:50
<sutula>
vagrantc: Probably not...this is the sort of stuff I don't know about
09:50
iirc, just ltsp-server and it's depends
09:51
<vagrantc>
sutula: well, that needs to happen on the thin client, not on the server
09:51
<sutula>
vagrantc: ...so installed in the chroot?
09:51
<vagrantc>
sutula: so, you've got it booting, and you can log in, but sound doesn't work?
09:51
<sutula>
vagrantc: Yes
09:52
<vagrantc>
sutula: you might need snd-mixer-oss and snd-pcm-oss loaded ...
09:52
sutula: so put them in /opt/ltsp/i386/etc/modules
09:53
<sutula>
vagrantc: Is the longer-term plan to make this more automatic (in terms of dependencies)?
09:53
<vagrantc>
sutula: yes
09:54
the long term plan is "everything just works" :)
09:54
or, to steal from k12ltsp, "it just works, duh."
09:54
<sutula>
vagrantc: If I get it working, should I just file bug reports for what I find is needed?
09:54
<vagrantc>
sutula: please do
09:54
<sutula>
vagrantc: k, thanks
09:54
<vagrantc>
sutula: and use reportbug, please :)
09:54
<ogra>
vagrantc, well, i assume you will switch to pulse as well
09:54
<sutula>
vagrantc: yep
09:55
<ogra>
/dev/dsp is moot then
09:55
<vagrantc>
ogra: sounds like the way to go, yes.
09:55
<sutula>
Is that a viable option today on Etch?
09:55
<vagrantc>
sutula: you'd have to hack a couple scripts
09:56
<ogra>
probably if you copy parts from ubuntu into your chroot ...
09:56
<sutula>
ogra: Is that what ubuntu uses today?
09:56
<ogra>
the scripts alone wont do it though you need a special config
09:56
sutula, feisty, yes
09:56mfdutra has joined #ltsp
09:56
<ogra>
the former versions also used esd
09:56
<vagrantc>
do you need to tweak ldm at all?
09:57
<ogra>
no but ltsp-client.init
09:57
ah, well, and one or two lines in ldm iirc
09:57* vagrantc hopes to merge debian and ubuntu's ltsp over the next few weeks
09:57
<ogra>
telling it about the default soundserver
09:57
yeah
09:57
<vagrantc>
and then hopefully i'll be able to release a backport to etch, too.
09:57
<sutula>
vagrantc: That would be sweet
09:57
<ogra>
well, you are 6 months behind atm ... the merge will be hard
09:57
<vagrantc>
and get a backport for ltspfs
09:58
ogra: has to happen :P
09:58
<moquist>
vagrantc: found someplace to hole up? :)
09:58
<ogra>
indeed :)
09:58
<vagrantc>
moquist: sort of... in a hostal in madrid
09:59
<arie-rsa>
ok iam burning 7.04 ubuntu now
09:59
fkn disc errors though :P
10:00
and i have a router, can i use that as tyhe dhcp server?
10:00
<vagrantc>
i'd recommend avoiding it if you can
10:00
<arie-rsa>
ok i will see what we can do :)
10:00
<vagrantc>
much easier to configure a real dhcp server under linux
10:01
<arie-rsa>
were just going to screw some time with this :)
10:01
<vagrantc>
the ideal is to have two network cards on your LTSP server, one for network access, and one for the thin-clients
10:01
<arie-rsa>
we have loads of old machines so we figured might aswell thy ltsp
10:01
we have extra network cards ^^
10:01
however no clue about drivers so i hope ubuntu has the correct drivers
10:01
<cliebow_>
arie-rsa, will they pxeboot..or are you going etherboot
10:02
<arie-rsa>
ethernet boot
10:03
well i hope they can pxe
10:03
<vagrantc>
well, they both do ethernet booting ...
10:03
<arie-rsa>
pxe seems to the way to go
10:03
<vagrantc>
pxe and etherboot are two different ways of booting over ethernet
10:04
<arie-rsa>
but atm iam first going to setup the server w. ubuntu :)
10:04
<vagrantc>
yeah, if your card supports PXE properly, that's definitely easier.
10:04
<arie-rsa>
i read about pxe bootp protocol :)
10:04
vagrantc u dont happen to live in haarlem do u?
10:04
<vagrantc>
nope
10:04
<arie-rsa>
damn :(
10:04
^^
10:05
<vagrantc>
bumming around spain for the next few weeks
10:05
<arie-rsa>
hablo un poco de espanol, pero nu me gusta hablo espanol
10:05
<vagrantc>
por que no?
10:05
<arie-rsa>
prefero hablo en englaisa
10:06
tengo espanol por mi collega
10:06
i have spanish at college and i hate it, es muy dificil
10:06
<sbalneav>
Argh. %&$! autoconf.
10:06
<mfdutra>
guess german is more difficult :)
10:06
<sbalneav>
A pox on thee.
10:06
<arie-rsa>
Me gustan las chicas muy guapas :P
10:06
nein, ich spreche auch deutsch
10:07
<mfdutra>
heh
10:07
<arie-rsa>
deutsch is der tale des reich
10:07
^^
10:07
<sbalneav>
ogra: BTW, Feisty upgrade here at work worked wunnerful.
10:07
<arie-rsa>
maar nederlands is het best, but dutch is the best
10:08
<ogra>
sbalneav, great
10:08
mvo will love to hear that
10:08
<sbalneav>
That fixed the bzr problem I was having too.
10:08
Now I just gotta move my edubuntu box at home to gutsy.
10:08
<mfdutra>
hey guys. I'm playing here with local devices. it's working ok when I insert a cd into the drive. but no way to open the drive tray any more
10:08
<arie-rsa>
mfdutra why? its a killer drive?\
10:09
<mfdutra>
probably
10:09petre has quit IRC
10:09
<ogra>
mfdutra, well, in ubuntu the nolock option is set, so just pressing the button should give it back
10:09
donno about other OSes
10:09petre has joined #ltsp
10:09
<mfdutra>
I mean that's happening in my thin client
10:09
so it's running ltsp 4.2
10:09
<arie-rsa>
mfdutra: it starts spinning at boot and doesnt stop spinning till poweroff? i have that at home, turning off the pc doesnt matter :D keeps spinning
10:09
<sbalneav>
mfdutra: Hmm, have you waited 2 seconds?
10:09
<arie-rsa>
with or without disc in it
10:09
<sbalneav>
Should work.
10:10
<mfdutra>
sbalneav, I've waited for 10 minutes
10:10
<arie-rsa>
^^
10:10
<cliebow_>
anyone else pxe booting from winblows?
10:10
<sbalneav>
Something's not right then.
10:10
mfdutra: 4.2 you say?
10:10
<mfdutra>
yep
10:10
nothing happens when I press the drive button
10:10
it doesn't even blink
10:12
<petre>
cliebow_: what do you mean? I thought MS required windows to be installed on the local disk
10:12
<cliebow_>
naahh using winblows dhcp
10:12
<petre>
ah, windows server, you mean
10:12
<cliebow_>
yeccch
10:12
<petre>
agreed
10:14
<sbalneav>
mfdutra: Maybe bomb through the checklist, see how that works
10:15
<mfdutra>
what happens when I press the drive button? does it send an event that was supposed to umount the cd?
10:16
<vagrantc>
i don't think CDs tend to communicate much
10:16
we have to do ugly things to detect device insertion events
10:17
<cliebow_>
works fine on some machines..not worht a doggoon on others
10:18
<mfdutra>
that's really bad for a user
10:18
<vagrantc>
tell that the the hardware manufacturers
10:23
<sbalneav>
mfdutra: Theoretically, after 2 seconds, the cd's supposed to be unmounted, so pressing the button should work fine.
10:23
<mfdutra>
how could I debug that?
10:24
is there a file where I can check whether or not the drive is sending an event?
10:25
<sbalneav>
Take a look on the thin client itself, see if the cd's being unmounted.
10:25
<vagrantc>
are there any open files that could be keeping it mounted?
10:25
<mfdutra>
sbalneav, it's still mounted
10:25
<sbalneav>
If there's any open files, that's handled at the ltspfs level, not the cdrom level.
10:26
<mfdutra>
there is no lsof in ltsp
10:26
there should be :)
10:26
<vagrantc>
mfdutra: is it still showing as mounted on the serveR?
10:26
because you could use lsof on the server-side
10:26
<mfdutra>
it's not mounted
10:27
I've got ~/Drives/My_CDrom, but it's empty
10:27
but in ltsp it's still mounted
10:27
in /tmp/drives/My_CDrom
10:28
if I umount it, the tray opens ok
10:28
but the user isn't gonna do that
10:29vagrantc has quit IRC
10:29
<sbalneav>
mfdutra: is lbussd running?
10:30
<mfdutra>
in the server or the client?
10:30
<sbalneav>
client
10:30petre has quit IRC
10:30efra has joined #ltsp
10:30
<mfdutra>
I see /sbin/lbuscd
10:31
<sbalneav>
ah, ok
10:31
sorry, cd
10:31
hmm
10:31
cant remember where they are for 4.2 , but look for ltspfs_mount and ltspfs_umount
10:31
they're what should be mounting and umounting the cd
10:31
<mfdutra>
hmmm
10:32
I tested again and it worked
10:32
maybe it was open files
10:32
locked descriptors
10:32
<sbalneav>
Gremlins
10:34
<mfdutra>
now it's happening something more weird
10:34
I put a cd into the drive, the icon pops up in my screen (server), and it's empty
10:34
I switch to vt2 (client) and the cd isn't mounted
10:37
<sbalneav>
mfdutra: What kind of server is it>
10:37
?
10:37
ogra: Seen the announcement about logfs?
10:38
ogra: http://logfs.org/logfs/
10:38
<mfdutra>
ubuntu dapper
10:38
<sbalneav>
32 or 64 bit?
10:38
<mfdutra>
32
10:39
<sbalneav>
run through the checklist. Step 11 should be informative.
10:39
<ogra>
"LogFS runs on the OLPC. It will die a horrible death upon encountering the first bad block"
10:39
lol
10:39
yay for stability
10:39
<sbalneav>
yeah, it's in development
10:39
<ogra>
but smells good
10:39
<sbalneav>
but worth keeping an eye on
10:40
<ogra>
yep
10:42
<sbalneav>
10 years from now, when we're all sitting here with 250 gig flash disks in our boxes :)
10:42
<Burgundavia>
more like 2 or 3
10:53
<joebake1>
ogra: by bad block are you referring to it's storage media? A project called EVMS
10:53
Enterprise Volume Management Service was built to provide storage management
10:53
within a gui environment. They have a bad block relocation feature.
10:53
I ran into problems with bad blocks on my email server. I've never quite quite
10:53
learned enough about EVMS to put it to good use, but maybe their code
10:53
could help in this place.
10:54
Sorry I must have another IRC session opened. I'll stay logged in for a while.
10:54
My normal user name is joebaker.
11:00Burgundavia has quit IRC
11:06petre has joined #ltsp
11:06Q-FUN1 has joined #ltsp
11:07
<petre>
if I attach a scanner to the server, should the clients be able to see it, say, via xsane?
11:07
<yanu>
yes
11:07
<petre>
using ltsp 4.2 here
11:08
do I need to enable it via lts.conf? I thought that was just if the scanner was attached to the client
11:08
if I login as a muggle on the server, xsane sees the scanner
11:08
but if I run it on a client, it says it can't find any scanners
11:13Q-FUNK has quit IRC
11:18
<petre>
ok, if I su to root on a client and then run xsane, it finds the scanner
11:18
so it sounds like a permissions issue
11:19
anyone know what file I should be looking for to allow all users to access the scanner?
11:19
i.e., which file needs its permissions changed?
11:20
<sutula>
petre: perhaps some users need to be added to a group in /etc/group
11:20* sutula isn't a scanning expert
11:20* sutula usually just looks for likely groups and then finds it by trial and error
11:21
<petre>
in /dev I have a device called scanner-usbdev1.3 which is a symlink to /bus/usb/001/003
11:21
perms on the symlink are 777
11:21
<sutula>
petre: perhaps group plugdev
11:22* sutula also sees a messagebus, scanner, saned in his /etc/group file
11:23
<petre>
sutula: I don't have any of those in /etc/group
11:23
I'm on fedora 5 (k12ltsp 5)
11:23Gadi has joined #ltsp
11:24* sutula doesn't know how RH administers such permissions, but it probably has a way...maybe ask in a Red Hat support channel?
11:25
<petre>
I tried changing the perms to 744 on /dev/bus/usb/001/003, but it still can't find it
11:25
Gadi: !!!!!!!!!!!!!!!!!!!
11:25
<Gadi>
petre!
11:26
<petre>
know anything about scanners?
11:26
<Gadi>
Ive scanned
11:26
I admit to it
11:26
I was young and I needed the money
11:26
<petre>
scanner is attached to server, but clients can't see it when they run xsane
11:26
only root can find it from a client
11:26
but a muggle logged in on the server can find it
11:27
<yanu>
then it is defenetly a permission problem
11:27
<Gadi>
so anyone logged into the server can see it
11:27
<petre>
yes
11:27
<Gadi>
but if they are on a client they cannot
11:27
<petre>
yes
11:27
but root on a client can see it
11:27Q-FUN1 has quit IRC
11:28
<Gadi>
thats a fun one
11:28
:)
11:28
have you tried using any of the cmdline tools
11:28
<petre>
yeah, esp. since I was hoping to get this working without needing root access
11:28
<Gadi>
they may provide more insight
11:28
<petre>
like what?
11:28
<Gadi>
is it a usb scanner?
11:28
<petre>
yes
11:29
epson 1250
11:30
<Gadi>
do you have the scanimage command?
11:30
<petre>
yes
11:31
<Gadi>
log in as a user on a client, open a terminal, type scanimage -L
11:31
iirc
11:31
<petre>
same result with scanimage: root can see it from client, muggle cannot
11:31
<Gadi>
does it give you a reason
11:32
<petre>
no, just says no scanners were identified
11:32
for root, it does show the scanner, correct model, etc.
11:32Daggett has quit IRC
11:32
<petre>
I wonder if I can point scanimage at the device
11:33
<Gadi>
did you check perms on the device?
11:33
who owns it?
11:33
<petre>
well, I think so
11:33
<Gadi>
what is the device name?
11:33
<petre>
I found /dev/scanner-usbdev1.3 which is a symlink to /dev/usb/001/003
11:34
perms on the sym link are 777
11:34
<Gadi>
forget the symlink
11:34
<petre>
perms on /dev/usb/001/003 were 700
11:34
<Gadi>
owner?
11:34
<petre>
I changed them to 744, to no avail
11:34
root:root
11:34
<Gadi>
try changing to 777
11:34
(really should be 666
11:34
but, hey
11:35
funny that the users should see it on the server, tho
11:35
<petre>
ls shows it to be character device
11:36
sorry, original perms were 600, now are 644
11:36
I'll try 666
11:36
hey, that worked!
11:36
scanimage sees it now
11:36
<Gadi>
weird
11:36
you should change the udev rules
11:36
<petre>
so it has to be able to write to it
11:37
<Gadi>
so its a permannt fix
11:37
<petre>
where do I do that
11:37
<Gadi>
or
11:37
well, you said its root:root right?
11:37
<petre>
yes
11:37
<Gadi>
yeah, so change the udev rules
11:37
all udev rules should be in: /etc/udev/rules.d/
11:37
best to add your own file there
11:38
instead of mod'ing an existing one
11:38
so it doesnt get blown away with an upgrade
11:38
<petre>
well, there is a 60-libsane.rules file there
11:38
<Gadi>
ok, give it a read
11:39
<petre>
do udev rules work similarly to /etc/rc?.d files?
11:39
with the numbering I mean
11:39
<Gadi>
you can glean the magic words to put into your own: 65-petre.rules
11:39muh2000__ has joined #ltsp
11:39
<Gadi>
yes similar
11:40
in redhat, you may need to reload udev for your changes to take effect
11:40* Gadi has not worked on a redhat system in some time for this sort of thing
11:45
<mistik1>
petre: Yes indeed it does, The lower the number the earlier it is processed
11:45
<petre>
so I want my script processed after the 60-libsane.rules, right, thus Gadi's suggestion to name it 65?
11:45
<mistik1>
yes
11:46* Gadi just ate 5 peppridge farm cookies - that's not going to bode well...
11:46* petre is having a hard time grokking the correct syntax for udev rules
11:47
<mistik1>
Just when I got a handle on devfs they went and dropped udev in my lap ;-)
11:47
<petre>
pepperidge farm cookies are good
11:47
<mistik1>
dsd has a neat tutorial
11:47
<petre>
but maybe not so good for you?
11:47
mistik1: link?
11:48
<Gadi>
petre: does your libsane rules file have a bunch of lines with "MODE=600" in them?
11:48
<petre>
no
11:48
has a tone of 'definitions' for different scanners
11:48
<mistik1>
http://webpages.charter.net/decibelshelp/LinuxHelp_UDEVPrimer.html
11:48
<petre>
and prior to that, just two significant lines:
11:48
<Gadi>
ok, and is there a MODE="600" on those lines?
11:49
<petre>
no
11:49
SYSFS{idVendor}=="03f0", SYSFS{idProduct}=="0102", SYMLINK+="scanner-%k"
11:49
# Hewlett-Packard ScanJet 4200C | Hewlett-Packard ScanJet 4200Cxi | Hewlett-Packard ScanJet 4200Cse
11:49
<mistik1>
What are you trying to do
11:49
<Gadi>
hmm... does redhat still use: /etc/udev/permissions.d/ ?
11:49
<petre>
no
11:49
ls /etc/udev
11:49
makedev.d rules.d udev.conf
11:50
<mistik1>
petre: what are you trying to do?
11:50
<Gadi>
good - do this:
11:50
copy the libsane rules file to 65-petre.rules
11:50
<petre>
mistik1: scanner is attached to server, but muggles logged in on clients can't see it
11:50
<Gadi>
and at the end of each entry for each scanner, add: , MODE="666"
11:50
<petre>
root can see it from client, and a muggle logged into server can see it
11:50
<Gadi>
that oughtta do it
11:51
<mistik1>
http://www.reactivated.net/writing_udev_rules.html
11:51
gadi is correct there
11:51
<petre>
Gadi: or in this case, add MODE=666 to the scanner model in question
11:51
<Gadi>
might as well do 'em all
11:51
you never know
11:51
<petre>
true, there's a slightly different model in the next classroom
11:52
<mistik1>
,MODE="0666" would be more acurate
11:52
bah
11:53muh2000_ has quit IRC
11:53
<mistik1>
, MODE="0666" would be more acurate
11:53
Or better yet and more secure use groups
11:53
<petre>
owner on the device is root:root
11:53
<mistik1>
then you would add something like , GROUP="scanner"
11:54
<petre>
ah, I see; yes, that sounds better
11:54
<mistik1>
, GROUP="scanner", MODE="0600"
11:54
that should work nicely
11:54
<petre>
shouldn't that be 660?
11:54
<mistik1>
nice catch ;-)
11:55
<petre>
;-)
11:55
<mistik1>
You are now a udev pro ;)
11:55
<petre>
heh
11:56
okay, something didn't go right
11:56
I unplugged the scanner, then plugged it back in
11:57
<mistik1>
and?
11:57
<petre>
now I see a /dev/bus/usb/001/004; last time it was /dev/bus/usb/001/003
11:57arie-rsa has quit IRC
11:57
<petre>
and the perms are 600 again for 004
11:58
<mistik1>
then I would make a new file with the support of your scanners
11:58
<petre>
I added , MODE="666" to the only the line for this particular model of Epson scanner
11:58
<mistik1>
hey
11:59
<petre>
maybe I should try Gadi's suggestion to add it to all lines, but I don't think that's the problem
11:59
<mistik1>
the bus should make no different
11:59
did you restart udevd?
11:59
<Gadi>
petre: you need to restart udev
11:59
<petre>
ah, okay, didn't do that yet.
12:00
<mistik1>
I would also add another arg
12:00
<petre>
can I just kill -1 the pid?
12:00
mistik1: to the rules file?
12:00
<mistik1>
something like , SYMLINK+="epson"
12:00
then you dont need to worry about /dev/bus/usb/blah/blah
12:00muh2000_ has joined #ltsp
12:01
<mistik1>
it should just show up after as /dev/epson
12:01
No matter what bus it's pluggen into
12:01
That is the real power of udev
12:03
<dan__t>
We just took over another phone room last week.
12:03
I think this will be fun.
12:03twinprism has quit IRC
12:04
<petre>
hmm, added SYMLINK+="epson" to my 65-petre.rules file, did a 'kill -1' on udev's pid, unplugged the scanner, plugged it back in, but I don't see a /dev/epson
12:04
and perms for /dev/bus/usb/001/006 are 600
12:04
grrr...
12:05
<mistik1>
petre: let me see exactly what's in that file
12:06
<Gadi>
petre, did you run udevd?
12:06
<ltsppbot>
"petre" pasted "[root@cat rules.d]# cat 65-pet" (1066 lines) at http://pastebot.ltsp.org/139
12:07
<Gadi>
on ubuntu, it is a service, but on redhat iirc, it is not
12:07joebaker has quit IRC
12:07
<Gadi>
you should start it from the command line
12:07joebake1 has quit IRC
12:07twinprism has joined #ltsp
12:07joebaker has joined #ltsp
12:07envite has quit IRC
12:07
<petre>
I'm not sure where udev gets started, I don't see any mention of it in /etc/init.d
12:07
aside from ltsp-related
12:08
<mistik1>
damn
12:08
A few models in there, eh?
12:09
<petre>
yeah, two or theree
12:09
*three
12:09
<Gadi>
petre - just run it
12:10
su -
12:10
udevd &
12:10
<petre>
well, there's already a udevd -d running, but what the hey, I'll give it a shot
12:10
<mistik1>
um
12:10
<Gadi>
(in a different terminal window, if you like)
12:10
<mistik1>
wait
12:10
<petre>
mistik1: too late! ;-)
12:10
<mistik1>
which model are you working with
12:10muh2000__ has quit IRC
12:11
<petre>
I'm on a test box, so it's okay
12:11
<Gadi>
heh, too many cooks - Ill let mistik1 drive;)
12:11
<petre>
well, running udevd just dropped me back to a prompt, and ps shows only one udevd running still, same pid
12:12
<mistik1>
No wonder its not working
12:12parsley has quit IRC
12:12
<mistik1>
petre: You put the code on the wrong line
12:12
Note that the line you put it on is commented
12:12
<petre>
that's certainly possible
12:13
ah; well, they're all commented, aren't they?
12:13
<mistik1>
nope
12:13
You need to put it on the line that starts with SYSFS
12:13
I notice they have scanner-%k there
12:13
<petre>
doh!
12:14
<mistik1>
looking at this file though I would make my changes in another file
12:14
So as not to have them hosed on an update
12:14
<petre>
mistik1: what is the significance of scanner-%k?
12:14
<mistik1>
It should symlink the device to /dev/scanner
12:15
%k represents the kernel number
12:15
<petre>
I copied 60-libsane.rules to 65-petre.rules and have been modding 65-petre.rules
12:15
<mistik1>
use a lower number
12:15
or it will be ignored as udev will hit the default file first
12:16
as a rule I usually put any custom rules into something like 10-somename
12:16
<petre>
I was wondering about that, that's why I asked if it is controlled by first match or last match
12:17
<mistik1>
You guessed right
12:17
<petre>
I thought that's why Gadi suggested 65, to be later than the 60 rules
12:17Egyptian[Home] has quit IRC
12:17
<petre>
but now you're saying it should be lower?
12:17
<mistik1>
but that depends on what you're trying to do
12:17
<cliebow_>
cd /home/nwea
12:17parsley has joined #ltsp
12:18
<mistik1>
In this case to have your code work before the redhat defaults take hold I would definately make it lower
12:19
<petre>
ah, progress: I plugged the scanner in, and now I have a /dev/epson
12:19
<mistik1>
SYSFS{idVendor}=="04b8", SYSFS{idProduct}=="010f", MODE="666", SYMLINK+="epson"
12:19
indeed
12:20jammcq has joined #ltsp
12:20
<mistik1>
Hey Jimbo
12:20
<petre>
however, it's a symlink to /dev/bus/usb/001/008 and the perms on that are still 600
12:20
:-(
12:21
<mistik1>
does your line looks like what I just pasted?
12:22
<petre>
mistik1: not quite: I still have the SYMLINK+="scanner-%k" in there
12:22
should I remove it?
12:22
<mistik1>
I'm confused
12:22
<petre>
SYSFS{idVendor}=="04b8", SYSFS{idProduct}=="010f", SYMLINK+="scanner-%k", MODE="666", SYMLINK+="epson"
12:22
<mistik1>
then where did /dev/epson come from
12:22
<petre>
I've got two symlinks in there
12:22
<mistik1>
only one symlink at a time
12:23
<petre>
and they both seem to be created
12:23
<mistik1>
make it look exactly as it does in the line I pasted
12:23
hehe
12:23
funny
12:23
and use 0666
12:24
iirc you need the first octect
12:26
<petre>
nope, /dev/bus/usb/001/009 is still 600
12:26
btw, why does the number keep incrementing each time I plug the scanner back in?
12:27
and I did make it 0666 in 65-petre.rules
12:27
<Guaraldo>
sbalneav: I have a little problem with pen-drives on ltsp4.2...
12:27Egyptian[Home] has joined #ltsp
12:27
<Guaraldo>
sbalneav: I didn't find lbussd on my ltsp tree
12:28
<petre>
hmm, something else didn't go right, as there is still a /dev/scanner-usbdev1.9 being created
12:28
<mistik1>
petre: Part of the point to udev is that you dont need to care about that incrementing numbers
12:28
<petre>
I took that part out of the rule
12:28
<jammcq>
mistik1: hey
12:28
<mistik1>
petre: odd
12:28
<ogra>
jammcq, hey
12:29
<jammcq>
ogra: howdie
12:29
<ogra>
jammcq, i think i gained us several tens of seconds :)
12:29
<petre>
the 60-libsane.rules would still match, and it has the kernel symlink command, so maybe that's where that's coming from
12:29
<cliebow_>
jammcq:!!
12:29
<jammcq>
ogra: oh?
12:29
cliebow_: howdie
12:30
<cliebow_>
jammcq:how familiar are you with blade servers?
12:30
<ogra>
jammcq, i fiddled around with the udev initscript ... we now dont stop for it anymore ... but i'll need testers for that change it coud be dangerous
12:30
<mistik1>
petre: maybe we could add a lastrule to it but lets see what udevtest shows
12:30sepski has joined #ltsp
12:30
<mistik1>
petre: udevtest /dev/bus/usb
12:30
pastebot that for me
12:30
<jammcq>
cliebow_: not at all. My feeling about blad servers is that when I'm ready to add more blades, the chassis will be obsolete
12:30
<ogra>
jammcq, so far i had no client having probs here though
12:31
<jammcq>
ogra: wowwie
12:31
that's awesome news
12:31
<ogra>
sadly it doesnt help much on the e2300
12:31
<cliebow_>
..i got a call from someone from Atlanta schools..asking about deploying for 100 schools
12:32
<petre>
mistik1: yeah, it was the 60-libsane.rules file that was creating the /dev/scanner-usbdev1.9 file
12:32
<jammcq>
cliebow_: isn't there already a pilot project taking place down there?
12:32
<cliebow_>
i asked him to come in and talk with youse
12:32
i think 5 schools..
12:32
<ogra>
the small CPU is then constantly busy so the gain through not stopping is traded for an overall slowness while udev probes
12:32
it gains some seconds though
12:33
<petre>
mistik1: ah! udevtest says "parse_file: line too long, rule skipped /etc/udev/rules.d/65-petre.rules, line 773"
12:33
<Gadi>
ah, ogra, just the man I wanted to see
12:33
<jammcq>
ogra: can we remove some of the udev rules, for things like IDE ?
12:33
<ogra>
on the classmate and the 1220 its 15-20 secs faster :)
12:33
<Gadi>
ogra: been testing ur gtk/c greeter against the pygtk one
12:33
<jammcq>
wow
12:33
<petre>
what's the command in vi to turn on line numbers?
12:34
<ogra>
jammcq, nope i wouldnt do that ... else IDE devices wont work
12:34
<jammcq>
:set nonumber
12:34
<ogra>
:set nu
12:34
<petre>
tx
12:34
<jammcq>
ogra: umm, we're talking about thin clients.
12:34
oh, you wanted to turn ON line numbers :)
12:34
<ogra>
jammcq, i know and most of them can easily cope ...
12:34
<jammcq>
sorry
12:34
<mistik1>
petre: add this at the end of the line in your 60- file OPTIONS+="last_rule"
12:34
<jammcq>
ogra: it might be interesting to see what impact it has
12:34
<ogra>
jammcq, the 2300 is a rare thing i'd like to have a specia mode for or something
12:35
<mistik1>
That will prevent any more processing of that device after your file has fineshed with it
12:35
<Gadi>
ogra: I think there is not much speed gain to going with the c/gtk greeter
12:35
<petre>
mistik1: can I mod the 65- file instead?
12:35
I'd prefer to use a new file than modify an existing if possible
12:35
<jammcq>
Gadi: are you on feisty?
12:35
<Gadi>
I think its just speedier now because it doesn't grab session/lang info from server
12:36
jammcq: for my testing? yes
12:36
<mistik1>
petre: use the new file
12:36
<ogra>
jammcq, how about incorporating the 4.2 scripts for a "low-end" mode the e2300 isnt usable for me on either version btw (with a 1200x1024 display)
12:36
<jammcq>
ok, there was a pretty good speedup when some fonts were removed. but feisty already does that
12:36privet has joined #ltsp
12:36
<jammcq>
ogra: hmm, special startup script ?
12:36
<ogra>
if i force it down to 800x600 it stops showing firefox window trails on the desktop
12:36
<mistik1>
petre: rpm -Uvh udev would blow away your changes to 65-....
12:36
<Gadi>
jammcq: the issue with the ldm is that it needs to load so many libs, i think
12:37
<jammcq>
ogra: I had no problem at all with 1024x768
12:37
<ogra>
but above that it isnt usable either on ssh or XDMCP
12:37
<Gadi>
if it could be statically compiled, it might improve
12:37
<ogra>
Gadi, thats the plan
12:37
<Gadi>
ah
12:37
<ogra>
and currently it only has an #include gtk.h at the top ;)
12:37
<petre>
mistik1: why would it modify my 65- file? The system created the 60- file, the 65- is my creation
12:38
<ogra>
and i didnt plan to add much more :)
12:38
<Gadi>
ogra: what is the boot time on a good thin client (by your testing - and not vmware)
12:38
<ogra>
about 90secs on the 1220
12:38
<petre>
do you mean that -Uvh would blow away any changes I made to the system-created 60- file (which makes sense)?
12:38
<mistik1>
petre: oh, You had me confused there
12:38
<Gadi>
ogra: and how much of that is ldm?
12:39
I have been counting ~50 secs
12:39
<ogra>
the udev thingie should get that down to 70, the new greeter should cut down 10 secs
12:39
<petre>
hey, it worked this time!
12:39
<ogra>
so we're at about a minute
12:39
<mistik1>
petre: rename your created file to a LOWER number than the distro supplied file
12:39
<petre>
perms on /dev/bus/usb/001/011 are 666!
12:40
<ogra>
if we get all the new xorg 7.3 features in we can get rid of the xorg detection crap ...
12:40
<jammcq>
petre: I guess '/dev' stands for 'devil', eh?
12:40
<ogra>
that should gain 20secs
12:40
<mistik1>
jammcq: so not funny
12:40
<petre>
jammcq: heh, as in Cruella
12:40* jammcq goes back and hides under his rock
12:40
<ogra>
Gadi, so we *could* end up with 40-50 second boots
12:40
<Gadi>
ogra: on the good client
12:41
right?
12:41
<petre>
mistik1: so, does that mean that something else higher/after my 65- file is getting involved, setting perms?
12:41
<ogra>
is 500Mhz 128M good ?
12:41
<Gadi>
i would say
12:41
<ogra>
i'd define that as standard ....
12:41
<Gadi>
standard
12:41
zactly
12:41
<mistik1>
petre: It certainly can, which is why I said your file should process before and not after
12:41
<ogra>
good is rather 900Mhz+ and 128M+
12:41
<Gadi>
how much is ldm atm?
12:42
<ogra>
too much
12:42
<Gadi>
by your count
12:42
<ogra>
i have no numbers
12:42
15secs probably
12:42
probably eve 20 on first boot
12:42
<Gadi>
I am finding about double that
12:43
counting 'mississippis'
12:43
;)
12:43
<ogra>
heh
12:43
<Gadi>
btw, for us, its not how much ram at this point
12:43
its the ram speed
12:44
remember, we *are* a thin client
12:44
<ogra>
http://people.ubuntu.com/~ogra/ldm-new-greeter.tgz
12:44
try that
12:44
<Gadi>
is that newer than launchpad as of yesterday?
12:44
<ogra>
its a binary you can just unpack in the chroot
12:44
unstripped and not static though
12:44letalx has joined #ltsp
12:45
<letalx>
help not a valid image unable to load file
12:45
<ogra>
install gtk2-engines-ubuntulooks in the chroot as well if you want themeing
12:45
<letalx>
segment 00092200,00092400
12:45
ltsp install in slackware
12:45
<ogra>
its not much smaller than the python one atm, but a lot faster
12:46
<Gadi>
is it python or C?
12:46
<ogra>
C
12:47
Gadi, btw, session and lang info are only requested if the popup for them is drawn, not before
12:48vagrantc has joined #ltsp
12:48
<ogra>
it wont affect bootspeed
12:48
<jammcq>
vagrantc: hola
12:48
<vagrantc>
jammcq: buenas tardes
12:49
<ogra>
Gadi, for now the udev change will be the biggest speedup we ever had btw
12:50
<vagrantc>
ditching stuff in udev ?
12:50
in the initramfs?
12:50J45p3r__ has quit IRC
12:53
<ogra>
vagrantc, dpkg-diverting udevsettle to /bin/false
12:53
so the initscript wont wait for udev to finish before it goes on boting
12:53
*booting
12:53
<Gadi>
55 secs from black screen with X to login screen
12:53
with that new greeter
12:54
so, it comes in about the same for me
12:54
<ogra>
what kind of hw is that ?
12:54
<Gadi>
its a slower one - Geode GX2 ~400MHz, 128MB ram
12:54
<ogra>
well, 400Mhz sounds reasonable
12:54
<vagrantc>
ogra: how much does the udevsettle hack give you?
12:55
<ogra>
vagrantc, the complete time of udev stopping
12:55
<Gadi>
i was gonna ask about udevsettle
12:55
I was just looking at the initramfs
12:55letalx has left #ltsp
12:55
<Gadi>
is that what you comment out?
12:55
<ogra>
# This next bit can take a while
12:55
if type usplash_write >/dev/null 2>&1; then
12:55
usplash_write "TIMEOUT 360"
12:55
...
12:56
so udevsettle can take up to 360secs ... depending on the HW it has to initialize
12:56
<Gadi>
what about udevtrigger?
12:57
<ogra>
if you want to generate coldplug events for udev to make non hotplug HW appear to the system, you call udevtrigger
12:57
<vagrantc>
so, yaking udevsettle out of the picture risks things like mouse and hard drive swap not working ?
12:57
<ogra>
udevsettle is called afterwards and will only return if udevtrigger is done
12:58
vagrantc, we will have to switch MODULES=most in initramfs ....
12:58
<vagrantc>
ogra: switch to module=most ?
12:58
<ogra>
ten they will exists before
12:58
<Gadi>
doesnt mouse/hard drive swap happen in the regular boot sequence's udev
12:58
not initramfs
12:58
initramfs only needs enough to mount the nfs root
12:58
<vagrantc>
ah, we're only talking about the initramfs?
12:58
<ogra>
ubuntus initramfs detects mice and keyboards and creates the /dev/input links etc
12:58
<Gadi>
I don't see why we load anything but the network drivers
12:59
<ogra>
Gadi, to keep the initramfs small
12:59
<vagrantc>
ogra: what are you meaning with the MODULES=most thing ... ?
12:59
<Gadi>
doesnt it do that again inthe boot process?
12:59
<ogra>
vagrantc, currently we only include NIC drivers
12:59
Gadi, exactly
12:59
<vagrantc>
ogra: and we'll have to add additional modules because ... ?
12:59
<ogra>
Gadi, and it stops for 15-30 secs ..
13:00
vagrantc, --- beacuse you complained about local swap :P
13:00
<vagrantc>
ogra: oh, we don't need that in the initramfs
13:00
<Gadi>
but doesnt swap get turned on during the boot process?
13:00
<Guaraldo>
sbalneav: are you there???
13:01
<vagrantc>
i'm all for keeping it MODULES=netboot (or something even better)
13:01
<Gadi>
heh
13:01
<sbalneav>
Guaraldo: Yep!
13:01
<Guaraldo>
sbalneav: I'm needing some help...
13:01
<ogra>
we run an instance of udev in initramfs already, the second run by init is only to fix permissions and add missing devices ...
13:01
<sbalneav>
Sure, what's up?
13:01
<vagrantc>
ogra: yeah. i stumbled in the middle of the conversation.
13:01
<Gadi>
ogra: right - I think we're all saying similar things
13:02
<vagrantc>
or, i've stumbled in and out of the conversation
13:02
<Gadi>
my question is: do we need udevtrigger in initramfs?
13:02
<Guaraldo>
sbalneav: I've checked all steps to have ltsp-devices working well, but when I plug my pendrive it takes too long to create the icon... when I open it, it's clean...
13:02
<ogra>
vagrantc, what i say is that if we use MODULES=most we can be sure the devices exists before the udve initscript runs
13:02
vagrantc, if we do that we dont need to care if its finished or not
13:02
<sbalneav>
Guaraldo: ok, is it a 64 bit server?
13:03
<ogra>
so we can drop usdevsettle during the bootprocess and just go on with our business
13:03
<Guaraldo>
no...
13:03
<vagrantc>
ogra: hrm.
13:03
<ogra>
ltsp-client-setup will divert it and ltsp-client will switch it back ...
13:03
<Guaraldo>
it was an ubuntu dapper (now upgraded to feisty)... LTSP4.2
13:03
<ogra>
we wont loose anything but bootdelay
13:04
<Guaraldo>
32 bit
13:04
<vagrantc>
ogra: and why is it faster doing the probing from initramfs than from the bootscripts ?
13:04
<ogra>
vagrantc, i had the idea when i saw that 4.2 is doing pretty much the same with udev ;)
13:04
vagrantc, no idea, but it is
13:04
<vagrantc>
gross
13:05
<ogra>
likely because we work 100% from ram there
13:05
<vagrantc>
yeah, i suppose
13:05
<ogra>
while the initscript runs from nfs
13:05
<sbalneav>
Guaraldo: Give the checklist a try, run through it quickly
13:05
!checklist
13:05
<ltspbot>
sbalneav: "checklist" is The checklist for debugging problems with local device access is at http://wiki.ltsp.org/twiki/bin/view/Ltsp/LTSP-42-LocalDev#Troubleshooting. Please work through all 12 steps, record the results, and paste the results to the pastebot at http://pastebot.ltsp.org
13:05
<sbalneav>
Either the terminals don't know their names, or there's some missing fuse libraries.
13:06
<Guaraldo>
sbalneav: Ok... I'll do it again...
13:06
<ogra>
vagrantc, if we can get rid of x detection and i get the greeter fats enough we'll get below 60sec boots
13:06
*fast
13:06
<sbalneav>
LTSPFSD_OPTIONS = "-a" will turn off authentication for the ltspfsd, I beleive.
13:06ccjoe has left #ltsp
13:06
<vagrantc>
ogra: no x detection?
13:07
<ogra>
xorg 7.3 is supposed to not need xorg.conf anymore
13:07
it just hit the archive yesterday :)
13:07
<vagrantc>
but, presumably it does some sort of detection
13:07
<ogra>
i didnt play with it yet though
13:07
well, but noi debconf or xdebconfigurator
13:07
it probes the HW itself ... no scripts involved
13:08
<vagrantc>
yeah, that would be nice.
13:08
<ogra>
it should be a lot faster
13:08
<Guaraldo>
sbalneav: Version 4.2 update 3
13:08
<vagrantc>
hopefully there's some way to override it when it's stupid.
13:08
<ogra>
it will still accept xorg.conf
13:08
<sbalneav>
CHeck the hostname one first
13:08
step 3 or 4 i think
13:09
<ogra>
vagrantc, http://keithp.com/blog/randr_1.2_update.html
13:15
<Gadi>
ogra: do you have source for that greeter you just sent me?
13:15
my focus is a speedy ldm - more so than boot time
13:16
<vagrantc>
Gadi: like, speedy during use ?
13:16
<Gadi>
like when I log out, it should take less than the better part of several minutes to get a login screen ;)
13:16
<ogra>
http://video.google.com/videoplay?docid=-2992183880068262304
13:16
*giggle
13:17
http://people.ubuntu.com/~ogra/ldm-greeter.tgz
13:17
Gadi, ^^
13:17
<Gadi>
thx
13:18
i tell people not to bother turning thin clients off ;)
13:18
but, I can't tell them not to log out
13:19
ogra: I see a lot more than just gtk.h
13:19
;)
13:19
<ogra>
ot takes 5 secs here to get the greeter back
13:19
<cliebow_>
http://counterfeitmini.com/main.swf
13:19
<ogra>
Gadi, right, most of it is from plaing around
13:20
*playing
13:20
thats far from being done yet, its my first sketch :)
13:20vagrantc has quit IRC
13:20
<ogra>
written in an afternoon, dont expect to much
13:21
<sbalneav>
Wow, all that for $99?
13:21
And Reversi too?
13:22
Think they got us beat, ogra
13:22
<Guaraldo>
sbalneav: there are about 1300 workstations... how can i configure to workstations get their name bu dns reverse?
13:23
<sbalneav>
Is that it? They don't know their names?
13:23
<Guaraldo>
sbalneav: no... my test workstation knows...
13:23
<sbalneav>
probably far easier to just set LTSPFSD_OPTIONS = "-a"
13:23Q-FUNK has joined #ltsp
13:24
<Guaraldo>
sbalneav: hummm... I'll try this...
13:24
<Gadi>
ogra: didn't mean to knock it - its a great piece of work - masterful, in fact
13:24
I just want to help get it down from ~50 seconds on this box o mine :)
13:24
<ogra>
Gadi, well, i just wanted to point out that there is more to come
13:25
<Gadi>
great - we can compare notes :)
13:25
<ogra>
the real issue seems to be either the python wrapper or X being slow
13:25
<Gadi>
btw, do we need to kill X on logout?
13:25
why not respawn the greeter
13:25
that would save some
13:26* Gadi wonders why he thinks instead of tries....
13:27
<sbalneav>
Gadi: Hmmm, I think you may need to restart X so you can reset all the xatoms/props etc.
13:29
<Gadi>
nothing a little "while True:" line won't test ;)
13:30
<ogra>
well, we could run the greeter in an xnest fullscreen window :P
13:30
<Gadi>
?
13:30
i was just thinking putting spawn_session in a while loop
13:30
<ogra>
that wouldnt restart the main server
13:30
<Gadi>
ill let you know what i find
13:31
<ogra>
yep
13:31
<jammcq>
isn't there a "reset" to X, without relaunching the Xserver?
13:31
<ogra>
not according to the manpage
13:32
<jammcq>
I think display managers have a way of resetting the Xserver without it restarting
13:32
<ogra>
-noreset
13:32
prevents a server reset when the last client connection is closed. This overrides a previous -terminate command line
13:32
option.
13:32
aha
13:32
i was looking at man Xorg
13:32
<jammcq>
what is the definiton of 'server reset' ?
13:32
<ogra>
no idea, its in Xserver's manpage
13:35
just trying in virtualbox
13:40
hmm, -noreset doesnt help
13:43twinprism has quit IRC
13:44
<ogra>
Gadi, it takes 5 mississippies from clicking the logout buttton to having the greeter back in virtualbox btw
13:44
<Gadi>
heh, if only we could give folks virtual boxes
13:44
;)
13:44
and fyi: the "while True" works :)
13:45
no more respawning X
13:45PMantis has joined #ltsp
13:45
<Gadi>
that shaves a bit of time
13:45
but not much
13:45
<ogra>
how much did that gain you ?
13:45
<Gadi>
maybe 5 seconds
13:45
not much
13:45twinprism has joined #ltsp
13:45
<Gadi>
most of the delay is launching the greeter
13:46
<ogra>
the greeters first action is to switch the cursor btw ...
13:46
<Gadi>
right
13:46
noticed that
13:46
<ogra>
so if you see the rotating pizza you're in the greeter
13:46
<Gadi>
it rotates for about 50 seconds
13:46
:)
13:46
im familiar with its work
13:46
<ogra>
from then on it still takes that long ?
13:47
phew
13:47
<Gadi>
about that, yeah
13:47
<ogra>
thats strange
13:47
<Gadi>
thats what im keen on helping with
13:47
and its roughly same time whether its on boot or logout
13:47
now, this is xover cable to a slow laptop
13:47
as server
13:47
<ogra>
i wonder if you should try out nolapic and noacpi options in your pxe config
13:48
<Gadi>
so,maybe on a server, NFS would serve up faster
13:48
<ogra>
well, i'm doing all my work with my lappie as server
13:48
<Gadi>
is that 'nolapic' or 'noapic'?
13:48
<ogra>
and i dont see 50 seconds on 500Mhz clients
13:48
<Gadi>
(yeah, but my lappy is 5 yrs old)
13:48
<ogra>
lolapic
13:48
heh
13:48
nolapic
13:49
but you can try noapic as well
13:49
<Gadi>
ill also try acpi=off and apm=off
13:49
why not?
13:49
<ogra>
i cant imagine that 100Mhz should make such a big difference
13:49
<Gadi>
nah, its prolly the nfs speed
13:49
in part
13:49
<ogra>
do they have any nifty devices builtin ?
13:50
<Gadi>
nope
13:50
these are as non-nifty as they come
13:50
<ogra>
(i just fund the classmate has a SATA contoller onboard, that slows down the boot a lot)
13:50
<Gadi>
this only has a DOM slot and a lil ide
13:50
iirc
13:51
but, at this point, im just talking greeter loading
13:51
<ogra>
yeah, indeed
13:51
<Gadi>
which is way past device module loading
13:51Guaraldo has quit IRC
13:51
<Gadi>
i think its my lappy
13:51
<ogra>
you BIOS values are fine ?
13:51
<Gadi>
I need a better lappy or better test environment
13:51
<ogra>
on the TC i mean
13:51
<Gadi>
yup
13:55Guaraldo has joined #ltsp
13:56petre has quit IRC
13:59
<Gadi>
ogra: tried it with a faster client (533MHz) and it cuts some 20 secs off the greeter
13:59
down to ~30-35secs
14:00
<ogra>
still to much
14:00
<Gadi>
so, may very well be in part the tc and in part nfs
14:02nicoAMG has quit IRC
14:17
<Guaraldo>
!checklist
14:17
<ltspbot>
Guaraldo: "checklist" is The checklist for debugging problems with local device access is at http://wiki.ltsp.org/twiki/bin/view/Ltsp/LTSP-42-LocalDev#Troubleshooting. Please work through all 12 steps, record the results, and paste the results to the pastebot at http://pastebot.ltsp.org
14:18jumako has joined #ltsp
14:23petre has joined #ltsp
14:23
<ogra>
Gadi, not on relogin ... with ram above the 64M everything should be oaded into ram
14:23
<petre>
mistik1: do you have that link to that udev tutorial handy?
14:29krinns has quit IRC
14:31gepatino has joined #ltsp
14:31jumako has quit IRC
14:36
<sutula>
petre: I wrote down two this morning...not sure if this is what you're looking for:
14:36
http://webpages.charter.net/decibelshelp/LinuxHelp_UDEVPrimer.html
14:36
http://www.reactivated.net/writing_udev_rules.html
14:37
<petre>
yes, that's the one. thanks sutula
14:37
<gepatino>
i'm plannig to upgrade an edgy ltsp server to feisty... any considerations on how to upgrade? my concern is how to upgrade the client root
14:38bobby_C has joined #ltsp
14:40irule has quit IRC
14:42
<Guaraldo>
sbalneav: Is it normal to umount CD-ROMs just pressing the eject button on it?
14:43
sbalneav: Normaly I eject my cds...
14:43
<sbalneav>
Not normally, but thats how we do it on ltsp
14:43
<Guaraldo>
by the icon or by typing eject command...
14:43
Hummm and is it normal to take some time (about 5 secs)?
14:43
<sbalneav>
We wanted to emulate the way it's normally done in the windows world.
14:44
well, it won't allow you to eject until at least 2 seconds of inactivity.
14:45
<Guaraldo>
hummm...
14:45
sbalneav: Ok... thanks!
14:53
<mistik1>
petre: http://www.reactivated.net/writing_udev_rules.html
15:03cliebow_ has quit IRC
15:05Gadi has left #ltsp
15:09Blinny has quit IRC
15:12meduxa has quit IRC
15:14gepatino has quit IRC
15:16vagrantc has joined #ltsp
15:16avoine has joined #ltsp
15:17
<avoine>
Hi, It's possible to run Xorg on a ltsp server without a physical screen?
15:17
<sutula>
Avatara: Sure
15:18
drat...avoine ^^^
15:18
<avoine>
oh
15:18
ok
15:19
<vagrantc>
avoine: what exactly do you mean?
15:19
<sutula>
avoine: I assume there's more behind your question
15:20
<avoine>
I want a ltsp server running with Xen on a virtual machine
15:21
<vagrantc>
you don't need to run Xorg on the server at all, normally.
15:21
<avoine>
and I'm asking if it's possible to start X without a real screen
15:21a5benwillis has quit IRC
15:21
<vagrantc>
well, on the LTSP server, you don't normally need to run Xorg
15:22
<avoine>
ok
15:22PMantis has quit IRC
15:23
<sutula>
avoine: If you're really trying to run XDMCP on the server, you'll probably need some options in e.g. gdm.conf to tell it not to open a local screen
15:24
<avoine>
ok
15:25muh2000_ has quit IRC
15:30sbalneav has quit IRC
15:30dean_za_ has quit IRC
15:31Egyptian[Home] has quit IRC
15:35dean_za_ has joined #ltsp
15:36McDutchie has joined #ltsp
15:48joebaker has quit IRC
15:50joebaker has joined #ltsp
15:50
<joebaker>
Can anybody point me to the syntax for specifying auto login over XDMCP within gdm?
15:51
I've been used to doing this through KDM.
15:51
But after switching to ubuntu, I'm now using gdm.
15:54
I've found this:
15:54
http://64.233.167.104/search?q=cache:Q1Op6sM54BwJ:www.thealbertalibrary.ab.ca/files/FriE1.ppt+ltsp.org+gdm.conf+auto+login&hl=en&ct=clnk&cd=2&gl=us&client=firefox-a
15:54
AllowRemoteAutoLogin=true
15:57vagrantc has quit IRC
15:57monteslu has joined #ltsp
16:02
<joebaker>
I think I might just switch to KDM It's a very simple way to authenticate users against their MAC addresses.
16:06sepski has quit IRC
16:07mhterres has quit IRC
16:08test__ has joined #ltsp
16:11
<test__>
hi, whats wrong in this dhcpd.conf ? I want PXE booting http://pastebin.ca/495613
16:12avoine has quit IRC
16:14
<ogra>
root-path is missing
16:14
<test__>
ogra, root path for ltsp? what about if I want a generic pxe booting?
16:15
<ogra>
well your initramfs needs to know what nfsroot to mount
16:15
root-path holds the path info for that
16:16
<test__>
ogra, when I uncomment the lines 21 to 29, it works well
16:18petre has quit IRC
16:19efra has quit IRC
16:27irule has joined #ltsp
16:30Q-FUNK has quit IRC
16:36twinprism_ has joined #ltsp
16:37cliebow has quit IRC
16:38cliebow has joined #ltsp
16:39twinprism has quit IRC
16:40ogra has quit IRC
16:45ogra has joined #ltsp
16:46Guaraldo has left #ltsp
17:05mfdutra has quit IRC
17:08bobby_C has quit IRC
17:42gentgeen__ has quit IRC
18:08McDutchie has quit IRC
18:17
<jammcq>
Qiot
18:17
oops :)
18:17jammcq has quit IRC
18:34cliebow has quit IRC
19:02floatingman has joined #ltsp
19:05floatingman has quit IRC
19:13freet15 has joined #ltsp
19:14VladimirBG has joined #ltsp
19:15
<VladimirBG>
hello
19:15
is is posible to have 3d desktop effects on a thin client, and if so, does the 3d card needs to be in the server or in clients?
19:15
is it*
19:15gepatino has joined #ltsp
19:16
<VladimirBG>
anyone?
19:17* VladimirBG pokes sleepy people
19:25
<VladimirBG>
please, can anyone answer?
19:27nickbenj has joined #ltsp
19:27
<gepatino>
hi
19:27twinprism_ has quit IRC
19:27
<gepatino>
VladimirBG: can i help you somehow?
19:28
<VladimirBG>
is it posible to have 3d desktop effects on a thin client, and if so, does the 3d card needs to be in the server or in clients?
19:29
<gepatino>
i've never tried, but you should have the 3d card at the clients
19:29
since all the 'drawing' is done there
19:30
you should also install the video card modules in the nfs share that servers as root file system for the clients
19:30
mm... nice question..
19:30
now i file like testing it :)
19:30
but i won't have access to a ltsp server till tuesday :(
19:31
<VladimirBG>
so drivers need to be installed on the server, and config set in ltsp client config file for each clients graphics hardware?
19:33
<gepatino>
yes
19:34
in the ltsp config file, you can specify the modules you need to load.
19:34
maybe you could put there the drivers for the 3d card
19:36
<VladimirBG>
hmmm, now if only someone produced small form factor thin clients with some basic 3d chip, like intel maybe(OS drivers)
19:37
would be nice if someone added howto to ltsp's wiki on 3d on clients
19:38
<gepatino>
that would be really nice
19:38
if i have some time, maybe next week (or the other) could make some tests
19:39
but all my terminals have really old video cards
19:39
no 3d... :(
19:39
<VladimirBG>
:(
19:39
<gepatino>
try again later in this channel... it seems to be that all the wizards are gone :)
19:39
<VladimirBG>
I'm relatively inexperienced, so I doubt I could set it up
19:40
alone that is
19:41
will do, I mean it would be interesting to see a cheap thin client do eyecandy better than vista :P
19:43
also, I take it I can use different kernels on the server, for example, PPC, i686 and K7?
19:47
<gepatino>
on the server you always run the same kernel
19:47
and that has to match the architecture of the server
19:47
then, for the clients
19:47
you could load different kernels for each one
19:48
but the default i386 kernel works on almost any pc
19:49
<VladimirBG>
so basically server needs skill in setting up(to match all the client hardware properly), but clients are just start and run
19:49
?
19:50
<gepatino>
well.. server is not that hard to setup, and yes... you don't to setup anything at the client (just network boot)
19:51
<VladimirBG>
hmmm, nice
19:51
<gepatino>
the server hosts al the clients configuration
19:51
if the client breaks, dump it, plug a new one, and that's all
19:51
just works® :)
19:51J45p3r has joined #ltsp
19:53
<VladimirBG>
I take it server need to be in failsafe category with hot-swapable HDDs (in RAID) and hotswapable CPUs for business application?
19:54
<gepatino>
it depends on how critical the applications are
19:54
at the office, we don't have that much
19:55
<VladimirBG>
but if server goes, everything goes?
19:55
<gepatino>
and there are five persons programing on the same ltsp server
19:55
evry day
19:55
yes, if the server fails, you're stuck
19:55
but you have to balance price+security
19:56
<VladimirBG>
can 2 servers be used if one fails, the second could kick in?
19:56
(failsafe servers cost a mint)
19:56
<gepatino>
only if you set a high availability cluster
19:56
and never tried it with ltsp, but should work
19:57
<VladimirBG>
so, if I set up a high availability cluster with 2 servers, if one fails the other will continue working?
19:57
and data is synced between their HDDs?
19:58
sorry I'm bugging you, I just want to set up thin clients at my uni
19:58
they are about to buy new computers, and are skeptic about thin clients
19:59
<gepatino>
in that case, the best (imho) would be:
19:59
two servers in HA cluster
19:59
and a net area storage
20:00
the servers mounting the file system from the storage, so the data keeps always up to date
20:01
<VladimirBG>
hmmm, and I take it I should use really fast network connection with data storage, like a gigabit link?
20:02
<gepatino>
shure
20:02
sorry.. i have to go now... the pizza is waiting for me
20:03
to get used to ltsp, i recomend to try first with a single server
20:03
<VladimirBG>
ok, thanks, you helped me quite a bit
20:03
:)
20:03
I tried with edubuntu
20:03
I like it
20:03
<gepatino>
the server should have much memory
20:04
about 128Mb per client
20:04
plus 256 for the server
20:04
and if its multi-cpu or multi-core, the best
20:04
with that, and a 100Mbps switched network, you're done
20:05
with a 2Gb RAM server, you should be able to serve a dozen clients or so
20:06
well... see you later...
20:06
<VladimirBG>
so, a quadcore intel with HT and 8GB should be able to run a number of PCs?
20:06
bye
20:07
<gepatino>
with that amount of ram... maybe four dozen of clients... in that case, think about a gigabit ethernet for the server
20:07
since it will generate a lot of traffic
20:07
the clients can be 100mbit
20:07
good luck!
20:07gepatino has quit IRC
20:08VladimirBG has quit IRC
20:15freemind has joined #ltsp
20:29cliebow has joined #ltsp
20:38freemind has quit IRC
20:50cliebow has quit IRC
20:55J45p3r has quit IRC
21:11jammcq has joined #ltsp
21:11
<jammcq>
hello
21:14cliebow has joined #ltsp
21:19ogra has quit IRC
21:20ogra has joined #ltsp
21:22muh2000 has joined #ltsp
21:27sbalneav has joined #ltsp
21:27
<sbalneav>
Evening all
21:28
<jammcq>
!s
21:28
<ltspbot>
jammcq: "s" is Scotty!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
21:28
<sbalneav>
Evening jammcq
22:04irule has quit IRC
22:18Lumiere has quit IRC
22:18Lumiere has joined #ltsp
22:19Lumiere has quit IRC
22:19GiantPickle has quit IRC
22:19pipedream has quit IRC
22:19sutula has quit IRC
22:19sutula has joined #ltsp
22:22dberkholz has quit IRC
22:22dberkholz has joined #ltsp
22:23Lumiere has joined #ltsp
22:23fernando1 has quit IRC
22:23str4nd has quit IRC
22:23dan__t has quit IRC
22:23viking-ice has quit IRC
22:23str4nd_ has joined #ltsp
22:23dan__t has joined #ltsp
22:24viking-ice has joined #ltsp
22:24pipedream has joined #ltsp
22:24GiantPickle has joined #ltsp
22:29sbalneav has quit IRC
22:42fernando1 has joined #ltsp