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


Channel log from 23 February 2010   (all times are UTC)

00:01shogunx has quit IRC
00:29shogunx has joined #ltsp
00:33alexbodn has joined #ltsp
01:23Egyptian[Home]1 has joined #ltsp
01:25Egyptian[Home] has quit IRC
01:40
<alkisg>
vagrantc: hi, could you give me a quick explanation of how ltsp_config.d works? E.g. could I make some Ubuntu-custom/* ltsp-build-client plugins, which would generate some ltsp_config.d/* scripts, and those script in turn could e.g. set X_COLOR_DEPTH=16 for thin clients and leave it unset for fat clients?
01:43
<vagrantc>
alkisg: yup.
01:43
<alkisg>
Thanks! :)
01:43
<vagrantc>
alkisg: they're just files sourced from ltsp_config
01:44
in fact, i was thinking that some of the hard-coded getltscfg-cluster stuff could be easily integrated into that.
01:44
<alkisg>
So we could pretty much split ltsp_config into several files could go in there? Got it...
01:44
<vagrantc>
alkisg: you could also have your wget stuff that includes shell-sourceable values drop stuff in there from the initramfs.
01:44
<alkisg>
*syntax error :)
01:45
<vagrantc>
alkisg: i.e. wget http://server/mac=foo&ram=bar/lts.conf
01:45
<alkisg>
Ooooh nice idea :)
01:46* alkisg would also like a wget hook upon ldm login :)
01:46
<vagrantc>
saves from having to re-invent the whole idea.
01:46
alkisg: ah, yes. we could split ltsp_config into several files ...
01:46
it adds a slight bit of overhead to source several files instead of one ... not sure how much that matters.
01:47
<alkisg>
I don't think that would be more than a few msec...
01:47
<vagrantc>
i did think of that when i wrote it, but decided to leave it as is for now, and just make it an optional directory.
01:49
it does run before all the defaults... so all the [ -z "$FOO" ] checks should be fast if you set those values.
01:50
<alkisg>
Right, so I can't e.g. check for $LTSP_FATCLIENT, I'd need to check for -f /etc/ltsp_fat_chroot instead
01:51
<vagrantc>
i think the original reason i wanted it was to be able to set LTSP_FATCLIENT based on system resources.
01:52
<alkisg>
Hmmm or at least disable it for systems with < 256 ram... not a bad idea
01:52
<vagrantc>
though the screen locker issue is definitely a blocker for deploying it in many cases at freegeek.
01:53pts_ has quit IRC
01:53
<vagrantc>
i really need a per-session ability to set or disable LTSP_FATCLIENT ...
01:53
as i think guest logins should be able to use LTSP_FATCLIENT (if system resources are sufficient) but logins with actual user accounts shouldn't.
01:54
at least, that's what i want.
01:54
as the guest logins shouldn't have screen lockers anyways...
01:54
but actual user accounts should
01:54
and guest logins shouldn't need to do anything requiring authentication
01:58
<alkisg>
Something like ltsp_config upon login wuold be nice
01:58gnunux has joined #ltsp
01:58
<alkisg>
But actual users might also need to run multimedia apps, googleearth etc, so it would have to be more configurable
01:59
<gnunux>
hi
01:59
<vagrantc>
alkisg: the other idea i had was to run the handful of applications as remoteapps.
01:59
alkisg: all the screen lockers and whatnot...
02:00* vagrantc heads to sleep
02:01
<alkisg>
'night vagrantc
02:02vagrantc has quit IRC
02:41pts has joined #ltsp
02:54Comete has joined #ltsp
02:54
<Comete>
hello
02:59
i have stability problem with pulseaudio on my thinclients and i would like to use esd or nasd instead, but i would like to have sound with flash too, is it possible ?
03:18ZyTer has quit IRC
03:20bobby_C has joined #ltsp
03:40
<Comete>
is there any documentation somewhere explaining how to replace pulseaudio with esd ?
03:45atomic007za has quit IRC
03:46
<alkisg>
stgraber: SSH_FOLLOW_SYMLINKS causes problems for fat clients. I'd like the following weird, but conventient, behavior by default:
03:47
When SSH_FOLLOW_SYMLINKS is not set, and only for fat clients, mount /home/username without following symlinks, but mount any other extra mounts as usually, i.e. with following symlinks
03:48
Of course if the user sets SSH_FOLLOW_SYMLINKS to either true or false, the previous behavior won't happen. Would you agree to that?
03:48
highvoltage, also, any opinions on this? ^^^
03:51
<Comete>
stgraber: thanks for your email, i will stay with gcompris 8 for the moment
03:52atomic007za has joined #ltsp
04:04
<Comete>
no way to replace pulseaudio by esd ? i've tried SOUND_DAEMON=esd but i still get no sound...
04:15Comete has quit IRC
04:20ogra has quit IRC
04:32ogra has joined #ltsp
04:37Sarten-X has quit IRC
04:59mikkel has joined #ltsp
05:02hersonls has joined #ltsp
05:24ZyTer has joined #ltsp
05:25HardDisk has quit IRC
05:40
<highvoltage>
alkisg: I guess that would make sense
05:40
<alkisg>
highvoltage: thanks, I'll commit it and if anyone objects, we can just revert it :)
05:41
<highvoltage>
alkisg: heh, it's not like I have any say anyway :)
05:41F-GT has joined #ltsp
05:42
<alkisg>
highvoltage: sure you do... and I'd vote for you whenever you feel like joining the team "officially" ;)
05:54shogunx has quit IRC
05:59shogunx has joined #ltsp
06:00
<highvoltage>
alkisg: :)
06:08shogunx has quit IRC
06:20shogunx has joined #ltsp
06:23ogra has quit IRC
06:32ogra has joined #ltsp
06:32ogra_cmpc has quit IRC
06:34ogra_cmpc has joined #ltsp
06:35scottmaccal has joined #ltsp
06:46jcarter has joined #ltsp
07:06comfrey has quit IRC
07:07
<stgraber>
alkisg: setting SSH_FOLLOW_SYMLINKS to false by default should be ok. I know it's also breaking localapps like google chrome.
07:07
<alkisg>
stgraber: for all? Or just for fat clients?
07:08
<stgraber>
for all
07:08
<alkisg>
stgraber: I don't think it breaks thin clients
07:08
<stgraber>
it breaks some localapps
07:08
<alkisg>
Ah sorry yes it does break localapps
07:08
stgraber: but I would like for extra-mounts-follow-symlinks to be enabled by default
07:09pmatulis has joined #ltsp
07:09
<alkisg>
stgraber: so can we do this: when SSH_FOLLOW_SYMLINKS is unset => don't follow symlinks for home, but do follow symlinks for extramounts
07:09
<stgraber>
alkisg: yeah, that should be safe
07:09
<alkisg>
OK, thanks
07:14comfrey has joined #ltsp
07:15mgariepy has joined #ltsp
07:25shawnp0wers has joined #ltsp
07:25litlebuda has joined #ltsp
07:27thunsucker has joined #ltsp
07:33litlebuda has quit IRC
07:40mikkel has quit IRC
07:48alkisg has quit IRC
08:04ogra_cmpc has quit IRC
08:05ogra_cmpc has joined #ltsp
08:05otavio has quit IRC
08:05otavio has joined #ltsp
08:07Gadi has joined #ltsp
08:09rjune has joined #ltsp
08:28ogra_cmpc has quit IRC
08:30ogra_cmpc has joined #ltsp
08:30vvinet has joined #ltsp
08:46
<sbalneav>
Morning all
08:47
<mgariepy>
morning
08:58kebax has quit IRC
08:58kebax has joined #ltsp
09:01
<thunsucker>
sbalneav: morning
09:02johnny has joined #ltsp
09:13nokia3510 has quit IRC
09:13nokia3510 has joined #ltsp
09:25mikkel has joined #ltsp
09:28Ralth has joined #ltsp
09:29
<Ralth>
Should all usernames on the server show up in the list displayed at the end of the output from ltsp-update-image? If so, what would cause them to not appear? I'm getting the login screen on a client, but can't log in, and I'm thinking this might be the issue, though i don't know for sure.
09:30
<johnny>
no.. it doesn't
09:30
not that i've ever see nanyways
09:30
did you change the ip?
09:30
of the server*
09:32
<Ralth>
I did. But I did update the ssh keys.
09:32
Do I need to do anything else?
09:34
<johnny>
and then you rebuilt the image
09:34
that's all you should need to do
09:39
<Ralth>
Hmmm. Done that, and I still can't log in.
09:39
<thunsucker>
Ralth: does it say it can't connect the server or any other error?
09:40
<Ralth>
No, it says "verifying username and password" and then takes me back to the login page.
09:41
Sorry, the exact message is "verifying password. please wait."
09:41
<Gadi>
Ralth: are the NICs in your server on different subnets?
09:42
<Ralth>
I only have one NIC in the server ... I'm trying to set this up with virtual machines
09:42
<Gadi>
gotcha
09:42
you should set a shell on the thin client so you can investigate further
09:43etyack has joined #ltsp
09:43
<Ralth>
(Doing some testing/learning before I attempt to implement this for a computer lab.)
09:43
<Gadi>
in /var/lib/tftpboot/ltsp/i386/lts.conf , add:
09:43
[default]
09:43
SCREEN_07=ldm
09:43
SCREEN_08=shell
09:44
then, when the client boots, you can ctrl-alt-f8 to a shell
09:44
and check to make sure you can ssh in correctly to the server
09:44
also, make sure you have a desktop environment installed on the server (such as gnome)
09:45
<Ralth>
lts.conf does not exist in that folder, should I create it?
09:45
<Gadi>
yes
09:46
(a little background: when you log in with LDM, you are basically executing: ssh -X user@server /etc/X11/Xsession
09:47
so, you need to make sure that: a. the user can ssh to the server, b. that there is a desktop installed on the server
09:47
c. that the server is a known host to the client (thats what ltsp-update-sshkeys does
09:47
<Ralth>
ok. I'm running Gnome on the server right now, so that's fine
09:47
<Gadi>
if all the above is in order
09:48
you should investigate the user's ~/.xsession-errors file
09:48
sometimes compiz runs when it shouldnt and it crashes the client
09:48
<Ralth>
... right now, trying to figure out how to do ctrl-alt-f8 on the vm. :-)
09:48
<Gadi>
what are you using?
09:49
<Ralth>
vmplayer
09:49
<Gadi>
hmm... try just alt-f8 or alt-shift-f8
09:49
or some such
09:54
<comfrey>
hey all
09:54
i am getting the wrong kernel (64 bit) on my 32 bit client
09:55
i am using ubuntu atm, and wondering which image needs attention
09:57
<Ralth>
Gadi: a,b, and c in the list you gave are good. Here's the .xsession-errors for the user: http://pastebin.com/QTTvSrYF
09:59alkisg has joined #ltsp
09:59__nubae__ has quit IRC
09:59__nubae__ has joined #ltsp
09:59nubae has quit IRC
10:00nubae has joined #ltsp
10:00
<comfrey>
i think it is the nbi...
10:01Sarten-X has joined #ltsp
10:01
<comfrey>
updating the nbi is not update-image no
10:01
how to regen the nbi?
10:05
<johnny>
uhmm.. do you have any 64bit clients?
10:05
<comfrey>
no, just 32.
10:06
<johnny>
then that's not the problem
10:06
delete your chroot
10:06
and start over
10:06
ltsp-build-client --arch=i386
10:06
it builds 64bit chroots on a 64bit server by default
10:07
<comfrey>
ok, i'll take a swing. the chroot being /opt/ltsp/i386 ?
10:07
<johnny>
you shouldn't have an i386 one..
10:07
<Gadi>
Ralth: are you logging in at the server and the thin client with the same username?
10:07
<johnny>
you need to make one
10:07
<Ralth>
yes
10:07
<johnny>
that won't work
10:07
need a new username
10:07
<Gadi>
Ralth: at the same time?
10:07
<comfrey>
there is one , but it is using the 64 bit for some reason
10:07
<johnny>
yes.. won't work at the same time.. :)
10:08
delete the 64bit one
10:08
<Ralth>
Ah. Ok.
10:08
<johnny>
oh wait.. isn't there a purge command Gadi ?
10:08
<comfrey>
i did and created a symlink to the 32 bit one
10:08
<johnny>
for ltsp-build-client
10:08
comfrey, the problem is in your tftpboot
10:08
remove the symlink
10:08
the problem now is in /var/lib/tftpboot/ltsp
10:09
<comfrey>
ok, lets see about that
10:09
<Gadi>
comfrey: what version of ltsp do you have there?
10:09
<comfrey>
ok still need to regen the image then
10:09
ubuntu...
10:09
uh 9.04
10:09
<Gadi>
9.10?
10:09
ah, ok
10:10
then, yeah, you might want to delete your amd64 chroot and img file in /opt/ltsp/images and adjust your inetd.conf to use port 2000 for i386.img
10:10
and, of course build the i386.img
10:11
don't forget to restart openbsd-inetd service after updating inetd.conf
10:11
<comfrey>
i updated the image and that is when all this started
10:11
all was working fine before the upgrade
10:12
<Gadi>
ok lets take a step back then
10:12
<comfrey>
i see...
10:12
<Gadi>
so, you are getting a 64-bit kernel on boot?
10:12
<anivair>
problem: my ltsp server keeps locking up. the users will lock up across the board for a few minutes and then it frees up again. the server looks fine when it happens and I can run things normally (top shows no processor hog). but it is stopping me from logging into the server remotely (so maybe a networking issue?)
10:12
<comfrey>
looks like 2000 is amd64 in inetd.conf
10:13
<Gadi>
comfrey: are you getting a 64 bit kernel or the 64 bit chroot?
10:13
<comfrey>
kernel
10:13
<Gadi>
ok, then dont worry about inetd.conf
10:13
<Ralth>
Gadi: rebooted the server to be sure I was not logged in at all, and I'm still having the same problem.
10:13
<Gadi>
comfrey: what filename is in /etc/ltsp/dhcpd.conf?
10:14
<comfrey>
/ltsp/i386/pxelinux.0
10:14
and else nbi
10:14
<Gadi>
Ralth: try adding: LDM_DIRECTX=True to your lts.conf file (the one you created)
10:15
comfrey: then, you cannot really be getting a 64 bit kernel
10:15
<anivair>
restarting ssh may or may not have had an effect. checking again
10:15
<Gadi>
have a look at /var/lib/tftpboot/ltsp/i386/pxelinux.cfg/default
10:16
<comfrey>
DEFAULT vmlinuz ro initrd=initrd.img quiet splash nbdport=2001
10:16
looks right...
10:16
<Gadi>
yup
10:17
comfrey: try restarting dhcp3-server and openbsd-inetd
10:17
<Ralth>
Gadi: do I need to rebuild after that?
10:17
<Gadi>
and, for good measure, run ltsp-update-kernels
10:17
Ralth: nope
10:17
Ralth: just reboot
10:18
<Ralth>
client, server, or both?
10:18
<comfrey>
grr...
10:18
<Gadi>
Ralth: your .xsession-errors file complained about not being able to properly connect to localhost:11 display or to the xrdb (the X resource db)
10:18
Ralth: client
10:19
<comfrey>
so still getting the error
10:19
<Ralth>
Still not working.
10:22
Gadi: So, I tried creating a different user ... and it works fine
10:22
<anivair>
update: I had to reboot the server. I really hope that doesn't happen again.
10:22
rebooting takes us down for wuite a bit
10:24
<Ralth>
Gadi: There's not a big reason to get the other account working, since I'm just testing at this point. Thanks for the help!
10:27
<anivair>
update: rebooting forced me to rebuild the image and reload the kernel
10:27
not okay
10:28Ralth has quit IRC
10:28
<ogra>
thats why you define maintenance windows as a good admin and dont do such stuff during work hours :)
10:30gnunux has quit IRC
10:30__nubae__ has quit IRC
10:30nubae has quit IRC
10:31ogra_cmpc has quit IRC
10:33ogra_cmpc has joined #ltsp
10:36
<anivair>
anyone have any thoughts as to why the server may have locked up like that?
10:37
<ogra>
sounds like a NIC problem
10:37
have you checked dmesg ?
10:37
<anivair>
of course not ... but I will now
10:38alexbodn has quit IRC
10:39
<ogra>
well, if you rebooted thats wiped
10:39
so you can only wait until it happens the next time
10:39
probably kern.log has something though
10:40
in any case it sounds like a network prob, either your NIC has issues or probably your swithc is overloaded or something like that
10:40
was there any particular reason you rebuiolt your image after reboot ?
10:44Faithful has quit IRC
10:45
<anivair>
because when I rebooted the first time ltsp was sticking on the load
10:45
it was hanging on loading vmlinuz
10:46
<ogra>
and *that* made you rebuild everything ?
10:46
<anivair>
that's happened a few times lately and I always have to rebuild
10:46
it is what solved it
10:46
<ogra>
no, you dont
10:46
you should find out what blocks your network :)
10:46carbm1_ has joined #ltsp
10:47
<anivair>
well, i'd love to, but atm having people on is imperative
10:47
I'd love for these people to give me some time to actuially fix things
10:47
<ogra>
right, do it in your regular maintenance window wrie a script or something that produces some network load
10:47
*write
10:48
<anivair>
I could give that a shot. I'm a little afraid to fiddle with it too much right now
10:48
but I'll take a loko at it in a bit
10:49
I do have an ssh window running to the machine, but I'm a little hesitant to play with the network at all right now
10:50
<ogra>
yeah, dont do it during business hours
10:50
but inspect all logs etc
10:51
and check your network hardware too
10:51
<anivair>
wait ... let me see if my understanding of ltsp is right here
10:51
my thin clients are patched into a switch, but they are all getting their network connection from the ltsp server (ie, if they are going out to ther machines or ht net)
10:52
so if their desktops locked up, but the ltsp server didn't lock up, then it should be the nic and not hte switch, yes? Since nobody else plugged into that switch had any issues?
10:52
<ogra>
your thin clients are somewhat like ethernet based extensions of kbd/mouse/display of the server
10:53
so your users fully work on the server, all apps usually run there
10:53
if the *display* locks up then it is very likely you have a network saturation
10:54
note that mouse, kbd and display drivers run on the client ... but the content of your display comes from the server
10:54
<anivair>
I'm honestly not sure if it was the display or their whole connection. can't really tell the difference at a glance
10:54
<ogra>
i.e. your mouse and kbd can be usable but the content you see is completely unresponsive
10:54
<anivair>
so if that's the case, then it's the server nic?
10:55
<ogra>
or the connetion to it
10:55
i.e. switch cables etc
10:56carbm1_ has quit IRC
10:57
<ogra>
if your mouse locks up too on the client, its rather a ram issue of the client machine ...
10:58
how much ram do your clients have, do you know that ?
10:58
(though thats unlikely, since you said you cant even ssh)
10:59
<anivair>
yeah, it blocked my ssh login to the system
10:59
which makes me think it's definately network traffic
11:00
I'm running bmon right now
11:00
so if it goes down again, I'll at least be able to se if there as a spike in trraffic
11:25
hrm ... do I need to use update-rc.d to set dhcp services to run at boot? right now I have to start them manually
11:25
<ogra>
wow, your server seems really messed up
11:26
<anivair>
well, I turned it off when i installed it, I just don't recall how
11:26
so i'm not sure how to undo it
11:26
<ogra>
no user should ever use update-rc.d its a maintainer script tool that should only be used by package scripts
11:26
<anivair>
I can od it via the gui, but I'd rather use the command line if I can help it
11:26
true, but there is no dhcp3.conf script in /etc/init
11:27
<ogra>
hwat are you running there ?
11:27
<anivair>
ubuntu karmic
11:27
(or did you mean in /etc/init)?
11:27
<ogra>
then there shouldnt be a dchpd script in /etc/init :)
11:28
its in the old /etc/rc2.d dir for karmic
11:28
<anivair>
ahh
11:28
I see it there
11:28
it's a symlink
11:28
<ogra>
with S and a number ?
11:29
<anivair>
yes
11:29
<ogra>
good
11:29
then it starts automatically
11:29
<anivair>
it isn't doing that
11:29
<ogra>
(unless the link is broken, which shouldnt be)
11:29
<anivair>
nope, link is fine
11:30
however, dhcp is not starting when i boot
11:30
<ogra>
well, it does for everyone else :)
11:30
what did you do to make it not start
11:30
<anivair>
heh ... I wish I remembered.
11:31
it's been a while
11:31
<ogra>
check /etc/default/
11:31
there should be a file for dhcpd
11:32
<anivair>
there is a file, though it looks like all that's not commented out is "INTERFACES="""
11:33etyack has quit IRC
11:33
<ogra>
ok, thats fine
11:34
check your logs ... syslog and messages, one of them should show you dhcpd when it attempts to start
11:34
and it should also log errors
11:37carbm1_ has joined #ltsp
11:38
<anivair>
looks like I've got a profile load and a profile replace in dmesg for dhcp3
11:38
does that not mean it should be running?
11:39
<carbm1_>
I know I've asked a lot of questions lately and greatly appreciate the help in getting my first LTSP box up and running. But, I gotta ask another one and I'm hoping somebody can at least point me the right direction. I've got LTSP with LDAP authentication working. Our users home directories are on a Novell SLES10.2 box with Open Enterprise Server (eDirectory). Is there a way to mount the users home directories our Novell servers? Anybody doing som
11:39
ething similar? I know Open Enterprise Server supports Samba.
11:49
<ogra>
anivair, no, thats just apparmor
11:50
<comfrey>
so i moved over to the backport out of debian on lenny...
11:50
things are better. proper kernel and all
11:50
but i am unable to authenticate for valid users
11:50
it is not letting me login
11:51
<ogra>
comfrey, check auth.log if you see login attempts
11:51
if they succeed, check ~/.xsession-errors of the user you try with
11:51abeehc has quit IRC
11:52
<comfrey>
yeah, it is an ssh port thing
11:53
<anivair>
ahh
11:53
<comfrey>
what is the ssh port specification for lts.conf?
11:53
<ogra>
do you not use 22 ?
11:53
<comfrey>
man 5 lts.conf is missing in the package
11:53
i use some other port
11:53* ogra thinks its SSH_OVERRIDE_PORT or something like that
11:53
<comfrey>
yeah, that sounds right
11:57
ok, that is better
11:57
thanks ogra
11:57
<ogra>
:)
11:59tstafford has quit IRC
11:59tstafford has joined #ltsp
12:02alexbodn has joined #ltsp
12:09ogra_cmpc has quit IRC
12:15etyack has joined #ltsp
12:19rjune has quit IRC
12:23ogra_cmpc has joined #ltsp
12:26rjune has joined #ltsp
12:31etyack has quit IRC
12:32highvoltage has quit IRC
12:34highvoltage has joined #ltsp
12:38Lns has joined #ltsp
12:45
<comfrey>
ok, so all is looking good, xcept... this eeebox with dvi is failing to startx.
12:45
i have set in lts.conf X_DEVICE_OPTION_02 = "ForcePanel true"
12:47
<Gadi>
change to: X_DEVICE_OPTION_02 = "\"ForcePanel\" \"true\""
12:55
<comfrey>
no go.
12:57
<alkisg>
Would it be possible before the TC calls nbd-client, to send (write with nc) the desired image to the nbdrootd, so that nbdrootd gets is (read with nc) and calls nbd-server with the appropriate image as a parameter?
12:57
This way, we could just use one NBD_PORT for all images...
13:00bobby_C has quit IRC
13:00
<Gadi>
comfrey: why are you telling an eeebox to force its panel on?
13:01
that doesnt seem to make sense if your goal is to have dvi output
13:01
<comfrey>
i see, i misunderstood some documentation then,
13:02
is there some recomended way to force dvi on, or... some driver to specify perhaps?
13:02
i tried vesa
13:02
<Gadi>
can you get a shell on the client?
13:03abeehc has joined #ltsp
13:03
<comfrey>
yes, i can set that up i assume, hang on
13:10
so i have a login, but there are no users set up in the chroot
13:10
i think i am missing something
13:10
<Gadi>
did you set a screen to shell?
13:10
do this in lts.conf:
13:10
SCREEN_07=ldm
13:10
SCREEN_08=shell
13:10
<comfrey>
oh, i see getty is in the way
13:10
<Gadi>
then, when you boot the client, press ctrl-alt-f8
13:10
<comfrey>
moving it to 8
13:11
thanks
13:11abeehc has quit IRC
13:11
<Gadi>
once at a shell, scp /var/log/X*log user@server:
13:12
so you can paste the xorg log file
13:13abeehc has joined #ltsp
13:21vvinet has quit IRC
13:22
<comfrey>
http://pastebin.com/j7hDgNsM
13:26
<Gadi>
_UsUrPeR_: ping
13:26
<_UsUrPeR_>
Gadi: woop
13:26
what's up?
13:26
<Gadi>
hey, I think comfrey here needs that LVDS output disable thing you have
13:26
for the intel driver
13:26
can you paste it?
13:27
<_UsUrPeR_>
Gadi: sure. Let me read above for a second, comfrey
13:27
<Gadi>
he's trying to get an EEEbox to work
13:27
<_UsUrPeR_>
EEEbox = intel atom chipset?
13:27
<Gadi>
log file appears to enable all the outputs
13:27
yup
13:27
<_UsUrPeR_>
ok.
13:27
comfrey: what version of ubuntu are you running?
13:28
<comfrey>
debian, lenny backports
13:28* Gadi thinks he's on 9.04
13:28
<Gadi>
oh, I take that back
13:28
hehe
13:28
<comfrey>
i am leaving the big U behind
13:28
<Gadi>
fair enough - same driver tho with same quirks
13:28
:)
13:28
<comfrey>
vagrantc showed me the light
13:28
;)
13:28
<Gadi>
and same LTSP
13:29
so, we're all good
13:29
<ogra>
... different kernel and drm drivers
13:29
<Gadi>
quiet, you
13:29
:P
13:29* ogra hides
13:32vvinet has joined #ltsp
13:33
<_UsUrPeR_>
comfrey / gadi: here's the file http://pastebin.com/22ZbkiNG
13:34
comfrey: that belongs in /opt/ltsp/i386/usr/share/ltsp/screen-session.d/XS20-server-intel-desktop-hack
13:35
comfrey: wait, are you saying that dvi output is not working at all?
13:35
<comfrey>
it works, but no x
13:36
<_UsUrPeR_>
comfrey: can you tell if intel's video drivers are being loaded properly in /var/log/Xorg.7.log?
13:37
<Gadi>
they are
13:38
<_UsUrPeR_>
hmm. Ok. How many video outputs are on that client?
13:38
just single DVI? or DVI + VGA?
13:38
<comfrey>
just one dvi
13:39
the scipt appears to have not worked
13:39
<_UsUrPeR_>
:/
13:39
you did update the ltsp image, correct?
13:39
<comfrey>
nfs
13:39
<_UsUrPeR_>
ahh
13:39* ogra thinks we should all just switch back to matrix printers as output devices ... saves so many headdaches
13:40
<_UsUrPeR_>
:)
13:40* _UsUrPeR_ can hear the forests crying now
13:40
<ogra>
e-ink !
13:41
<Gadi>
comfrey: go back to the shell and check /var/run/ltsp-xorg.conf
13:41
that script has lspci and hal checks that may not work on lenny
13:42
<_UsUrPeR_>
Yeah, not quite sure this will work properly without some finagling.
13:43
Gadi: I believe that this is the correct place to start though. The intel LVDS bug is omnipresent in every Atom chipset I've tested.
13:43* Gadi nods
13:45
<comfrey>
ok, i'll start poking
13:47
<Gadi>
comfrey: does this return anything at the thin client shell: lspci -n |grep 8086:27ae
13:49cliebow has joined #ltsp
13:55
<Lns>
ogra: When thermal paper comes down in cost, this might be a better "e-ink" ;) http://en.wikipedia.org/wiki/Thermal_printer
13:57* Gadi keeps thinking of saturday night live and german humor whenever ogra makes a funny ;)
14:00alkisg has quit IRC
14:00alkisg has joined #ltsp
14:02
<cliebow>
all i can think of is nat. Lampoons vacation..
14:02
in Germany..
14:02
"Who the hell were they??
14:05prpplague has joined #ltsp
14:06carbm1_ has quit IRC
14:09
<prpplague>
any recommendations on extremely cheap thin clients with minimal specs?
14:12
<Appiah>
what are the minimal specs?
14:13
<sbalneav>
prpplague: what means "extremely cheap"? :)
14:13
Sub $500? sub $400? Sub $100?
14:14shawnp0wers has quit IRC
14:14
<prpplague>
sub $300, 64MB ram, 64MB flash
14:15
<comfrey>
prpplague: arm or i386?
14:15
<prpplague>
comfrey: either is fine
14:15
<Gadi>
cherry or blueberry?
14:15
<prpplague>
snausberry
14:15
<comfrey>
the little hp boxes look interesting.
14:15* comfrey looks for link
14:16
<Gadi>
why are you looking for flash?
14:16vagrantc has joined #ltsp
14:16* Gadi waves to vagrantc
14:16* vagrantc waves
14:17cliebow has quit IRC
14:17
<prpplague>
Gadi: not going to use true ltsp style thin clients
14:17
<comfrey>
http://www.retrevo.com/s/HP-t5145-Desktops-review-manual/id/23192ag120/t/1-2/
14:17cliebow has joined #ltsp
14:17* prpplague looks
14:17
<prpplague>
comfrey: yea i was looking at some of the devon it stuff, pretty darn good pricing there
14:18
<Gadi>
dont forget ebay
14:18
<comfrey>
prpplague: yeah, thats what i am running here
14:18
devonit boxes
14:18
<prpplague>
Gadi: yea well i can't exactly purchase 2500 from ebay
14:18
<comfrey>
just via mini-itx boxes.
14:18
<Gadi>
prpplague: youd be surprised
14:18
<comfrey>
vagrantc: what was the thin client you mentioned
14:18
the india thing
14:19
<vagrantc>
comfrey: based on the sheevaplug
14:19
<comfrey>
right
14:20
<vagrantc>
the sheevaplug is headless, although with usb video you *might* make a nice little thin client out of it. but supposedly there are some newer designs that also have VGA (or some other video output)
14:20
<prpplague>
vagrantc: yea i'm familar with the sheeva, hehe
14:21
<vagrantc>
it's what i used to test the ltsp-build-client cross-architecture support
14:21
well, that an qemu
14:22
<prpplague>
vagrantc: oh? ltsp has cross arch builds these days?
14:22
<vagrantc>
prpplague: yes, using qemu-user-static packages.
14:22
<comfrey>
so it looks kike i messed up something good....
14:22
<prpplague>
interesting
14:23
<comfrey>
now it is not making it through init and i an getting can't open /tmp/net-eth0.conf
14:23
<vagrantc>
prpplague: only works for armel, in my experience.
14:23
<comfrey>
also seeing "sleep" invalid number '0.1'
14:24* vagrantc disappears to a meeting
14:24
<prpplague>
vagrantc: well it's nice to see someone actually moved arm support into mainstream ltsp
14:24* prpplague 's first ARM ltsp install was back in 2001
14:26* vagrantc is hoping to get access to some arm and mips hardware to test ltsp
14:29
<Lns>
I've had my eye on these, can't beat $202 for a price... Might order one today in fact
14:29
http://www.mini-box.com/Mini-Box-M200
14:29GodFather_ has joined #ltsp
14:29
<prpplague>
vagrantc: any particular ones?
14:30
<vagrantc>
prpplague: something reasobly well supported by debian's kernel with video(or built-in lcd)/mouse/keyboard
14:30
<alkisg>
comfrey: where's that sleep problem? lenny initramfs?
14:31prpplague is now known as prpplague_afk
14:31
<prpplague_afk>
vagrantc: i'll have a look around the office
14:31cliebow has quit IRC
14:33
<comfrey>
alkisg: yup looks like it
14:33
it popped up out of the blue. i must have touched something...
14:34
it cant find the /tmp/net-eth0.conf
14:34
init line 174:
14:34
<alkisg>
That might be a "module not found" problem
14:35
<comfrey>
i see, so i need to update initrd
14:35
and specify the modules
14:36prpplague_afk is now known as prpplague
14:36
<prpplague>
vagrantc: not tried the beagleboard with ltsp?
14:36pmatulis has quit IRC
14:39
<Lns>
Atom 2x1.6GHz + 1GB RAM should be great for localapps, right?
14:45
Also, is there a hook for env. variables to set themselves when you chroot to the client filesystem? Probably not...is there any reason not to provide some sort of script (ltsp-update-chroot ?) that automatically does stuff for you, such as LTSP_HANDLEDAEMONS = False and mounting /proc, etc.. while updating ? I just updated my Lucid chroot and it tried restarting cron..and failed on some other things cuz I forgot to do all of the other steps befo
14:45
rehand.
14:45mikkel has quit IRC
14:45
<_UsUrPeR_>
Lns: that sounds like a fine machine :)
14:45
shoot, 1x1.6GHz is quite adequate for localapps
14:47
<Lns>
_UsUrPeR_: Right on =) I'll probably be testing one of those guys out real soon with LTSP to see how it handles the load..need to see what vid. chipset it's got on it
14:47
<_UsUrPeR_>
Lns: can you name the unit?
14:47
<Lns>
? It's the link I posted above
14:47* _UsUrPeR_ assumes it's an atom processor
14:49
<Lns>
I just ran into another update issue on lucid.. while updating packages, dbus fails with "The system user 'messagebus' already exists. Exiting. -- Failed to open connection to "system" message bus: Failed to connect to socket /var/run/dbus/system_bus_socket: No such file or directory" .. then consolekit fails, and it's a domino effect after that. Any clues? Did I bork it?
14:49
(this is the chroot update btw)
14:52carbm1_ has joined #ltsp
14:57thunsucker has quit IRC
15:10artista_frustrad has joined #ltsp
15:15
<vmlintu>
Lns: Atom 330 2x1.6GHz + 2GB RAM has been working nicely for fat clients also
15:17
<vagrantc>
prpplague: nope. only real hardware i've tried was a sheevaplug
15:18
<Lns>
vmlintu: that would be SO nice for media heavy workstations. Man I can't wait till next school year when I roll out 10.04 =) I'm hoping some of the TCs they have will be OK (maybe a ram upgrade) for at least SOME multimedia stuff
15:20mgariepy has quit IRC
15:21
<vmlintu>
Lns: I wouldn't do video editing with atom 330, but the newer atoms probably can handle that too..
15:21
<prpplague>
vagrantc: i'll see what kind of spare boards i have around that might be useful
15:27AndyGraybeal has joined #ltsp
15:29carbm1_ has quit IRC
15:31artista_frustrad has quit IRC
15:31vvinet has quit IRC
15:31ogra_cmpc has quit IRC
15:32vvinet has joined #ltsp
15:32ogra_cmpc has joined #ltsp
15:32artista_frustrad has joined #ltsp
15:38
<AndyGraybeal>
okay.. so i got a question, and i'm a newb, so go easy on me... i'm looking at the 1500 series thin client on disklessworkstation.com -- it's got dualheads, i'm wondering if i can run dual-seated terminals on it... in otherwords -- can i run two monitors and two keyboards and mice off that station seperately? does that make sense? i'm sorry if this is an ignorant question.
15:45
<vagrantc>
AndyGraybeal: it's probably technically possible, though i don't know if it woudl be easy or not.
15:45
i think the term for such a setup is called "multi-seat"
15:47khris4 has joined #ltsp
15:47
<khris4>
hello
15:49
<alkisg>
AndyGraybeal: I've done that with my laptop. It isn't easy, and the new gdm made it even harder. And to do that in an LTSP terminal would be even more harder, so I don't think you'd want to even try it ;)
15:49alexbodn has quit IRC
15:50
<vagrantc>
it's an interesting model, for sure, as modern hardware is way overkill for a tradition thin client
15:51
<alkisg>
I think I had tried this one: http://wiki.c3sl.ufpr.br/multiseat/index.php/Mdm
15:53
So... a networking question: would it be possible for nbd-client to send (with nc) which nbd image it wants, and for nbdrootd to receive it (again with nc) and call nbd-server with the appropriate parameter? This way we could use the same NBD_PORT for all images...
15:54
*maybe nbdrootd would use stdin, not nc, as it runs from inetd.
16:01
<Lns>
AndyGraybeal: Using a thin client as a multi-seat system, ...at the least, you'll probably want to have Gbit networking going to it from the server =)
16:02
I've been wanting to do that with my home workstation so my wife and I can be on it at the same time =)
16:02
but that's a much simpler setup
16:02Lns has quit IRC
16:04
<AndyGraybeal>
carbm1, alkisg, lns .. thank you for your responses.. sorry i had to renew my IP address for some strange reason and didn't se your responses.
16:04
until now i mean.
16:04abeehc has quit IRC
16:07
<AndyGraybeal>
oh vagrantc, just so you know, i realiz that multi-seat is the correct term.. recently in documentation and forums.. i've been reading that if it's only two seats.. they call it dual-seat.. i just thought that was cute and more descriptive... but yes, multi-seat :)
16:09
so aside from the multiseated thinclient idea, i'm wondering what is the best documenation to read for setting this up with CentOS? i work at a restaurant, and we need atleast 11 workstations but we can't really afford nice shiny new equipment. this is the first time that i've actually looked into thin clients... and seeing that they go for $350 is a relief.
16:10
is there a way i can run virtual machines inside ltsp?
16:10
i'm sorry - i don't know the foundation of ltsp.. so this might be a really dump questin
16:11
everyone that works here is used to windows, so i am wondering if i can run something lke qemu/kvm with windows on top to ween them off of windows and head towards linux eventually.. when they can handle it.
16:14
<vagrantc>
you can connect to windows terminal services via RDP using rdesktop, and ltsp has some built-in support for that
16:15
<AndyGraybeal>
ah interesting.
16:16
thank you that's awesome.
16:16
<vagrantc>
LTSP does tend to bring up terms such as "awesome" now and again. :)
16:16
<AndyGraybeal>
:)
16:16
i'm sorry if it's non-formal.
16:16vvinet has quit IRC
16:17abeehc has joined #ltsp
16:17
<vagrantc>
no need to be sorry... this is suppose to be fun, not formal. :)
16:18
<AndyGraybeal>
thank you :)
16:20
<Gadi>
vagrantc: please pass the Grey Poupon
16:20* Gadi forgot to roll down his window
16:21rjune has quit IRC
16:26johnny has left #ltsp
16:28GodFather_ has quit IRC
16:29
<alkisg>
For my local setups, I'm thinking of diverting /etc/init/cron.conf, to completely disable cron on thin/fat clients (to get rid of apt-xapian-index, logrotate, mlocate etc). Should I expect any bad side effects from doing so?
16:31
<AndyGraybeal>
since i'm new to this, what kinda hardware do you guys use?
16:33alexqwesa has quit IRC
16:33alexqwesa has joined #ltsp
16:33
<AndyGraybeal>
http://www.disklessworkstations.com/200118.html <-- this thing looks great
16:34
please tell me what you all thinka bout this - is it just a newb reaction that this client looks good?
16:35
<khris4>
anyone have issue with remote terminal service on unbuntu? I keep getting a message "peer reset connection" and then the windows close every few minutes
16:40* alkisg also wonders if we should be running mksquashfs with `nice` in front of it :)
16:41
<AndyGraybeal>
is there documentatin for setting up LTSP on CentOS?
16:43
<vagrantc>
AndyGraybeal: your closest bet for anything recent (i.e. supported) woudl be fedora.
16:45
<AndyGraybeal>
ah fedora, thank you
16:46
<vagrantc>
i've heard of people setting up an ubuntu LTSP chroot on a centos server...
16:46
something like that might be another way to go
16:46
<AndyGraybeal>
ah interesting, 'm still confused about chroot
16:49khris4 has quit IRC
16:53
<AndyGraybeal>
something like a virtual server?
16:55alkisg has quit IRC
17:01ogra_cmpc has quit IRC
17:01
<vagrantc>
AndyGraybeal: in the case of LTSP, the chroot is just the environment where the root filesystem for the thin-clients is built and maintained, and is what's exported over the network for the thin-clients.
17:01ogra_cmpc has joined #ltsp
17:32Gadi has left #ltsp
17:32alexbodn has joined #ltsp
17:41vagrantc has quit IRC
17:54prpplague is now known as prpplague_afk
18:01
<comfrey>
AndyGraybeal: a bit spendy
18:18nokia3510 has quit IRC
18:19nokia3510 has joined #ltsp
18:36ola has joined #ltsp
18:48mordocai has joined #ltsp
18:49mordocai has quit IRC
18:49johnny has joined #ltsp
18:50mordocai has joined #ltsp
18:53shogunx has quit IRC
18:55shogunx has joined #ltsp
18:55Faithful has joined #ltsp
18:55litlebuda has joined #ltsp
19:10litlebuda has quit IRC
19:11ogra_cmpc has quit IRC
19:11litlebuda has joined #ltsp
19:13ogra_cmpc has joined #ltsp
19:16Faithful1 has joined #ltsp
19:35
<comfrey>
open rd looks neat, but a bit of overkill for a thin client
19:36
http://www.globalscaletechnologies.com/p-20-openrd-base.aspx
19:37litlebuda has quit IRC
19:47
<comfrey>
i guess the guruplug is not out till april
19:47
http://www.globalscaletechnologies.com/p-33-guruplug-display.aspx
19:47Sarten-X has quit IRC
19:48Sarten-X has joined #ltsp
19:52mushroomblue has quit IRC
19:55Faithful1 has quit IRC
19:56mordocai has quit IRC
20:11johnny has left #ltsp
20:49vagrantc has joined #ltsp
20:54johnny has joined #ltsp
21:08vagrantc has quit IRC
21:44artista_frustrad has quit IRC
22:41
<map7>
I'm having trouble with a thin client which uses the Intel GMA950 graphics driver, is there anyway to add this module 'intelfb' to the client image?
22:43
I already have package 'xorg-video-intel' installed, but that doesn't seem to include the intelfb driver
22:43
<sbalneav>
Which package does? Just install that in the chroot.
22:43
<map7>
I don't know which package does
22:43
that's the problem
22:45
I know that if I boot off an Ubuntu 9.10 32bit Live CD the driver works, but if I boot off my client image (Ubuntu 9.10 32bit as well) it doesn't work unless I use the VESA driver
22:45
but the VESA driver does not do enough for my liking
22:47
<sbalneav>
Don't know myself.
22:47
<map7>
Maybe it's not in a package I might have to compile the module myself would that be possible?
22:47
or would it most likely be in a package?
22:48
<sbalneav>
Not sure.
22:55
<map7>
I might try this, 1. upgrading to the latest kernel 2.6.32-8, 2. install all packages which the live cd has into my client image, 3. upgrade to ltsp 5.2 and upgrade my image
22:59
<johnny>
why do you want fb?
22:59
that's not related to using vesa in X
22:59
or do you mean the vesafb on the console?
22:59
/me pokes sbalneav
22:59
sup :)
23:00
<map7>
I would like full 3d support and high resolution (1920x1080) support for the Intel GMA950 chipset on my thin client
23:00
when I load the ubuntu live cd and do a 'lspci -v' I see it's using the kernel module intelfb
23:01
I must not have this on my client image as I cannot set it to use intelfb and have to set it to VESA to get any gui happening on my thin client
23:01
<johnny>
tha'ts not related to intefb
23:01
intelfb is only for console
23:01
not X
23:01
doesn't do 3d
23:01
the drm driver is what you need
23:02
not that i can offer any more information than that..
23:02
but i can definitely tell you intelfb isn't gonna be helping
23:02
i'm suprised ubuntu ever shipped that
23:03
<map7>
thank you
23:03
Doesn't the Intel GMA950 use an open source driver?
23:08
johnny would upgrading to LTSP 5.2 and rebuilding my image help at all?
23:15
<johnny>
no.. ltsp 5.2 wouldn't help at all
23:15
i don't know anything about the 950
23:16
is that the same as poulsbo ?
23:16
no
23:16
that's the 500..
23:17
the kernel you get in in the chroot should be good enough to do the right thing
23:17
the kernel package should be the exact same between your server and client
23:17
so.. you should have the intel drm drivers