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


Channel log from 15 February 2009   (all times are UTC)

00:01head_ has quit IRC
00:30alkisg_ has quit IRC
00:58alkisg has joined #ltsp
00:58japerry has quit IRC
01:00ball has quit IRC
02:12DarKnesS_WolF has joined #ltsp
02:12
<DarKnesS_WolF>
good morning
02:12
nubae2: i am testing ur fat client script right now on ubuntu 8.10 :-) one question.. how big the highfat image will be ?
02:23vagrantc has quit IRC
02:41
<nubae2>
morning
02:41
DarKnesS_WolF: about 1 gig
02:42
takes a while to build too
02:42
<DarKnesS_WolF>
nubae2: okay cool :-)
02:42
yes
02:42
i am building it but i had to remove edubuntu stuff from teh script i don't need them right now
02:42ogra has quit IRC
02:43
<DarKnesS_WolF>
nubae2: right now looking into getting pxeboot.0 to be a small boot menu so i can chose should i go thin client or fatthin client
02:43
<nubae2>
yeah no problem, the idea is to choose only the things u want... lke u can make a kde or xfce version easily too
02:44
ah right... thats definetly here: help.ubuntu.com/Community/UbuntuLTSP
02:44
<DarKnesS_WolF>
nubae2: but i like the plugin so flexable
02:44
<nubae2>
it will be much better next version, everything visual... so u can pick and choose packages with mouse
02:45ogra has joined #ltsp
02:46
<nubae2>
ogra: u have an idea what the edubuntu DVD is going to be filled with?
02:46
wbat will make up the gigs?
02:46
<DarKnesS_WolF>
nubae2: great .
02:47
nubae2: but now if i did install new application " using the root acount of the ltsp image " i'll need to rebuild the image ? ? don't think so right ?
02:47
<nubae2>
if u use nbd, I'm afraid yes
02:47
but u can keep a copy of packages using apt-cacher
02:48
there s something about it also in the help docs, which imo are some of the best help files available for any project...
02:50
<DarKnesS_WolF>
nubae2: okay will check but i find that strange i mean if i am accessing the ltsp image as root and installting anything it should be installed in teh chroot /opt/ltsp/fati64 or whatever why then i need to rebuild the image ?
02:56
<nubae2>
because u are loading an image and not the directory
02:57
if you were to use nfs, it would load directly the directory, and with nbd, it is like it makes a cd image for u
02:57
btu across the network... the result is that this loads much faster
02:57
and is more secure
02:59bobby__C has joined #ltsp
03:03
<DarKnesS_WolF>
nubae2: ok got the idea ..i only need to update the image right ?
03:04
ltsp-update-image
03:04
right ?
03:05
<Ryan52>
stgraber, sorry, I didn't get to it today...tomorrow morning I'll try to.
03:28
<ogra>
nubae2, not really, likely what is on the ubuntu DVD plus edubuntu-desktop
03:28
stgraber, are you aware that your recent change breaks nfs on ubuntu ?
03:29
stgraber, lts.conf needs to exist (and need to be in the bind mounted files) on nfs setups
03:29
(unless something massively changed i'm not aware of
03:29
)
03:48ogra_ has joined #ltsp
03:52ogra has quit IRC
03:53ogra_ is now known as ogra
03:59artista_frustrad has quit IRC
04:26nubae2 has left #ltsp
04:30philippe__ has joined #ltsp
04:33philippe_ has quit IRC
04:37bobby__C has quit IRC
04:38pasmen_ has quit IRC
04:39nubae2 has joined #ltsp
04:40Remaille has joined #ltsp
05:00pasmen_ has joined #ltsp
05:04
<DarKnesS_WolF>
mmm
05:04
nubae2: something not right about the NFS home dir.
05:29
<Remaille>
hi
05:29
on my thin clients, il would like ton use windows too (we have sift than can't be run with linux)
05:30
I have heard that it was possible on another tty
05:30artista_frustrad has joined #ltsp
05:30
<Remaille>
we have a win 2003 server
05:30
how to do that ? do you have any links ?
05:30
thanks
05:54
<knipwim>
johnny: still working on the gentoo port?
05:57pasmen has quit IRC
06:05artista_frustrad has quit IRC
06:05
<DarKnesS_WolF>
i have question can i run multi. images on the same LTSP server ? i am always getting kernel panics on one of them and also when i do update the image it is only updateing one image ..
06:11alkisg has quit IRC
06:14nubae2 has left #ltsp
06:22DarKnesS_WolF has quit IRC
06:28Egyptian[Home] has quit IRC
06:28DarKnesS_WolF has joined #ltsp
06:31artista_frustrad has joined #ltsp
06:35shogunx has quit IRC
06:42Egyptian[Home] has joined #ltsp
07:06artista-frustrad has joined #ltsp
07:17artista_frustrad has quit IRC
07:52DarKnesS_WolF has quit IRC
07:54Remaille has quit IRC
08:00pmatulis has joined #ltsp
08:09shogunx has joined #ltsp
08:24hanthana has quit IRC
08:24hanthana_ has joined #ltsp
08:32shogunx has quit IRC
08:33shogunx has joined #ltsp
08:35F-GT has quit IRC
08:37F-GT has joined #ltsp
08:40shogunx has joined #ltsp
08:42shogunx has quit IRC
09:04shogunx has joined #ltsp
09:44alkisg has joined #ltsp
09:51DarKnesS_WolF has joined #ltsp
09:55
<DarKnesS_WolF>
anyone did multi OS on the same LTSP server ? i have problem with SSH key if the machine loged to any img. 2nd time it get connection refused and i have to upgrade the sshkeys am i missing anything ?
10:02
<johnny>
knipwim, yes
10:07
<knipwim>
i was wondering what's the way to go for the ltsp installation
10:07
install ltsp-server from git and do ltsp-build-client?
10:16hanthana_ is now known as hanthana
10:17alekibango has quit IRC
10:19
<johnny>
knipwim, basically
10:20
<knipwim>
so no need to emerge ltsp-client?
10:20
<johnny>
altho there's one modification to the genkernel script needed
10:20
no.. please don't
10:21
<knipwim>
what's the modification? the /usr/share/genkernel/x86/modules_load one?
10:22
<johnny>
yes
10:22
the genkernel initrd stuff needs to be rewritten or somethin
10:23
so it doesn't manually load everything..
10:23
and uses more detection
10:24DarKnesS_WolF has quit IRC
10:27alkisg has quit IRC
10:28pmatulis has quit IRC
10:32makghosh has joined #ltsp
10:46japerry has joined #ltsp
10:46johnny has left #ltsp
10:52Gadi has joined #ltsp
10:54hanthana has quit IRC
10:54rjune_ has quit IRC
10:54rjune_ has joined #ltsp
10:54johnny has joined #ltsp
10:56hanthana has joined #ltsp
11:13ilaiho has joined #ltsp
11:13
<ilaiho>
hi!
11:13
could someone explain what is that "LTSP thick client" thing?
11:15
<stgraber>
ogra: why would not having a lts.conf file break NFS ? the current lts.conf however displayed an ugly error message at every boot ...
11:16
<ilaiho>
does it use the local hard drive as cache?
11:17
<stgraber>
ogra: oh, I guess I understand why it'd break, I'll see what I can put in /etc/lts.conf that won't make the parser to complain :)
11:19p4ch0 has joined #ltsp
11:21p4ch0 has quit IRC
11:25
<johnny>
ilaiho, no.. it means that it runs the programs locally
11:25
using the client's ram and cpu
11:26
even if it gets the programs themselves from the netowrk
11:28
<stgraber>
ogra: fixed it, that way you get the previous file and also a valid lts.conf
11:32vagrantc has joined #ltsp
11:35* vagrantc dances the debian stable release dance
11:41
<Gadi>
stgraber: maybe you can put the k12linux default lts.conf there
11:41
thats already in upstream
11:42
<vagrantc>
Gadi: so, the main problem with the local ltspfs mounting is that it doesn't play well with cdpinger
11:42
<Gadi>
vagrantc: do you know why?
11:42
<vagrantc>
Gadi: nope.
11:43
<stgraber>
Gadi: ./server/configs/k12linux/lts.conf ?
11:43
<Gadi>
cdpinger *should* just call the ltspfs_entry script
11:43
<vagrantc>
but the local mount doesn't respond to insertion/removal events in a predictable fashion, and even interferes with insertion/removal events on the remote ltspfs mounts for CDs
11:43
Gadi: it *does*
11:43
<Gadi>
stgraber: yeah, I guess - it would be nice for ubuntu to have a default lts.conf with comments
11:44
<stgraber>
Gadi: well, I don't really want my lts.conf to do anything useful :) we have a commented one in /usr/share/doc/ anyway
11:44
<Gadi>
stgraber: no one looks in /usr/share/doc (a) and (b) newbies are quite confused as to the syntax of lts.conf
11:45
it would cut down on lots of questions to simply have a doc to start with
11:45
*conf file
11:46
unless you want to create a man page for lts.conf :P
11:46
<stgraber>
Gadi: man isn't installed in the chroot so wouldn't be that useful :)
11:46
<Gadi>
for the server
11:47
<stgraber>
Gadi: for now I'll go with that empty lts.conf and make sure lts-parameters and the example lts.conf are present on the server. If I have to set a variable to a different value to what we've in LTSP, then I'll probably put it in lts.conf
11:48
<sepski>
are there anything i can do in the chroot to lean up the thin clients ? i have tried 2 different clients now. they both start to lag after a while. I notice this as very varying pingtime between the server and the tk. often between 30-50 ms, pinging from the server to a pc/printer/any other device in the same switch as the tk, gives a rock steady 0.55 ms.
11:48
<stgraber>
Gadi: though current lts.conf points the user to /usr/share/doc on the server so as long as the user actually reads the lts.conf it'll find the doc
11:48
*he'll
11:49
hmm, it actually points the user to the in-chroot doc ... I'll need to make sure I haven't moved it to the server :)
11:49* Gadi wonders how many users stgraber has talked to recently :)
11:49
<Gadi>
users cant read
11:49
its a universal fact of nature
11:51
<vagrantc>
so then they can't read the configuration file either, no? :P
11:51
need to give them a GUI button that says "
11:51
"GO"
11:53sepone has joined #ltsp
11:53
<Gadi>
well, it seems like we have the users going on a scavenger hunt to find info
11:53sepski has quit IRC
11:53sepski has joined #ltsp
11:53
<Gadi>
when we could just as easily give them stuff to uncomment
11:53
:)
11:54
<stgraber>
Gadi: well, the main issue there is that lts.conf in the chroot shouldn't be used
11:54
Gadi: so if you give them things to uncomment they'll just do the wrong thing :)
11:55
<Gadi>
well, shouldn't be used for NBD
11:55
<stgraber>
and NFS
11:55
<Gadi>
you tftp for NFS now, too?
11:55
<stgraber>
as far as I understood, the only reason why it's there is so we can bind-mount on it
11:55
then NFS will behave the same as NBD
11:55
with it being overwritten by tftp
11:56
<Gadi>
last I checked, you only tftp in ltsp-nbd
11:56
but I checked a long time ago
11:56
:)
11:57
<stgraber>
Gadi: just check the code, it's actually what we're doing at the moment
11:57* stgraber is confused by ogra's comment then :)
11:57
<Gadi>
what is?
11:57
you are tftp'ing for NFS and NBD?
11:57
or just NBD?
11:57
<stgraber>
Gadi: downloading lts.conf is done in ltsp_nbd
11:57
<johnny>
if not.. you can.. and that would be nice..
11:57pasmen_ has quit IRC
11:58
<stgraber>
johnny: yeah, I'd have to make sure the bind mounting is in place at this point first
11:58pasmen has joined #ltsp
11:59
<Gadi>
well, as it seems this is all unclear to the developers, whatever we wind up doing should be nicely commented in lts.conf
11:59
:)
11:59
<johnny>
it'd be nice if we could all stop bind mounting.. but there's still no perfect solution
11:59
or even close to it..
11:59sepski has quit IRC
11:59sepski has joined #ltsp
11:59
<Gadi>
and if etc/lts.conf remains minimal for bindmounting only, then there should be a l-b-c plugin to generate a nicely commented default tftpdir/lts.conf
11:59
<sepski>
let's see if ndswap helps any :s
12:00
<stgraber>
johnny: the perfect solution would be aufs on top of NFS but last I checked it triggers a kernel panic
12:00
<johnny>
Gadi, and then there will be too many options :)
12:01
<Gadi>
what do you mean?
12:01
<johnny>
stgraber, aufs will not make it to kernel
12:01
that is the real problem :(
12:01
Gadi, too many options in lts.conf .. users will not like it :)
12:01
<stgraber>
johnny: right but that's what I'm using anyway for nbd :)
12:02* Gadi has never heard a user complain about too many options in a conf file
12:02
<johnny>
i have..
12:02
<Gadi>
*never*
12:02
unless the file is confusing
12:02
<johnny>
have trouble finding the necessary options
12:02
<Gadi>
smb.conf is huge by default
12:02
<johnny>
yeah.. i hate that file
12:02
<stgraber>
Gadi: all users who once read squid.conf complained about it containing too many options :)
12:02
<Gadi>
that's why there's a seach feature
12:02
*search
12:03
well, warren's k12linux lts.conf file is nicely commented
12:03
and the k12linux community has never complained of too many options
12:04
more users are frustrated when their conf file does not work, because of a missing [default]
12:04
:)
12:04
<stgraber>
Gadi: warren's lts.conf doesn't contain half of the available options :)
12:05
<Gadi>
fine. its better than none
12:05
dear newbie, I want you to guess as to the lts.conf file format and options
12:05
:)
12:06
oh, and location
12:06
<stgraber>
yeah, well, I guess we'll first fix the "Where's lts.conf" question, then talk about its content :)
12:06
<Gadi>
since there is no default file
12:14
<vagrantc>
the NFS implementation on debian does not download lts.conf from TFTP, and i really have little interest in making it do so.
12:14
<johnny>
vagrantc, even if we all asked nicely?
12:15
<vagrantc>
johnny: you're going to have to do more than ask nicely, yes.
12:15
though probably making a symlink in the tftp dir would work...
12:15
thta way the location appears the same, even if it isn't actually downloaded via tftp
12:16xjwprygzc has joined #ltsp
12:16|Paradox| has quit IRC
12:16xjwprygzc is now known as |Paradox|
12:17
<vagrantc>
but the symlink wouldn't work well if they actually need to download via tftp, as tftp servers may not support symlinks outside of the tftp server's "chroot"
12:19
<Gadi>
we could always have a tftp thing in the init scripts
12:19
<vagrantc>
it just seems hairbrained to download something via tftp just for consistancy.
12:19
there's no technical reason to download lts.conf using tftp for NFS implementations.
12:20
<Gadi>
indeed
12:20
<vagrantc>
in fact, there's a technical argument against it- it adds more code.
12:20
<Gadi>
and more network traffic
12:20
:)
12:20
<vagrantc>
think of the bytes!
12:20
<Gadi>
poor lil bytes
12:21
<vagrantc>
we'll have to set up code to fail on anything less than gigabit networks!
12:24
<stgraber>
ok, so I'll just keep that lts.conf as it's and maybe ship a default /var/lib/tftpboot/ltsp/i386/lts.conf. Newbies aren't supposed to do NFS anyway as it's no the default setup and will require more changes than just editing lts.conf in the chroot anyway
12:30
<Gadi>
sounds good to me
12:30
that will also close the bug I filed against Ubuntu :)
12:31* Gadi runs away from stgraber
12:34
<stgraber>
:)
12:35
good, I now get a clean boot when usplash is disabled :) (well, except nbd but the reconnect is handled by the kernel so can't silence this one ...)
12:37
next on my list is checking why ipconfig is so slow ... it currently takes 6s to get the IP and then the system boots in 5s :)
12:37
<Gadi>
wow
12:37
5s?
12:37
<stgraber>
well, my thin client is a Core2Duo, it helps :)
12:37
<Gadi>
is this a VM?
12:37
oh
12:37
hehe
12:38
btw, did you sort out the issue with that geode?
12:38
<stgraber>
nope, it's still using the gateway as tftp server ... I guess I'll just redirect the gateway port to the tftp server :)
12:38
<Gadi>
sounds like a config issue
12:39
<stgraber>
maybe it needs another dhcp option than next-server
12:39
<Gadi>
shouldnt be hardware-related
12:39
<stgraber>
well, I don't know I have another geode based thin client with the exact same config on that network and it boots just fine :)
12:39
<Gadi>
unless you are tempted to blame the PXE code in that unit
12:40
weird
12:40
but it booted up fine with previous ltsp versions, right?
12:40
<stgraber>
don't know, I found it at the office on thursday :) we couldn't make it to boot at the office and at one of our customers
12:41
so I brought it back home as I can more easily tweak the network to debug that kind of issue
12:41
<Gadi>
gotcha
12:41
let me know what you find
12:41
I know I was having issues with the new geode driver
12:41
(or maybe with the new xserver who knows)
12:42
btw: another qu for you:
12:42
<stgraber>
hmm, we have a few hundreds of Geode on Intrepid and they usually work fine
12:42
<Gadi>
have you tested X_VIRTUAL recently?
12:42
<stgraber>
though you must disabled usplash
12:42
and use the hack.d script in upstream ltsp
12:42
(basically force the geode driver and XAA instead of EXA)
12:42
<Gadi>
right
12:43
<stgraber>
nope, haven't used VIRTUAL for a while, mainly because except my laptop we don't have any dual-head capable thin clients ...
12:43
I'm supposed to get a few EEE box this week that should be able to do VGA and DVI with an intel video card
12:43
<Gadi>
have you tested X_VIRTUAL? I was working with _UsUrPeR_ a couple weeks back, and it looked like we may need to change the hack from a "screen_hack" to a "dsiplay_hack" for it to work
12:43
<stgraber>
so will be able too test with them
12:43
<Gadi>
*display
12:44
<stgraber>
hmm, why changing it to display ?
12:44
<Gadi>
that adds a Display subsection to SCrren
12:44
*Screen
12:44
and puts it in there
12:44
<stgraber>
ah, yes
12:44
<Gadi>
seems thats what xorg likes
12:44
but, I remember the current version working for you
12:45
so, I didnt want to push a change upstream
12:45
<stgraber>
yeah, both work for me
12:45
<Gadi>
ok, so maybe we should change it.
12:45
<stgraber>
I just checked and I have a SubSection display on the installed Ubuntu I have on that laptop
12:46
http://ubuntu.pastebin.com/f2788e6d5
12:46
I currently have that on my laptop
12:46
<Gadi>
ok, lemme change it then upstream
12:48
<stgraber>
ok, just tried dual-head with that config (I believe it's what Ubuntu auto-generated) and it works fine (using it now)
12:49
<vagrantc>
stgraber: i see ldm 2.0.34 uploaded to ubuntu, but no commits or tags on ldm-trunk...
12:49
<stgraber>
vagrantc: oh, wait a sec
12:49
doh, works better when pushed :)
12:50
I guess I tagged this one offline and forgot to push it afterwards ... sorry for that
12:50aglio2 has joined #ltsp
13:02
<ltsppbot>
"vagrantc" pasted "set volume defaults based on volume if unset" (19 lines) at http://pastebot.ltsp.org/249
13:02
<vagrantc>
anyone see an issue with that ^^
13:03
basically allows setting a relative default for all channels ... although some of those channels play off one-another ...
13:03
<stgraber>
I have no problem with that
13:04
Gadi: does that change from screen_hacks to display_hacks will create a subsection display in screen or move it to the display section ?
13:05
<vagrantc>
display_hacks is a subsection of screen, so a screen section will be generated if display_hacks is set
13:05
<Gadi>
stgraber: creates a subsection Display and puts Virtual in the subsection
13:05
<stgraber>
good
13:06
<Gadi>
vagrantc: the only issue I see is that VOLUME=PCM_VOLUME=100 will often saturate the speakers
13:07
you almost never want PCM_VOLUME=100
13:07
I think
13:07
perhaps set all of them except PCM_VOLUME to VOLUME by default
13:08
and keep PCM_VOLUME at a fixed 90
13:08
that would handle more cases well, I think
13:09
<vagrantc>
Gadi: that's a problem if they set VOLUME=100 and don't set PCM_VOLUME ... is that what you're saying?
13:09
<Gadi>
right
13:09
which I would expect to be common
13:10
<vagrantc>
i was thinking there might be some issue with PCM ...
13:10
<Gadi>
lets max out the master volume
13:10
I have a script somewhere that actually intelligently finds the channels
13:10
<vagrantc>
i honestly have very little idea what all the channels do, and how they relate to one another
13:10
<Gadi>
maybe I can paste that for you to pick at
13:10
:)
13:11
<stgraber>
Gadi: I'd love that :) we've tons of weird sound cards with non-existing channels and similar stuff
13:11
<vagrantc>
well, i should head out to my local debian release party soonish :)
13:12
<johnny>
release??
13:13
how soon?
13:13
<ltsppbot>
"Gadi" pasted "amixer channel volume set script" (13 lines) at http://pastebot.ltsp.org/250
13:13
<johnny>
lol
13:13
<Gadi>
I wrote this, but don't use it in production yet
13:13
stgraber: feel free to play
13:13
:)
13:14
<vagrantc>
johnny: yesterday?
13:14
<Gadi>
stgraber: I think I wrote it, trying to find something to address VIA and their weird channels
13:15
<johnny>
missed it :)
13:16
<vagrantc>
well, the party is today, the release happened yesterday ... though which exact day depends on your timezone.
13:18
<stgraber>
Gadi: thanks, I'll test it with some of our broken systems :) (VIA and SIS mainly)
13:19
<Gadi>
I was thinking, it may make sense to deal with channels like xrnadr ouputs
13:19
*xrandr
13:19
and have a:
13:20
AUDIO_CHANNEL_01 = "Master"
13:20
AUDIO_CHANNEL_VOLUME_01 = 100
13:20
etc
13:20
(making VOLUME an alias
13:20
or some such for backward compat)
13:20
then the script I pasted can easily set the volume if the channel matches
13:21
and you have limitless flexibility to name channels
13:26
ah, to be quoted in a changelog... is there any quicker route to immortality?
13:30hanthana has quit IRC
13:30
<vagrantc>
:)
13:40
<stgraber>
Gadi: writing your name anywhere on internet is probably faster :) then google will take care of the immortality part
13:42pasmen has quit IRC
13:49bobby__C has joined #ltsp
13:51
<mistik1>
good day #ltsp
14:00vagrantc has quit IRC
14:03
<sepski>
my klient still lagg down after a while's usage nbdswap is active but unused. how can i troubleshoot
14:05japerry has quit IRC
14:11
<johnny>
why do launchpad feeds include newlines in their defintions.. weird
14:11
or some other odd character
14:40agike has joined #ltsp
14:49alkisg has joined #ltsp
14:58shogunx has quit IRC
15:03ace_suares has quit IRC
15:09mistik1_ has joined #ltsp
15:13sistemas has joined #ltsp
15:18
<sistemas>
hi everyone
15:18
i wanna know about local apps
15:20mistik1 has quit IRC
15:20mistik1_ is now known as mistik1
15:36sistemas_ has joined #ltsp
15:36sistemas has quit IRC
15:47shogunx has joined #ltsp
16:02sepski has quit IRC
16:07japerry has joined #ltsp
16:20
<ogra>
stgraber, the nfs bind mounting to make files writable only works for files that actually exist
16:21
stgraber, i fixed the error thats shown with an empty file ages ago, did that get lost in a merge ?
16:21
<johnny>
i haven't see nthat in ages..
16:21
if you're referring to lts.conf
16:21
err getltscfg
16:21
<ogra>
yes
16:21
<johnny>
uggh.. actually the opposite
16:22
it has errored out for ages
16:22
<ogra>
there is a wrapper script around it in ubuntu that only calls getltscfg if the file is nonempty
16:22
<johnny>
oh.. for ubuntu.. thanks..
16:22
lol
16:22
why not fix getlscfg instead..
16:22
it shouldn't even need a [default] section
16:22
it should just ignore it
16:22
everything is default until you get to the first []
16:23
<ogra>
right, that was not possible back then, ubuntu was in deep freeze ... later i forgot about it
16:23
else i would have fixed getltscfg
16:23
<johnny>
you could have fixed it for us :)
16:23
i don't understand all that lexer stuff going on :)
16:23
<ogra>
in any case the initscript that parses it has that fix for ages, i wonder why stgraber sees the error
16:23
well its two lines in the script
16:24
<alkisg>
getltscfg even complains for the default lts.conf file (the one with the comments)
16:25
<stgraber>
ogra: no idea, but I got the error on Jaunty ...
16:25
<johnny>
alkisg, and it will.. until getltscfg is fixed
16:25
<alkisg>
I think I can send a patch for the parser file, if you like...
16:25
<johnny>
not just in the init script
16:25
<alkisg>
I've done my share of bison/flex..
16:26
<sistemas_>
please where do i found local apps manual
16:26
<johnny>
alkisg, neat.. go right ahead :)
16:26
<alkisg>
johnny: ok, but lets wait to hear from ogra - maybe he has one already :)
16:26
<stgraber>
ogra: and I don't see anything in the initramfs that actually modifies lts.conf
16:26
<ogra>
# Read in lts.conf entries.
16:26
if [ -f /etc/lts.conf ] && [ -n "$(grep -v ^# /etc/lts.conf)" ]; then
16:26
eval $(getltscfg -a) || true
16:26
fi
16:27
from ltsp_config in hardy
16:27
<johnny>
that's a lame fix
16:27
lol
16:27
<ogra>
johnny, thats all you can do in a deep freeze
16:27
<johnny>
yeah.. looks like it got lost
16:27* johnny puts ogra in deep freeze
16:27
<johnny>
hehe
16:27
<ogra>
as i said, it was a week before release or some such when i added that
16:28
<stgraber>
ogra: look at ltsp-core and ltsp-setup, it's not done for these two
16:28
<ogra>
and since it didnt show up anymore i forgot about the error
16:28
<Gadi>
sounds like the default lts.conf file just has a syntax error :P
16:28
<stgraber>
so the init scripts still show the error :)
16:28
<ogra>
Gadi, no, getltscfg spills an error if it finds an empty file
16:28
<johnny>
Gadi, the syntax error is missing [default]
16:28
<stgraber>
Gadi: well, getltscfg fails if the file is empty so ... :)
16:28
<alkisg>
Gadi: did you try to support an empty lts.conf and got reduction warnings in the parser?
16:29
<johnny>
Gadi, my suggestion is it ignore default from now on
16:29
<stgraber>
johnny: nope, you don't need [default]
16:29
<Gadi>
how come the default file is missing [default]?
16:29
:)
16:29
<johnny>
stgraber, you sure?
16:29
i'm pretty sure i ran into that
16:29
<stgraber>
johnny: yes
16:29
johnny: my current file is:
16:29
<Gadi>
no wonder none of the newbies know how to make an lts.conf file
16:29
<stgraber>
[example]
16:29
key=value
16:29
and it works :)
16:29
<ogra>
stgraber, well, it was an ubuntu fix that didnt go upstream because i found it to hackish, a better fix should be found ...
16:29
<Gadi>
what was the old default file?
16:30
<johnny>
ah.. ok.. i see
16:30
it just needs something []
16:30
<stgraber>
ogra: right, ideally getltscfg should just be fixed :)
16:30
<johnny>
stgraber, but.. can it work with
16:30
key=value
16:30
[something]
16:30
key=value
16:30
<Gadi>
no
16:30
<johnny>
that would be nice
16:30
<Gadi>
nor should it
16:30
<stgraber>
ogra: what do you think is best ? adding your grep to all call to getltscfg or use the two-liner I added to the lts.conf ?
16:30
<johnny>
i disagree
16:30
<ogra>
Gadi, http://paste.ubuntu.com/118616/
16:30
<Gadi>
thats not proper ini format
16:30
<johnny>
no need for [default]
16:30
<Gadi>
:)
16:31
<johnny>
oh... proper ini format.. lol
16:31
didn't know we were going for proper ini format..
16:31
<ogra>
stgraber, your fix looks cleaner
16:31
<Gadi>
ogra: didnt the default file once have: [default] VERSION = blah
16:31
?
16:31
<ogra>
Gadi, in hmm, feisty
16:31
<Gadi>
lol
16:31
<ogra>
in some antique version
16:31
<Gadi>
goes to show you when I last looked ;)
16:32
see, when you moved from that, it broke!
16:32
:P
16:32
<alkisg>
Gadi: http://en.wikipedia.org/wiki/INI_file - Parameters may be grouped into arbitrarily-named sections.
16:32
<stgraber>
ogra: ok, I'll drop the grep from ltsp_config and say in the commit that we should fix getltscfg (we'd need a C coder there)
16:32
<alkisg>
Notice the "may" word there :)
16:32
<ogra>
stgraber, jammcq or sbalneav should be able to fix it, one of them is upstream :P
16:33
<Gadi>
getltscfg is the only code from old LTSP!
16:33
<ogra>
nah
16:33
<Gadi>
it must be preserved and revered
16:33
:)
16:33
<ogra>
there are some script snippets
16:33
<stgraber>
Gadi: I could just write a getltscfg.sh that'd check then start getltscfg but that's even more hackish :)
16:33
<ogra>
yeah, doesnt buy anything
16:35
<stgraber>
ogra: I guess that error appeared when I started using upstream's init scripts in Ubuntu
16:36
<ogra>
well, the fix was in ltsp_config
16:36
<stgraber>
yeah but for some reason the init scripts also load lts.conf :)
16:36
<ogra>
ugh
16:36
thats messy
16:36
drop that
16:37
<stgraber>
well, that's upstream :)
16:37
<ogra>
no, its your cluser hack
16:37
if [ -f "/etc/ltsp/getltscfg-cluster.conf" ]; then
16:37
# Tell the control center that we are booting and get lts.conf
16:37
eval $(getltscfg-cluster -a -l refresh) || true
16:37
fi
16:38
ltsp-trunk/client/initscripts/ltsp-setup
16:38
<stgraber>
ah right, it's only ltsp-cluster :) so that's fine
16:38
<ogra>
ergh
16:38
and ltsp-trunk/client/initscripts/ltsp-core too ?!?
16:39
<stgraber>
yeah but not doing the same thing
16:41
<ogra>
woah, you actually call it three times
16:41
<stgraber>
yes :)
16:41
<ogra>
in client/screen_session as well
16:41
<stgraber>
indeed and screen_session is called every-time X dies
16:41
so it reloads the config every-time
16:42
that way we can change SCREEN_ / AUTOLOGIN / ... without rebooting
16:42
<ogra>
ah
16:42
<stgraber>
basically first one tells the control center the terminal is booting, next gets the config for SOUND / JETPIPE / ..., next gets the config for the SCREEN_XX and X settings
16:43
the first also sends the hardware information so at the second request it gets the config including hardware-specific variables
16:43
<ogra>
i think it costs about a second to exec getltscfg ....
16:44
so you add three seconds to your boot ... but well... its cluster ...
16:44
<stgraber>
well, what we're deploying now is either VIA C7 or Intel Atom so ... :)
16:44
and we don't affect regular LTSP's boottime
16:45
ogra: speaking of boottime, something I've noticed on really fast thin clients what's taking the most time is ipconfig
16:46
ogra: looking at it (and at the switch attached to the client), ipconfig seems to be starting before the interface is up making the first few requests to fail. Any idea on that ?
16:46
ogra: I have some clients where ipconfig takes 6s and then they boot in 5s or so
16:46
<ogra>
talk to Keybuk, its his code
16:46
afaik
16:47
<stgraber>
k
16:47
<ogra>
though it might predate him touching initramfs and it might be from jbailey
16:47
i'm not 100% sure on that
16:48
does debian still use ipconfig ?
16:48
in their nfs code
16:48
<stgraber>
no idea, but would interested by the answer :)
16:49
<ogra>
well, vagrant could anser it, but he is probably at a lenny releaseparty
16:49
;)
16:51
<alkisg>
johnny: all that's needed is an "|" in parser.y, like this:
16:51
configuration
16:51
:
16:51
| sections
16:51
;
16:52
number of conflicts remain the same as before: conflicts: 2 shift/reduce, 1 reduce/reduce
16:54
stgraber: ipconfig has a *lot* of problems, I've send some patches but only one of them got accepted before I "resigned"... :(
16:54
<stgraber>
ogra: is it possible to start dhclient from the initramfs ?
16:55
<ogra>
you would bloat it and i doubt it will be faster
16:55
<stgraber>
ogra: that'd also let us renew our dhcp lease
16:55
<alkisg>
ipconfig looses packets, so dhclient would be faster
16:55
<stgraber>
one issue I have with ipconfig is that when you have your thin clients weeks without reboot, the IP is re-assigned to another thin client
16:55
making both of them to crash
16:56
<ogra>
i would like alkisg's solution but i'm somewhat not feeling good mangling pxelinux.cfg by default
16:56
we tried to avoid that for the default case all the time so i'd not be happy to start with it now
16:59* alkisg thinks uclibc would be the best path here... but yup, /me knows it can't :(
16:59* stgraber adds dhclient3 to its chroot and does some testing
17:02
<ogra>
stgraber, we're free to do any hack we like in ltsp_nbd, but nfs is different
17:04Gadi has left #ltsp
17:07
<stgraber>
ogra: I can easily skip the networking part of initramfs if required, I just have to do the request before myself and touch /tmp/net-iface.conf
17:10sistemas_ has quit IRC
17:10
<ogra>
stgraber, i'm open to anythng in nbd since only we use it ... but nfs needs to be done with debian/ubuntu dev reviews
17:12
<stgraber>
ogra: sure, I don't plan on modifying anything outside ltsp if I can do it in ltsp :)
17:12
<alkisg>
stgraber: some information from net-iface.conf is needed, e.g. the tftp path to get lts.conf is worked out from the "filename" line in net-iface.conf.
17:13
So if you use dhclient, you'll either have to modify the scripts or generate that file...
17:13
<ogra>
that should be an env var
17:13
handed over by dhcp
17:13
<stgraber>
ogra: net-iface.conf is then sourced
17:14
<ogra>
sure, but i think filename and rootserver are exported before anyway
17:14
by the kernel
17:14
<alkisg>
ogra, I don't think so...
17:14humbolt has joined #ltsp
17:15
<alkisg>
ipconfig gets this info from dhcp, writes it to net-iface.conf, and then it's sourced. IPAPPEND or ipopts has to be used for the kernel to know such things before ipconfig is ran...
17:15
<humbolt>
hi guys, how is the ltsp-fat-client project doing?
17:15
<stgraber>
ogra: I don't see them in the environment at least
17:16
<ogra>
hm, i always thought they are handled like cmdline kernel vars
17:16
<humbolt>
As there are all these cheap eeebox pcs out there, I would like to use them as clients and make use of the extra processing power and graphic chips.
17:16
<alkisg>
No, only if ipopts=xxx kernel parameter is used :( (which is what IPAPPEND does)
17:17
<humbolt>
I found a couple of docs on the matter, but am wondering if there is one ltsp recommended doc?
17:18
<stgraber>
ogra: oh, ltsp_nbd is really great for the network part, it's just one line to comment to get rid of the current network code :)
17:18
<ogra>
heh
17:19
<humbolt>
It is somewhat clear to me, how to get a fat client setup, but I am wondering, if you guys really recommend nbd over nfs in this case, as each software upgrade would require the image to be rebuilt.
17:20
<stgraber>
ogra: that dhcp lease renewal thing is currently my #2 bug so if I can just use dhclient, I'll get a lot of happy users :)
17:21
<humbolt>
It also appears like a waste of space to me, if I have the dir tree on there and an image of it as well. Makes sense for small thin client images, but for full blown fat systems?
17:21
<ogra>
stgraber, if it works, why not
17:21vagrantc has joined #ltsp
17:22
<humbolt>
ogra: I remember your ltsp-fat-client blueprint for feisty or what it was back then. Still says deferred. Did you make another attempt for another release?
17:22
<ogra>
vagrantc, !!!
17:22
<vagrantc>
ogra: hi!
17:22
<ogra>
humbolt, i dont do much official ltsp work anymore, stgraber took over most of my duties in ubuntu
17:22* ogra hugs vagrantc
17:23
<ogra>
CONGRATS !!
17:23
<vagrantc>
yay!
17:23
ogra: thanks!
17:23
<ogra>
just in time ...
17:23* ogra goes afk to celebrate into his b-day :)
17:24
<alkisg>
Ah! Lenny released on the 14th???!!! Yeah, congrats!!! :)
17:25
<ogra>
well, 15th actually
17:25
<humbolt>
ogra: I see.
17:26
stgraber: Can you give me an update on LTSP fat client support?
17:26
<stgraber>
humbolt: nubae has done some work on that but it's not yet in a state where I'd include it in Ubuntu
17:26
humbolt: http://www.nubae.com/creating-a-low-fat-client-environment-for-ltsp
17:28
good, I get my initramfs to use dhclient, now I need to do the same as ipconfig's /tmp/... file
17:28
I can probably just write a /sbin/dhclient-script for that
17:28
<vagrantc>
you'll have to parse dhclient's output/logs and figure out all the correlary variable names ...
17:31
<stgraber>
vagrantc: that's what dhclient-script is for AFAIK
17:31
<alkisg>
stgraber: if you do include dhclient, please provide an lts.conf parameter for its "require" option! It would make it possible for clients to boot in multiple dhcp server environments. :)
17:32
<stgraber>
alkisg: won't work
17:32
<alkisg>
Why?
17:32
<stgraber>
alkisg: at this point I don't have access to tftp or the nbd
17:32
alkisg: as I'm starting dhclient to get access to the network
17:33
<alkisg>
Ah, right, it can't be an lts.conf option
17:33
It has to be in a static file in the chroot. OK... :(
17:33
<stgraber>
vagrantc: all variables are in dhclient-script's environment :) I just need it to write to /tmp/ and it should work
17:42
<vagrantc>
stgraber: they're the same variables?
17:43humbolt has quit IRC
17:43
<stgraber>
vagrantc: no, that's why I'll need a minimal dhclient-scripts to generate that file in /tmp
17:43
<vagrantc>
that's what i though :)
17:43japerry has quit IRC
17:43
<stgraber>
though, for now I'm trying to understand why I get all the variables from my DHCP except next-server ...
17:44
<vagrantc>
siaddr?
17:44
i've been down this road years ago ... but i haven't touched it since ipconfig has been mostly working
17:49bobby__C has quit IRC
17:51bobby__C has joined #ltsp
17:56nothingman has joined #ltsp
18:10alkisg has quit IRC
18:17artista-frustrad has quit IRC
18:27
<stgraber>
argh, won't work ... there is no code in dhclient to get next-server ...
18:29* stgraber tries other way to speed up the boot process then ...
18:37vagrantc has quit IRC
19:13artista_frustrad has joined #ltsp
19:20Gadi has joined #ltsp
19:21
<Gadi>
stgraber: why does your dhclient in initramfs need next-server?
19:24
<nothingman>
we're speeding up boot processes? I wanna watch!
19:25
<stgraber>
Gadi: to get lts.conf
19:29hanthana has joined #ltsp
19:41pmatulis has joined #ltsp
19:41
<Gadi>
stgraber: you don't get it from rootserver?
19:41
or does rootserver rely on next-server?
19:43
<stgraber>
Gadi: ROOTSERVER is a variable exported from ipconfig
19:43
Gadi: which actually is next-server
19:44
I'm currently playing with udhcpc, it's huper-light and does exactly what i need
19:44
only problem is that it's in universe
19:44
and I have real doubt I can get it in main by thursday
19:48alekibango has joined #ltsp
19:48
<Gadi>
stgraber: have you tried dhclient + tftp-server-name?
19:49
<stgraber>
Gadi: that probably works though I don't feel like asking all my users to change their dhcp config :)
19:49
<Gadi>
asking?
19:49
I thought your configurator does that
19:49
<stgraber>
and you'd need next-server + tftp-server-naam ...
19:49
Gadi: I'm talking about changing it in Ubuntu not only for my customers
19:50
<Gadi>
I thought ubuntu installs a default dhcpd.conf
19:50
:)
19:50
<stgraber>
:)
19:52
<Gadi>
you could always make a plugin to l-b-c: --use-tftp-server-option
19:52
:)
19:53Nubae has joined #ltsp
19:53
<stgraber>
doh, that thing gets an IP in less than a second :)
19:53
<Nubae>
wtf... johnny u there?
19:54
I cant find where to set the keyboard anzwhere on the gentoo lived
19:54
with xfe
19:54
<stgraber>
setxkbmap -layout de ?
19:55
<Gadi>
stgraber: which thing the udhcpdc?
19:55
<stgraber>
Gadi: yeah, udhcpc
19:55pscheie_ has joined #ltsp
19:55
<Gadi>
cool - so you actually have a 7 sec boot?
19:56
<Nubae>
thanks!hpefullyalso valid for the rest off the apps amd mjot just ter,inal
19:56
<stgraber>
something like that yeah :)
19:56
<Gadi>
complete to login screen?
19:56
<stgraber>
yup
19:56
<Gadi>
wow
19:56
jaunty?
19:56
<Nubae>
ah ??? it is
19:56
<stgraber>
I'm cleaning up the script :)
19:56
<Gadi>
or intrepid?
19:56
<stgraber>
yeah, jaunty
19:57rjune_ has quit IRC
19:57
<Gadi>
does it use the in-kernel modesetting and such?
19:57
<stgraber>
the good thing with that hack is that I don't touch any existing file so I can package it :)
19:57
<Gadi>
you just hack the initramfs hooks and scripts?
19:58
<stgraber>
yeah
19:58
<Gadi>
nice
20:02chrisinajar has joined #ltsp
20:10pscheie has quit IRC
20:19
<nothingman>
has anyone here done a test thin client using virtualbox?
20:26makghosh has quit IRC
20:30chrisinajar has quit IRC
20:42alekibango has joined #ltsp
20:49makghosh has joined #ltsp
20:49w0lter has joined #ltsp
20:49w0lter has left #ltsp
20:54try2free has joined #ltsp
20:57try2free has left #ltsp
21:07li-la has quit IRC
21:08w0lter1 has joined #ltsp
21:09w0lter1 has left #ltsp
21:20pmatulis has quit IRC
21:28
<stgraber>
Gadi: from the boot splash to ldm => 12s
21:28
and you also get real dhcp client that renew its leases :)
21:29
oh and my thin clients also start a ssh server, it probably slow things down a bit
21:39
Gadi: I think I'll ship that initramfs script anyway and make it conditional to udhcpc being installed in the chroot
21:39
so if I can get it in main for Jaunty it'll just be on by default, otherwise you'll just have to install it :)
21:51
<cyberorg>
stgraber, 25s to gnome :) http://hpjansson.org/blag/2009/02/15/shoveling-snow/
21:58
<stgraber>
cyberorg: well, I'm probably under 20s to gnome, it's just that I don't have an appserver around :)
21:58kaos01 has joined #ltsp
21:59
<cyberorg>
stgraber, work on boot speed by everyone is great news for users, all next editions of distros would be boasting their boot time slashed by half :)
22:00
<kaos01>
hi, trying to get localdevice working in ltsp5, on the client i cansee /var/run/drives/ and on the server I can see /media/$USER but nothing seems to be in neither directories
22:00
spped up boot -> dont reboot :)
22:00hanthana has quit IRC
22:00
<cyberorg>
kaos01, install ltspfs package on the server, and if using ubuntu add user to fuse group
22:01
<kaos01>
yes i have ltspfs on server , although i dont have fuse group i just have crw-rw-rw- 1 root root 10, 229 2009-02-06 05:53 /dev/fuse as I had in ltsp4.2
22:01CaScAdE^FarAway has quit IRC
22:02CaScAdE^FarAway has joined #ltsp
22:03
<kaos01>
well i guess the first problem is that in /var/run/drives/FLASH_DRIVE i cant see the content of my usb
22:04
<cyberorg>
kaos01, that shows up when you access it from the client session
22:04hanthana has joined #ltsp
22:04* cyberorg got to go
22:05kaos01 has left #ltsp
22:05kaos01 has joined #ltsp
22:18|Ryan52 has joined #ltsp
22:19Ryan52 has quit IRC
22:21|Ryan52 is now known as Ryan52
22:37
<nothingman>
hi, my ltsp client (a virtualbox vm) hangs on trying to load pxelinux.0
22:38
<Ryan52>
nothingman, that's a familier problem. xD
22:38
is your tftpd running?
22:38
is it serving from the right place?
22:38
can you manually download the file?
22:38
does it work on physical thin clients?
22:39
(the third will answer the first two, but ya..)
22:39
<nothingman>
tftpd is running; real clients work fine
22:39
<Ryan52>
hmm.
22:40
<nothingman>
or, at least they did on Friday
22:40
haven't changed anything since
22:44
is the prepended / necessary in naming files in dhcpd.conf?
22:49li-la has joined #ltsp
22:54Nubae has quit IRC
22:57
<nothingman>
seems to me to be the same problem I had with my regular clients early Friday, with PXE-TFTP errors
22:57
but this just seems to hang
22:58
<Ryan52>
ya, sometimes that happens.
22:58
when tftpd is broken it sometimes just hangs.
22:58
so, can you manually download the pxelinux.0?
22:58
(just from localhost or something)
22:58
<nothingman>
how would I tell?
22:58
I have no OS running
22:59
<Ryan52>
use a tftp client.
22:59
<nothingman>
just a vm with pxe
22:59
<Ryan52>
ok...
22:59
but...
22:59
please use a tftp client (a program) from the server itself.
23:00
<nothingman>
OK
23:01
I get "unknown host"
23:01
<Ryan52>
what?
23:02
that's really....irrelevant.
23:02
<nothingman>
using "tftp localhost:/ltsp/i386/pxelinux.0"
23:03
also using "tftp 127.0.0.1:/ltsp/i386/pxelinux.0"
23:04
<Ryan52>
$ tftp asdf
23:04
tftp> get /ltsp/i386/pxelinux.0
23:04
Received 15987 bytes in 0.1 seconds
23:04
tftp>
23:04
^^ that's how you use it.
23:04
<nothingman>
sorry, that worked
23:05
<Ryan52>
k. still no clue then.
23:06shrek has joined #ltsp
23:31alkisg_ has joined #ltsp
23:38makghosh has quit IRC
23:47makghosh has joined #ltsp
23:48nothingman has quit IRC
23:51makghosh has quit IRC
23:53
<kaos01>
trying to get local devices workingon ltsp5
23:54
on the client where is the local device soposed to apear ?
23:54
<johnny>
on the desktop usually
23:54
in gnome
23:54
<kaos01>
well im not gettign that, so i have a shell on the client
23:54
and trying to see if the device actually got mounted somewhere
23:54
<johnny>
and you are using gnome?
23:54
<kaos01>
yes
23:54
<johnny>
look on the server in /media
23:55
<kaos01>
ok ok , thee is a directory there but access is denied
23:55
<johnny>
you should only be able to access it as the client's user
23:55
<kaos01>
even to root
23:55
<johnny>
yes, that is normal
23:56
<kaos01>
wel laccess is also denied by the user, as group permissions are not set
23:56
<johnny>
so.. open up a terminal IN the logged in session
23:56
hmm.. what distro?
23:56
<kaos01>
fedora
23:56
<johnny>
Ryan52, you know anything about this?
23:57
<Ryan52>
wasn't paying attention...lemme read teh backlog real quick.
23:58alkisg_ has quit IRC
23:58
<Ryan52>
hmm. no clue.
23:59
<kaos01>
can i ask this, is the local device (in this case my usb stick) soposed to get mounted on the client somewhere ?
23:59
so if i have a shell open on the client shoudl i see it mounted somewhere ?
23:59
<johnny>
it should be in /media on the server
23:59
gnome looks there
23:59
<kaos01>
im mean ton the client