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


Channel log from 24 November 2008   (all times are UTC)

00:09Xperiment62 has joined #ltsp
00:10
<Xperiment62>
can LTSP boot original powerpc iMacs?
00:18
<sbalneav>
You'd have to build the chroot on one, but yes, it can
00:19
<johnny>
sbalneav, i know i can cross compile powerpc chroots in a build from src distro
00:20Ahmuck has quit IRC
00:20
<johnny>
but since ubuntu/deb is all binary, it seems like you should be able to fake it
00:20dirigeant has quit IRC
00:21aristidezzz has quit IRC
00:23Ahmuck has joined #ltsp
00:25
<sbalneav>
You could use qemu
00:27
No, you wouldn't be able to do it on gentoo either
00:27
you'd have to use qemu as well
00:30
heading to bed
00:39
<johnny>
cross compile, yes, run no
00:41Xperiment62 has quit IRC
00:49try2free has joined #ltsp
00:59vagrantc has quit IRC
00:59alkisg has joined #ltsp
01:10Ahmuck has quit IRC
01:10Ahmuck has joined #ltsp
01:16nothingman has quit IRC
01:16cyberorg has quit IRC
01:17cyberorg has joined #ltsp
01:19try2free has left #ltsp
01:39Ahmuck has quit IRC
01:45gate_keeper_ has joined #ltsp
01:50Pascal_1 has joined #ltsp
01:51Pascal_1 has left #ltsp
01:53dirigeant has joined #ltsp
01:58alkisg has quit IRC
02:28Q-FUNK has quit IRC
02:30alkisg has joined #ltsp
02:39alkisg has quit IRC
02:39alkisg has joined #ltsp
02:41alkisg has quit IRC
02:54X0d_of_N0d has quit IRC
02:54X0d_of_N0d has joined #ltsp
03:04tjikkun_work has joined #ltsp
03:07dirigeant has quit IRC
03:07dirigeant has joined #ltsp
03:08dirigeant has joined #ltsp
03:20mikkel has joined #ltsp
03:24ltsppbot has quit IRC
03:27siki has joined #ltsp
03:32ltsppbot has joined #ltsp
03:38tjikkun_work has quit IRC
03:47Q-FUNK has joined #ltsp
03:49tjikkun_work has joined #ltsp
03:50siki has quit IRC
04:14randra has joined #ltsp
05:03randra is now known as randra`offz
05:12jstephan has joined #ltsp
05:26cdshan has joined #ltsp
05:32cdshan has quit IRC
05:38cdshan has joined #ltsp
05:41
<cdshan>
Hello ! I am currently using CentOS 5.2 and would like to install ltsp5 on it. Could someone point me to the suitable page to do the same.
05:48Q-FUNK has quit IRC
05:50
<stgraber>
ogra: ping
05:52cdshan has quit IRC
05:57cdshan has joined #ltsp
06:00
<cdshan>
Could someone help me with installing ltsp5 on centos 5.2
06:00
I would like to know where I could get the ltsp rpm's for centos 5.2
06:12
<jammcq>
there aren't any RPMS for ltsp for centos
06:12
the closest we have is LTSP-5 for Fedora 9
06:12alkisg has joined #ltsp
06:13
<jammcq>
but there's some instructions for how to install a F9 LTSP environment on a RHEL5 system
06:13
take a look at k12linux.org
06:15
hmm, I thought there was notes about RHEL5. I don't see them now
06:23BrunoXLambert has quit IRC
06:28gate_keeper_ has quit IRC
06:28
<ogra>
stgraber, you are missing my last changelog entry
06:29
(in ltsp)
06:34Q-FUNK has joined #ltsp
06:36nubae has joined #ltsp
06:37Appiah has quit IRC
06:41randra`offz is now known as randra
06:43gate_keeper_ has joined #ltsp
06:48jammcq has quit IRC
06:52BrunoXLambert has joined #ltsp
06:53
<cdshan>
Jammcq! Sorry I couldn't catch you when you were online. But as you said there are no documentation of installing ltsp5 on CentOs 5.2
06:53
Is there some other site which I could refer to?
06:55Blinny has joined #ltsp
07:01six2one has joined #ltsp
07:03hanthana has joined #ltsp
07:03alkisg has quit IRC
07:05alkisg has joined #ltsp
07:09cdshan has left #ltsp
07:15randra is now known as randra`offz
07:15gate_keeper__ has joined #ltsp
07:17
<stgraber>
ogra: oh ? Weird, I apt-get sourced it
07:18
<ogra>
from jaunty ?
07:18
<stgraber>
ogra: nope, on Intrepid. Did you upload something to Jaunty ?
07:18
<ogra>
yes
07:19
the acpid fix
07:19
to make it installable on armel
07:19
<stgraber>
oh, ok. I'll do a new package then :)
07:19
<ogra>
thanks, apart from that it looks great
07:21
<Q-FUNK>
armel. sounds like caramel.
07:23
<Blinny>
apt-get caramel
07:24
<stgraber>
ogra: updated
07:28gate_keeper_ has quit IRC
07:34gate_keeper__ has quit IRC
07:40dmaran has joined #ltsp
07:46cliebow has joined #ltsp
07:47pscheie has quit IRC
07:47Eghie has joined #ltsp
08:00CAN-o-SPAM_ has joined #ltsp
08:09try2free has joined #ltsp
08:09Q-FUNK has quit IRC
08:09try2free has left #ltsp
08:20Q-FUNK has joined #ltsp
08:26
<dmaran>
Morning all- I am trying to install x11vnc on some clients using LTSP5 Ubuntu 8.04 I have followed the instructions here: https://wiki.edubuntu.org/InstallX11VncOnLtspClients to a "t" and still get no love. The response I get is "xvncviewer: ConnectToTcpAddr: connect: Connection refused
08:26
Unable to connect to VNC server " Any ideas? I have triple checked the files and permissions and rebuilt the client image a few times. Has something changed since this article was last updated?
08:30
<Eghie>
you could check the log files, maybe a firewall or check if the x11vnc is even running or not
08:31Gadi has joined #ltsp
08:33hanthana has quit IRC
08:34
<dmaran>
That is the issue I don't think it is running at client boot
08:44F-GT has quit IRC
08:46
<Gadi>
dmaran: log into the client console on ctrl-alt-f1 and try running by hand
08:46
better yet add: -o /tmp/x11vnc.log to the arguments
08:47
and cat /tmp/x11vnc.log after it fails
08:47
personally, I would run it out of inetd
08:47
but, that's personal preference
08:47* ogra would simply use italc instead
08:47
<Gadi>
or that
08:48
but, dont some prefer to use italc+local vnc?
08:48
to cut down on bandwidth?
08:48
<ogra>
well ,there are two ways to use italc
08:48
either on the servers session or directly installed on the client
08:49
both should work, the session variant has some feature limitation
08:49
(no shutdown/reboot)
08:49
<dmaran>
We are using iTalc now, BUT with the Geographic separation it is unusable
08:50
outside the local site
08:50
the latency is too much to handle we have about a 45 second delay
08:50
<Gadi>
dmaran: do you instal italc on the client?
08:50
<dmaran>
Yes
08:50
<Gadi>
or in the session?
08:51
ah - i thought it used x11vnc itself
08:51
<ogra>
no, it has ts own vnc implementation
08:51
iirc it only uses th libs
08:51
<Gadi>
ah - but still vnc
08:51
right?
08:52* Gadi wonders why x11vnc would work any better
08:54
<cliebow>
!g
08:54
<ltspbot>
cliebow: "g" is Gadi!!!!!!!!!!!!!!!!!!!!!!!!
08:54
<rjune>
!g
08:54
<ltspbot>
rjune: "g" is Gadi!!!!!!!!!!!!!!!!!!!!!!!!
08:54
<dmaran>
We don't know if it will, but we need to try it out. As the 45 second delay makes remote admin impossible
08:55
<ogra>
Gadi, different ports, italc uses ssh eys to authenticate etc
08:55
*keys
08:55
<rjune>
ogra: filled out the form, here's hoping
08:55
<ogra>
great, lets see what comes out
08:55
<Gadi>
ah
08:56
the "directx" of the desktop shadowing world ;)
08:56
morning, cliebow, rjune
09:00
<cliebow>
top o the mornng oh sage
09:01
<rjune>
howdy gadi.
09:03monteslu has joined #ltsp
09:07jstephan has quit IRC
09:10dirigeant has quit IRC
09:12
<stgraber>
ogra: when do you plan to upload ldm and ltsp (both should be uploaded at the same time as I don't think one would work without the other) ?
09:20F-GT has joined #ltsp
09:21pscheie has joined #ltsp
09:39mikkel has quit IRC
09:43
<Blinny>
We installed iTalc on the server, not in the client chroot
09:46
<Gadi>
can anyone help me test a new xrandr script: lp:~gideon/ltsp/ltsp-trunk-xrandr ?
09:47
I am hopeful it will provide backwards compat with current X_MODE_0, while providing automatic multihead and some other xrandr goodies
09:47
oh, and obviate the need for CONFIGURE_X
09:48
but, I need some help testing
09:52
<alkisg>
Gadi, what do you need from the tester hw?
09:53
E.g. right now I got a TC with an embedded SiS graphics card and 1 screen, will this do?
09:53selffik has joined #ltsp
09:53
<Gadi>
alkisg: sure thing
09:53
<Blinny>
Sorry Gadi slamming with programming now :(
09:53
<Gadi>
I first need to make sure that if you do:
09:53
CONFIGURE_X=False
09:53
X_MODE_0=<some mode or modeline>
09:54
that it sets the mode for you
09:54
<alkisg>
Gadi, OK, first things first: what do I need to install? :)
09:54
<Gadi>
well, here - I'll make it easier on u since, it is only one script
09:54
let me pastebot it
09:56
<ltsppbot>
"Gadi" pasted "This goes in: /opt/ltspi386/usr/share/ltsp/xinitrc.d/I10-xrandr" (73 lines) at http://pastebot.ltsp.org/111
09:57
<Gadi>
of course, you need to be running the latest ltsp from -trunk
09:57
with the xinitrc stuff in it
09:57
this is newer stuff than what is in intrepid
09:58
ah, damn - mtg
09:58
bbiab
09:59Eghie has quit IRC
10:00
<alkisg>
Gadi, so... put the trunk in the chroot sources.list, update, put the script, ltsp-update-client, right?
10:00
Trunk link? :)
10:11Devilmaster has joined #ltsp
10:12
<Devilmaster>
hi
10:13
has anyone some ideas, why ltspfsmounter does not create /tmp/.$user-ltpfs ?
10:23Devilmaster is now known as bstorp
10:23
<bstorp>
hello?
10:26
hm okay. maybe someone is alive a little bit later
10:26bstorp has quit IRC
10:38staffencasa has joined #ltsp
10:39jammcq has joined #ltsp
10:39
<jammcq>
Good morning #ltsp
10:45alkisg has quit IRC
10:52tjikkun_work has quit IRC
11:00
<Q-FUNK>
!j
11:00
<ltspbot>
Q-FUNK: Error: "j" is not a valid command.
11:00
<Q-FUNK>
bad robot!
11:00
hiya jammcq!
11:06
<jammcq>
hey Q-FUNK
11:09prpplague has joined #ltsp
11:09
<prpplague>
jammcq: ping
11:10
<jammcq>
hey prpplague
11:10
how you doing?
11:10
<prpplague>
jammcq: not too bad, always could be better
11:10
<jammcq>
cool
11:11
<prpplague>
jammcq: quick question for ya, iirc you guys have done some ltsp kiosks before, ever do anything for a pricechecker style kiosk?
11:11
<jammcq>
nope
11:11
we did some point-of-sale stuff, and there's a way to put Firefox into a kiosk mode
11:11
but I've not taken it any further than that
11:12
<prpplague>
hmm, just curious, we have new kiosk design and got a request for it work with ltsp
11:12
<jammcq>
wow, cool
11:12
you still playing with ARM chips ?
11:13
<prpplague>
yea, exclusively, not much $$ is the x86 dev world
11:13japerry has quit IRC
11:13
<jammcq>
you should talk to Ogra. he's getting into ARM stuff too. ubuntu has a port to arm
11:14
<prpplague>
jammcq: yea, i sent the ubuntu guys a bunch of our arm boards, david mandala
11:14
jammcq: unfortuantely, i'll also do some wince dev arm as well
11:14Ahmuck has joined #ltsp
11:20
<prpplague>
jammcq: http://imagebin.ca/view/WsElZHD.html
11:20
<jammcq>
hmm, very cool
11:21
<prpplague>
guess i'll have to get the latest ltsp and give it a try
11:22
<ogra>
prpplague, hey, my EVM board came from you ?
11:22
<prpplague>
jammcq: ever go to wal-mart or somewhere else for the auto section and try to find a battery or windshield wiper blades? they usually have a small kiosk to help you make the selection
11:22
ogra: it is possible, i don't know which evm you have
11:22
<jammcq>
yeah
11:23
<ogra>
i havent ad time to even remotely care for ltsp on arm yet ... we are still shuffling the archive aroud to make everything build first
11:23
<prpplague>
jammcq: this is targeted at that and price checker
11:23
<jammcq>
I always prefer the kiosk, cuz the employees are mostly worthless
11:23
<prpplague>
indeed
11:23
<ogra>
prpplague, TI OMAP35x EVM ... the mistral board
11:23
<prpplague>
ogra: ahh, nope not mine
11:24
<ogra>
oh, what did you send then ? i know there is nore to come
11:24
*more
11:24
<prpplague>
we sent a couple hammer kits and nail boards
11:24
http://www.tincantools.com
11:24
<ogra>
i'm mainly waiting for a beagleboard and some other onknown thing
11:24
<prpplague>
ogra: yea we sell adapters and such for the beagle as well
11:24
<ogra>
ah, sweet !
11:25
<prpplague>
ogra: we have a couple expansion boards that will be available soon
11:25
<ogra>
the hammer is arm9 though
11:25
<prpplague>
yea
11:25
<ogra>
wont work with ubuntu
11:25
we're focusing on arm7
11:25
<prpplague>
ogra: armV7
11:26
<ogra>
right
11:26
<prpplague>
ogra: hehe, big difference between arm7 and armv7
11:26
<ogra>
i hate how confusing this arm naming is
11:26
<prpplague>
ogra: yea, tincantools was just giving aways some stuff to developers
11:26
i didn't expect much to come of it
11:27vagrantc has joined #ltsp
11:28japerry has joined #ltsp
11:34Blinny has quit IRC
11:46Blinny has joined #ltsp
11:50alkisg has joined #ltsp
11:59chrisinajar has joined #ltsp
12:11alkisg has quit IRC
12:13Q-FUNK has quit IRC
12:35kharloss has joined #ltsp
12:38Q-FUNK has joined #ltsp
12:43Tomaatti has joined #ltsp
12:46selffik has quit IRC
12:46selffik has joined #ltsp
12:46Tomaatti has quit IRC
12:52mikkel has joined #ltsp
12:54Q-FUNK has quit IRC
12:55Q-FUNK has joined #ltsp
13:06
<Gadi>
vagrantc: ping
13:07
<vagrantc>
Gadi: heya
13:07
<Gadi>
hi
13:07
can you help me test something?
13:07
or can I add it to a queue you may have?
13:07
:)
13:07
<vagrantc>
Gadi: there's an alternative to cdfs called... cddfs...
13:08
<Gadi>
lp:~gideon/ltsp/ltsp-trunk-xrandr
13:08
<vagrantc>
Gadi: currently trying to get new ltsp/ldm packages uploaded within the next half hour
13:08
<Gadi>
ah, ok
13:08
after that?
13:08
<puppy dog eyes>
13:08
<vagrantc>
i like the sounds of that, though :)
13:08
<johnny>
hmm.. maybe ltspbot should have a message thing..
13:08
i found a neat feature of a bot
13:08
to do that
13:08
<Gadi>
vagrantc: its a single scriptie
13:09
johnny: you mean like a punce?
13:09
*pounce
13:09* ogra tries to imagine the sound of puppy dog eyes
13:09
<Gadi>
its like the sound of one dog clapping
13:09
<johnny>
yeah
13:09
<Gadi>
:)
13:09
<ogra>
heh
13:18Q-FUNK has quit IRC
13:23cosf has joined #ltsp
13:26alkisg has joined #ltsp
13:29
<Blinny>
How do you turn off vino? Seems like all my users have it enabled and so port 5900 on the server is open, and when I vnc to it I get one of my user's sessions.
13:30
<johnny>
maybe it starts in /etc/xdg/autostart ?
13:31bobby_C has joined #ltsp
13:33
<Blinny>
Not that I can see
13:34
<johnny>
it isn't starting by default for me..
13:34
<Blinny>
Could it be connected to iTalc at all?
13:35
<johnny>
not sure
13:35
i don't use anything like that
13:35
<Gadi>
sounds like italc would be the one opening the port
13:35
vino may run even if no one enables its use
13:35
i believe
13:36
<Blinny>
Strange that it arbitrarily picked one of my users' sessions to display
13:36F-GT has quit IRC
13:36F-GT has joined #ltsp
13:37
<Blinny>
connected to host localhost port 5900
13:38twinprism has quit IRC
13:39
<alkisg>
Blinny, iTalc uses 5900... so even if someones wants to run Vino, he can't, he has to use another port for either of these two...
13:40
<Blinny>
iTalc uses 5900 on the server or when installed in the chroot ?
13:40
<alkisg>
Ehm... both? :)
13:41
<Blinny>
Hmm.
13:42
<alkisg>
Blinny, I think that if it is installed in the chroot, it'll use 5900 on the clients as well as on the server
13:42
<Blinny>
So what about all these enabled=true in $/.gconf/desktop/gnome/remote_access ?
13:42
<alkisg>
If it's not on the chroot, only on the server
13:42
<Blinny>
er, ~/.gconf ...
13:42
<alkisg>
Well, when I tried to enable remote access, I couldn't, I had to change the vino port to something <> 5900...
13:43
<Blinny>
Where did you do that?
13:43
vino-preferences ?
13:44
<alkisg>
Yeap
13:44
<ogra>
vino isnt enabled by default on any ubuntu system
13:45
can only be italc you are seeing there unless you fiddled with the gconf keys of vino yourself
13:46
<Blinny>
Thanks guys.
13:50
<ogra>
stgraber, both uploaded, do you ping ubuntu-archive about binary new for ldminfod ?
13:50
(and we'll probably need a MIR)
13:50
<stgraber>
ogra: do we really need a MIR for something that was already in Main ?
13:51
<ogra>
not sure, last i looked we didnt, but policies change :)
13:51
<stgraber>
ok, I'll ping steve once the package appears in the binary new queue
13:51
<ogra>
in any case it will lock up in the queue until someone manually unqeuese it
13:52
un-queues
13:52
<vagrantc>
hah! debian beats ubuntu again!
13:53cosf has quit IRC
13:53
<ogra>
ltsp_5.1.35 and ldm_2.0.19 here
13:54
<vagrantc>
we'll have the same upstream again
13:54
<ogra>
oh
13:54
yeah, and i see ou beat me by 30 min :)
13:54
<vagrantc>
exactly
13:54
<ogra>
or 20
13:54
<vagrantc>
thereabouts
13:54
we'll see how long it actually takes to hit the mirrors, though.
13:54
<ogra>
well, thats because stgraber has no upload rights yet
13:55
he always has to wait on his slacking upload bitch
13:55
;)
13:55
ubuntu will take about 3h
13:55ltsppbot has quit IRC
13:55
<ogra>
netx build queue run is at :03
13:55
<stgraber>
a bit longer as we may need some paperwork to get ltsp in (ldminfod)
13:55
<vagrantc>
of course, i cheated. i'm still shipping ldminfod in ltsp-server
13:56
<ogra>
ah
13:56
so we're more *modern* :)
13:56
<vagrantc>
in some regards.
13:58
<Gadi>
hmm...
13:58
ldminfod is now where? ldm-trunk?
13:58
<ogra>
yes
13:59* Gadi wonders if there is an ldm server package?
13:59
<vagrantc>
there will be
13:59
<Gadi>
ah
13:59
ok
13:59
stgraber: do you already make an ldm-server pkg?
13:59
<ogra>
stgraber, btw, we should drop the xinetd configs from warren somewhere into the docs/examples section
13:59
<vagrantc>
the ironic thing is, even though i had gently pushed for it for years, i'll probably be the last to upload it :)
14:00
<ogra>
Gadi, its called ldminfod
14:00
<Gadi>
ah
14:00
<ogra>
not ldm-server
14:00
<Gadi>
so, if I want an ldm-ready app server, I should install ldminfod and ltspfs
14:00
atm
14:00
:)
14:00
<stgraber>
Gadi: yes
14:00
<vagrantc>
i would think ldm-server would be better, just in case there's anything else we might want to add to it
14:00
<stgraber>
Gadi: that's what I do here, no need of ltsp-server anymore
14:00
<Gadi>
excellent
14:01
<vagrantc>
of course, keeping it optional things, not mandatory things
14:01
<Gadi>
vagrantc: I would suggest a meta-pks
14:01
*meta-pkg
14:01
<ogra>
Gadi, well, ltsp-server recommends it
14:01
<vagrantc>
Gadi: why?
14:01
<ogra>
so it gets installed automatically
14:01
<Gadi>
that way, you can have ldm-server depend on ldminfod, ltspfs, etc
14:01gbolte has joined #ltsp
14:01
<ogra>
no need for any meta stuff
14:01
<vagrantc>
Gadi: why not have ldm-server include ldminfod and all that other stuff?
14:01
<ogra>
why should ldm depends on ltspfs
14:01
<gbolte>
hi
14:02
<ogra>
its a dep of ltsp-server anyway
14:02
<Gadi>
hmm
14:02
<vagrantc>
recommends would be appropriate
14:02
<ogra>
no need for extra dupication
14:02
right
14:02
<vagrantc>
or suggests
14:02
<Gadi>
ogra: suppose you want just the app server
14:02
<ogra>
Gadi, so ?
14:02
<Gadi>
not the ltsp booting part
14:02ltsppbot has joined #ltsp
14:02
<vagrantc>
Gadi: i don't see why we need ldminfod separate from an ldm-server package...
14:02
<Gadi>
so, I dont want to install ltsp-server
14:02
<ogra>
right
14:03
<Gadi>
just ldminfod and ltspfs
14:03
<ogra>
you install openssh-server, ldminfod and ltspfs
14:03
not the common case
14:03
<Gadi>
but, as more things come out, it would be nice to have an app server pkg that pulls in all
14:03
<vagrantc>
ogra: but not uncommon either
14:03
<Gadi>
not common for you :)
14:03
quite common for me, and most likely stgraber
14:04
<vagrantc>
it's how we configured stuff at freegeek before ltsp5 was a gleam in the milkman's eye
14:04
<ogra>
pfft
14:04
do what you want
14:04* vagrantc will
14:04
<Gadi>
hehe
14:04
<ogra>
i dont have to maintin it anymore (soon) :P
14:04
<Gadi>
just asking
14:04
:)
14:04
<ogra>
if stgraber wants to add overhead, he is free to do so
14:04
<vagrantc>
Gadi: so i still don't see why ldminfod should be separated out from an ldm-server package...
14:05
<ogra>
there is no ldm-server package
14:05
<Gadi>
vagrantc: agreed
14:05
I was confused
14:05
mea culpa
14:05* vagrantc intends to make an ldm-server package, rather than ldminfod
14:05
<ogra>
why should a package that only contains ldminfod be called different than the only binary stuff it sips ?
14:05
*ships
14:05
<vagrantc>
if there's every anything else it would contain, or if ldminfod is renamed
14:05
<Gadi>
vagrantc: would your ldm-server depend on ltspfs and openssh-server?
14:06
<ogra>
surely only recommend
14:06
if even that
14:06
<vagrantc>
Gadi: definitely openssh-server, might recommend or suggest ltspfs...
14:06randra`offz has quit IRC
14:06* Gadi nods - ok
14:07
<vagrantc>
Gadi: i'd also put some sort of dependency on: gnome | x-desktop-environment | x-window-manager or some such
14:07* ogra suggests x-session-manager
14:07
<Gadi>
it will still be nice in my docs to be able to tell an admin: when using the ldm session, make sure your Linux appliaction server has the "ldm-server" package installed ...
14:07
and know that that would make everything work
14:07
<vagrantc>
ogra: sure... the exact details i'll work out later...
14:08
Gadi: recommends will typically be pulled in by default
14:08
<ogra>
x-session-manager | x-window-manager | xterm :)
14:08
<Gadi>
oh really?
14:08
cool
14:08
<vagrantc>
ogra: yeah... getting barebones.
14:08
<ogra>
heh
14:08
<Gadi>
I always thought recommends are not installed
14:08
<vagrantc>
Gadi: they weren't.
14:08Eghie has joined #ltsp
14:08
<ogra>
they sadly are now
14:08
<Gadi>
oh, ok
14:08
<ogra>
stgraber, btw we should work that out in jaunty
14:08
<Gadi>
then carry on
14:09
:)
14:09warren has quit IRC
14:09
<ogra>
in intrepid i disabled it in ltsp-build-client since it made the chroots explode
14:09
<vagrantc>
recommends as "install-by-default but don't fail to install if they're missing" is great.
14:09warren has joined #ltsp
14:13
<vagrantc>
i wonder if virtualbox thin clients support xrandr...
14:13gbolte has quit IRC
14:17
<ogra>
sure
14:17
just not many settings :P
14:19
<vagrantc>
stgraber: why do you tweak /root/.ssh/known_hosts rather than /etc/ssh/ssh_known_hosts in the ltsp-cluster stuff?
14:23* alkisg wonders why stricthostchecking isn't completely disabled, to save us all the ssh_known_hosts trouble...
14:23
<vagrantc>
because that's evil.
14:23
<ogra>
because its a security flaw
14:24
<alkisg>
vagrantc, I can't think of any possible security flaws
14:24
I'm sure there are, but I can't think of any :)
14:24
<vagrantc>
alkisg: someone plugs their laptop into your network, pretends to be your server, and logs all your keystrokes?
14:24
<alkisg>
vagrantc, nope, afaik the passwords are not send
14:25
<ogra>
but the whole connection is sniffable
14:25
<vagrantc>
alkisg: the ssh connection does pass passwords around
14:25
alkisg: and if someone logs into a website that uses a password ...
14:25
<ogra>
so indeed also the passwords are
14:25
<alkisg>
ogra, but not reproducable (if I get this straight)
14:26
<jammcq>
someone could pretend to be the server and accept connections from thin clients
14:26
<vagrantc>
alkisg: if you disable stricthostkeychecking, then you can spoof *anything*
14:26
<ogra>
yeah
14:26
and do a pretty MIM :)
14:26
<jammcq>
if the thin client isn't checking host keys, nothing would stop the client from logging into the rogue server
14:26
<alkisg>
vagrantc, but you'll have to emulate everything, up to the user profile and data to succeed with this
14:26
<vagrantc>
alkisg: you just pass the whole session off the real server, logging the whole session
14:27
<alkisg>
I don't think that would be really feasible...
14:27
<vagrantc>
alkisg: trivial.
14:27
<ogra>
you dont need to do anything but route through
14:27
<alkisg>
vagrantc, no, I don't think this can be done (passthrough)
14:27
<ogra>
it can
14:27
be sure
14:27
<alkisg>
Because you can't authenticate to the real server
14:27
(the man in the middle)
14:27
<ogra>
you dont need to ... you just froward keys
14:28
<vagrantc>
alkisg: you can pass through by starting another ssh connection to the real server, and pass the whole session through.
14:28
<ogra>
right
14:28
<vagrantc>
there's a reason openssh has stricthostkeychecking enabled by default, and i won't override that.
14:28
<alkisg>
But the real server will ask different "questions", ones who you won't know the answer for
14:28
<ogra>
but the client does
14:28
you just pass them on
14:28dmaran has quit IRC
14:28
<vagrantc>
alkisg: you pass all those questions on to the client...
14:28
<alkisg>
Anyway, I'll have to take your word for that!!! :)
14:29
<ogra>
all the server and client see here is a slight delay
14:29* vagrantc doesn't want to see another "-ac" bug in LTSP
14:29
<ogra>
heh
14:29
yeah
14:29
<alkisg>
heh
14:29
<ogra>
even though that was added by a security guy :P
14:30Appiah has joined #ltsp
14:30
<vagrantc>
i sent some email saying: hey, isn't this a security risk: "-ac disable access control restrictions"... the response was something to the effect of "no, no, everything's fine"
14:31
<Gadi>
hehe - that must have been someone who didnt have a firend enable a screen-melting screensaver on him
14:31
*friend
14:31Q-FUNK has joined #ltsp
14:31warren has quit IRC
14:31
<vagrantc>
heh
14:32warren has joined #ltsp
14:32
<vagrantc>
that's gotta be one cruel trick.
14:32
<stgraber>
vagrantc: we used to have /etc/ssh/ on a read-only fs, it's not that long ago we moved to aufs
14:32
<vagrantc>
stgraber: ah, got it.
14:33
<ogra>
well, and we only did for ubuntu
14:33
or did debian find a way to use aufs with nfs ro-root
14:33
<vagrantc>
haven't tried yet.
14:34
my last go with nbd+squashfs+aufs went fine
14:34
haven't tried nbd+ext2+aufs ...
14:34
but nbd+ext2+bindmounts worked fine
14:34* ogra is talking nfs+tmfs and aufs on top
14:35
<ogra>
*tmpfs
14:35
<vagrantc>
yeah, haven't tried that
14:35
<ogra>
it never worked
14:35
<vagrantc>
the debian-live folks were using it, i thought
14:36
<ogra>
if it works now it wuld be a cool option
14:36
i didnt bother to try nfs since we switched to nbd
14:38
<Gadi>
not to belabor security, but how does fetching the known_hosts key from the server you are trying to verify help security?
14:38
<vagrantc>
i actually had an idea to try nbd+squashfs+aufs with an nfs mount bind-mount of volatile files like lts.conf and known_hosts and such
14:38
Gadi: not so great, really.
14:38
<Gadi>
hehe
14:38
not so "Strict" host key checking
14:39
:)
14:39
<vagrantc>
well, yeah.
14:39
<ogra>
the only fact that helps a bit is that known_hosts is on a ro fs
14:39
<Gadi>
that should be a flag
14:39
<vagrantc>
but it raises the bar
14:39
<Gadi>
NotSoStrictHostKeyCheecking
14:39
<ogra>
NearlyStrictHostKeyCheecking
14:39
<vagrantc>
PonderingWhatItMightBeLikeToHaveStrictHostKeyChecking
14:39
<ogra>
heh
14:39
<Gadi>
ogra, Im talking about the cluster stuff: (echo -n "$LDM_SERVER " ; echo "getkey" | nc $LDM_SERVER 8001) > /root/.ssh/
14:39
known_hosts
14:40
<vagrantc>
that seems pretty ugly
14:40
<Gadi>
hehe
14:40
<ogra>
yeah
14:40
that highly insecure
14:40
*thats
14:40
<stgraber>
ogra: do you happen to know whose archive day it's today ?
14:40
<ogra>
th ewiki can tel
14:40
l
14:41
<stgraber>
oh, we have that on the wiki ?
14:42
<ogra>
https://wiki.ubuntu.com/ArchiveAdministration
14:42
slangasek
14:45alkisg has quit IRC
14:45alkisg has joined #ltsp
14:50Blinny has quit IRC
14:50* alkisg also wonders, if there was a man in the middle, why would he bother with ssh instead of sending a customized nbd image with a modified ldm that doesn't use ssh at all... :P
14:51
<vagrantc>
alkisg: more work
14:51
<alkisg>
Heh... good enough
14:51
<Gadi>
...and where are all the women if not in the middle?
14:53
<vagrantc>
besides, just because it's impossible to implement security perfectly, that's no reason to ignore all risks entirely.
14:53
<alkisg>
vagrantc, yeap, you're right.
15:00cliebow has quit IRC
15:01ogra_ has joined #ltsp
15:02ogra has quit IRC
15:02ogra_ is now known as ogra
15:03kharloss has quit IRC
15:05vvinet has joined #ltsp
15:06twinprism has joined #ltsp
15:08dirigeant has joined #ltsp
15:09gbolte has joined #ltsp
15:10
<gbolte>
hello
15:10
anyone know of a decent graphics card to use with ltsp?
15:11dirigeant has quit IRC
15:11
<ogra>
wow ...
15:11
http://me-am-brasil.livejournal.com/
15:13
<gbolte>
interesting ogra
15:13
:P
15:13
<ogra>
gbolte, surely every intel card would be good btw ...
15:14
so if you can find any clients with intel onboard that wuld be the best you could get
15:14
<gbolte>
yeah thats what I was thinking we have had noting but bad luck with our nvidia setup
15:15
<ogra>
heh
15:15
well, its nvidia after all
15:15
i think the ATI/AMD cards will get better soon as well
15:15
nvidia will be bad until they decide to open up
15:16
<gbolte>
we seem to have a constant issue with workstations locking up and slowing to a crawl with things like thunderbird and adobe pdf
15:24
<japerry>
ogra: are you currently using intel integrated with your LTSP clients?
15:27
<alkisg>
If the dhcp server doesn't send hostnames along with ip addresses, all TCs are named 'ltsp'. I think it would be more helpful in some setups if they were named ltsp138, from the last IP byte, or even 192-168-0-138. Does anyone think this isn't a really bad idea? :)
15:28
<vagrantc>
alkisg: yeah, there's a debian bug about that
15:28
alkisg: i looked into it some...
15:28
<Eghie>
alkisg: that would only working in a /24 network
15:28
and smaller
15:28
<alkisg>
Eghie, it would be better than nothing, but the long form would work for all ipv4 networks
15:28
<vagrantc>
alkisg: http://bugs.debian.org/483760
15:29
<Eghie>
it is indeed better than nothing yes
15:29
<alkisg>
vagrantc, I may try a patch for it, if anyone's interested...
15:29
<vagrantc>
alkisg: go for it.
15:29
<alkisg>
vagrantc, thanks!
15:29
<Eghie>
but I guess a UUID kind of thing or make use of the MAC address would be better
15:29
<alkisg>
the short form or the long one?
15:29
<Eghie>
kind of CRC or something
15:29
<vagrantc>
the mac address gets a little visually ugly
15:29
<Eghie>
just a small hash
15:29
<alkisg>
what about the pxelinux encoding of the ip?
15:29
CA080001
15:30
<ogra>
japerry, i have one, generally intel is supported best for ll features, simply due to the fact that they opensource all of their drivers
15:30
<vagrantc>
although at freegeek, we just end up having a script that sets the hostname early in the boot process
15:30
<Eghie>
alkisg: could be helpfull
15:30
alkisg: or IP to long
15:30
<alkisg>
Eghie, so, 192-168-0-123 ?
15:30kharloss has joined #ltsp
15:30* ogra goes afk
15:30
<johnny>
alkisg, fedora does that
15:30
client-$(ip)
15:30
<vagrantc>
alkisg: the real trick would be to get hostnames that were useable usernames
15:31
<Eghie>
alkisg: see: http://gtools.org/tool/ip-long-convert/
15:31
<johnny>
i integrated warren's work
15:31
<alkisg>
vagrantc, any hostnames would be better than no hostnames... :)
15:31
<vagrantc>
indeed.
15:31
it used to be "(none)"
15:31
<johnny>
it'd be nice if fedora/gentoo/ubuntu/debian would have the same default hostnames
15:31
<warren>
what work?
15:31
<alkisg>
I mean, if I have a router=dhcp server, and it doesn't send any hostnames,
15:31
<johnny>
warren, client-$(ip)
15:31
for default hostnames
15:31
<alkisg>
then I can just put all ltsp123, ltsp124 etc in my /etc/hosts...
15:32
<johnny>
warren, i try to run revisor in fedora10, and it wants me to do someting with selinux
15:32
and won't work :(
15:32
is that the tool i should be using?
15:32
<warren>
I've never used revizor ever
15:32
<johnny>
oh.. ok
15:32
<warren>
what are you trying to do?
15:32
<johnny>
i thought it was the preferred spin creator tool
15:32
<warren>
no
15:32
livecd-tools
15:33
<johnny>
that's a gui thing ?
15:33
warren, http://revisor.fedoraunity.org/ FYI
15:34
<japerry>
ogra: yah I'm thinking having them open source is the key to making it work
15:34
I thought the nvidia drivers were fairly stable, and are for thick-clients.. but something is leaking on thin-client applications
15:35
<warren>
johnny: revizor is a competing project
15:35
<johnny>
oh
15:35
<warren>
livecd-tools is the official thing, it isn't pretty
15:35
revizor is a fork that is pretty
15:35
<johnny>
aha..
15:35
<warren>
revizor does a lot more than livecd-tools
15:35
but not as stable
15:36
<johnny>
thanks warren
15:36
i'm trying to get some huge flash drive
15:36
so i can boot off this fedora10 install wherever i go
15:36
<warren>
why do you need revizor for that?
15:36
<johnny>
i wish i could just carry one big key..
15:36
and have a fat32 partition for portable apps..
15:36
<warren>
you can either do a plain install onto the flash drive, or install the live img with persistent overlay and separate /home
15:37
johnny: https://fedorahosted.org/liveusb-creator/
15:38
<_UsUrPeR__>
johnny: I've done the plain install on a 2.5" USB drive
15:38
<alkisg>
vagrantc, Eghie, I think it would be better if I checked the subnet mask, and if it's /24 => hostname = ltsp254, if it's /16 => ltsp65534 etc... So, hostname = 'ltsp' + 1-to-max-number-of-clients-based-on-subnet-mask.
15:38
<_UsUrPeR__>
worked great until my car pushed the drive off my desk
15:38
:(
15:38
car = cat
15:38
<vagrantc>
alkisg: sounds reasonable
15:39
<Eghie>
alkisg: wouldn't it be better to have a hostname which will stay even when IP changes?
15:39
alkisg: like using the MAC address?
15:39
<johnny>
warren, is there any way you'd want to fllow the same standard here for generated hostnames ?
15:39
follow*
15:39siki has joined #ltsp
15:39
<johnny>
i'd go the same for gentoo
15:39
<alkisg>
Eghie, but how would the server know to map this hostname to an IP adress?
15:39
<warren>
johnny: what standard?
15:40
<Eghie>
alkisg: using it´s MAC address
15:40
<vagrantc>
alkisg: basically, there are probably several ways to configure it, and the part that stalled me was not knowing how to make it configurable from the initramfs
15:40
<Eghie>
alkisg: you can do some kind of "hash calculation" on the mac address
15:40
<johnny>
warren, there isn't one yet.. :)
15:40
i'm just using what you have
15:40
<vagrantc>
Eghie: it would also be nice to have something that was simply readable
15:40
<Eghie>
maybe just the last nrs + first nr's combined
15:41
<alkisg>
Eghie, suppose a very simple dhcp server is used. A client with mac address = X gets an ip of 192.168.0.123. How does the server know what to put in its /etc/hosts file?
15:41
<johnny>
but it'd be nice if we could all use the same function
15:41
<vagrantc>
Eghie: i.e. i can remember 3 digits... so sometimes i boot a client, look at the hostname, and then go check it in the server logs
15:41
<alkisg>
vagrantc, what do you mean? After ipconfig is called, the ip is known, so hostname can be modified (if it is empty, that is)
15:41
<Eghie>
alkisg: using the MAC address of the host getting the IP and combine the first and the last NR's for example
15:41
gives you 4 characters
15:42
<vagrantc>
alkisg: but i've had requests to base it on ip address, on mac address, and numerous other ways...
15:42
alkisg: so how to decide which to use in a configurable way?
15:42
<alkisg>
vagrantc, ok, you can't satisfy them all.. :)
15:42
<vagrantc>
alkisg: exactly
15:43
<Eghie>
hostnames are easy if they can be used even with change of IP address
15:43
<vagrantc>
alkisg: so instead i left it as is, and it's easy enough to include an extra script if they want it
15:43
<alkisg>
vagrantc, I think basing this on mac address is useless in most cases. Let me finish the talk with Eghie for this.
15:43
<Eghie>
MAC addresses don't change that often
15:43
<vagrantc>
alkisg: i.e. if [ "$hostname" = "ltsp"]; then... my_crazy_idea_for_how_it_should_be ; fi
15:43
<alkisg>
Eghie, my point is that the server won't know how to map the given hostname to an ip address
15:44
<vagrantc>
mac address is great if you want consistant information across boots with dynamic ip addresses
15:44
<Eghie>
alkisg: why not?
15:44
<alkisg>
The client will know, but not the server, because it won't know the ip address that the client got
15:44
<vagrantc>
maybe that's not necessary for some uses
15:44
<alkisg>
vagrantc, Eghie, give me an example...
15:44
<Eghie>
alkisg: when the DHCP gives out an IP address, it knows 3 things, the client his IP which he is going to give, Client MAC and the generated hostname
15:45
alkisg: the server does know which MAC is getting which IP
15:45
alkisg: that's what a DHCP server is all about
15:45
<alkisg>
Eghie, the DHCP is not always on the ltsp server... If it is, then they may just use static host entries in dhcpd.conf
15:45
<vagrantc>
alkisg: an inventory system based on mac address?
15:45
<Eghie>
alkisg: even then
15:46
<alkisg>
vagrantc, I lost you! :)
15:46
Eghie, suppose the dhcp server is a simple router
15:46
<Eghie>
ok
15:46
arp?
15:46
<alkisg>
If dhcp server == ltsp server, then there's no point to generated hostnames...
15:46
Eghie, or nsupdate, but we don't have this on the chroot
15:47
<vagrantc>
alkisg: http://bugs.debian.org/483760
15:47
<Eghie>
alkisg: is that not serverside?
15:47
<vagrantc>
alkisg: that's the main use-case i've heard about, other than our simple case at freegeek
15:47* alkisg is reading the bug report...
15:48
<vagrantc>
at freegeek, it just requires some crazy stuff that i wouldn't recommend pushing upstream ... though it's basically based on ip with the trailing /24 part of the ip address... and the rest of the ip address determines the host name
15:49
<Eghie>
you have various ways to communicate your data to the server from the client
15:49jammcq has quit IRC
15:49
<Eghie>
so even things like MAC + IP
15:49
<alkisg>
vagrantc, so an lts.conf option like GENERATED_HOSTNAME='ltsp$a$b$c$d' would satisfy them all?
15:50
<vagrantc>
alkisg: maybe
15:50
<alkisg>
Where $a = ip bytes, or $xyz = mac address bytes...
15:50
<vagrantc>
some sort of templating would probably do... but that seems like a lot of work to implement in initramfs
15:51
<alkisg>
vagrantc, implementing this wouldn't need more than 10 lines, I think...
15:52
<vagrantc>
alkisg: i await the results :)
15:53
<alkisg>
vagrantc, which ones? I still prefer ltsp+ip part based on subnet... :)
15:53
templating?
15:53
<vagrantc>
alkisg: that seems like the most flexible.
15:53
<Eghie>
hmm, another use-case: client1 gets errors and is a system with a lot of problems and has a log under his hostname
15:53
<alkisg>
OK, templating it is. No additions / sums etc thought! :)
15:53
<Eghie>
on reboot client1 gets another ip, so another hostname
15:54
that would mean, logs under another hostname
15:54six2one has quit IRC
15:54
<vagrantc>
right
15:54
<Eghie>
so, system administrator will have difficulties to know that is 1 client which has problems, instead of 2 clients
15:54
<alkisg>
Eghie, you think a templating system based on ip/mac digits is reasonable?
15:55
<Eghie>
alkisg: I do indeed think that
15:55
but for debugging reasons I would default that to MAC
15:55
<alkisg>
OK, it fits my needs too, so templating it is. And defaulting to MAC is not a problem for me.
15:55
<Eghie>
because it's easier to look for a certain machine
15:56
<vagrantc>
alkisg: it needs to at least not explode if lts.conf isn't available, though
15:56
<Eghie>
sed is a powerfull language to use for templating
15:56
<vagrantc>
alkisg: i.e. default back to previous behavior
15:56
<alkisg>
vagrantc, previous behavior? Not default = mac-address ?
15:57
<vagrantc>
alkisg: oh whatever. come up with something and i'll mangle it, if needed :)
15:57
<alkisg>
Heh, ok.
15:57
A last think: should I do it in the initramfs or later?
15:57
I think both can be done...
15:58
<vagrantc>
the initramfs ensures it's early enough
15:58selffik has quit IRC
15:58
<vagrantc>
later gives you more available tools
15:58
<alkisg>
OK... I think the tools will be enough in initramfs
15:58
<Eghie>
on what point logging kicks in?
15:58
<vagrantc>
at freegeek we run a custom init script very early in the process
15:58
Eghie: pretty late in the process
15:59
<Eghie>
vagrantc: then it could be even outside of the initramfs I think
15:59
<johnny>
it is outside of the initramfs in gentoo and fedora
15:59
<alkisg>
There's a hostname.sh somewhere, I'll check when it's called, maybe I'll put it near there
15:59
<Eghie>
but I guess the basic tools you will already have in the initramfs
16:00
so, technically both can, and I prefer early as possible
16:00
<vagrantc>
alkisg: we insert it into hostname.sh at freegeek
16:00
<Eghie>
to keep processes happy which use the local hostname
16:01selffik has joined #ltsp
16:01
<alkisg>
OK, thank you all... good night!
16:02BrunoXLambert has quit IRC
16:02
<Eghie>
good night, alkisg
16:08alkisg has quit IRC
16:09puff has quit IRC
16:13rafaelmf has joined #ltsp
16:13CAN-o-SPAM_ has quit IRC
16:21Eghie has quit IRC
16:29kharloss has quit IRC
16:31selffik has quit IRC
16:32
<stgraber>
warren: does the "groups" command also exist on fedora ?
16:33
<warren>
[root@newcaprica ~]# which groups
16:33
/usr/bin/groups
16:33
[root@newcaprica ~]# rpm -qf /usr/bin/groups
16:33
coreutils-6.12-17.fc10.x86_64
16:33
<stgraber>
warren: I'm looking at a replacement of "/usr/bin/getent group|egrep "[,:]${LDM_USERNAME}(,|$)"|cut -d: -f1| tr '\n' ',' | sed -e 's/,$//g')" that also shows the result of pam_group
16:33
good, so I'll just use "groups" instead, way shorter and similar result :)
16:34
<johnny>
yes..
16:34
<stgraber>
warren: are you a BSG fan ? :) (noticed the "newcaprica")
16:35
<warren>
stgraber: New Caprica is about hope.
16:35
<johnny>
warren you big dork
16:35
i'm not that interested in most sci series
16:37
<warren>
I revoke your dork membership
16:38kharloss has joined #ltsp
16:48Egyptian[Home]1 has joined #ltsp
16:51mikkel has quit IRC
16:55
<Gadi>
stgraber: you should put full path to groups
16:55
otherwise it is a security hole
16:56prpplague has quit IRC
16:57
<dberkholz>
i suspect that if you can inject path like that, you've already lost
16:57
<Gadi>
dberkholz: it is run as the user logging in
16:57
on many distros, someone can easily put a script in ~/bin/groups
16:58
and have it execute first
16:58
<stgraber>
Gadi: are you sure groups is always in /usr/bin/ ?
16:58
<dberkholz>
so the user can put a script in their own home directory and have it execute as themselves?
16:58
<Gadi>
were we sure that getent was?
16:58
<dberkholz>
i must be missing some context here. /me reads up
16:59
<Gadi>
:)
16:59
<stgraber>
Gadi: I'd hope so :) I never like to put a full path when I'm not sure it'll work with all distros
16:59
<Gadi>
I think we erred on the side of caution
17:00
otherwise, I will write a virus to send to ltsp users that downloads a script in ~/bin/groups
17:00
rm -rf $HOME ?
17:00
<dberkholz>
if you're worried about path, explicitly set it at the top of the script
17:00
keeps things a lot more readable imho
17:00
<Gadi>
dberkholz: its not our script
17:00
:)
17:01
though, you could do: ssh..... "PATH=... groups ..."
17:02
<stgraber>
well, I'm fine putting /usr/bin/, it seems to be where it's at least in Ubuntu and Fedora, I didn't know that some distros had ~/bin in their PATH
17:02bobby_C has quit IRC
17:03
<vagrantc>
stgraber: the user can set whatever they want in their path
17:03Egyptian[Home] has quit IRC
17:03
<vagrantc>
#!/bin/sh ; echo root
17:04
the whole localapps infrastructure introduces a few security nightmares...
17:04
<stgraber>
vagrantc: well, the user can also do some LD_PRELOAD in its .bashrc then, overwrite any function and just do whatever they want with it
17:05
<vagrantc>
indeed.
17:05
if /usr/bin/group is standard enough, though, that's a simple precation
17:05
<stgraber>
the problem is that we ask the user to tell us in what group he's
17:06
<vagrantc>
yeah, that is a bit on the sketchy side
17:07
<Gadi>
we ask each user's server what group the user's in
17:07* Gadi is unclear what is sketchy about that
17:08
<stgraber>
Gadi: as long as the user can do LD_PRELOAD every command is unsafe, even with a full path. full path just makes things a bit harder
17:09
<Gadi>
are you sure ssh <command> honors .bashrc?
17:09
<stgraber>
if it doesn't then PATH isn't a problem either
17:10
<Gadi>
good point - perhaps there is a flag to ssh to not import server env
17:11
then, we set our own PATH
17:11
which is the only PATH
17:11
<stgraber>
Gadi: I just tried, with hardy's ssh, .bashrc is executed even it not starting a shell (ssh login@host ls)
17:12
<Gadi>
ah - by default, ssh sets PATH
17:13
(just checked man ssh)
17:13
so, it uses its own compiled-in PATH
17:13
rather than server PATH
17:14
<stgraber>
good feature
17:15randra`offz has joined #ltsp
17:15randra`offz is now known as randra
17:18
<Gadi>
vagrantc: I just found a quotation bug in my xrandr script
17:18
let me fix and push
17:19
<vagrantc>
Gadi: i've taken a glance at it... definitely don't have anything to test the 9-headed monster on... or even two headed monster...
17:20
Gadi: i was also curious why not just use X_MODE_0 instead of XRANDR_MODE_0 ? and then there's XRANDR_NEWMODE_0 ... ?
17:20
<Gadi>
well, first - I am trying to make things clean
17:20
believe it or not
17:20
:)
17:20
in LTSP 4, you could specify X_MODE_0 = <mode> or <modeline>
17:21
<vagrantc>
and i'm just kind of randr clueless, so i'm just trying to figure things out :)
17:21
<Gadi>
in 5, we dropped the support for <modeline>
17:21
<vagrantc>
right
17:21
<Gadi>
but, with xrandr, you can set mode or newmode (modeline)
17:21
so, I thought it would be good to bring that back in
17:21
<vagrantc>
ah.
17:21
<Gadi>
now, in current LTSP,
17:22
the digit in X_MODE_<blah>
17:22
refers to one of up to three modelines
17:22
but, having more than 1 modeline only makes sense when you are forcing a limited subset of modes
17:22
and in practical use, almost everyone uses 1 -> X_MODE_0
17:22
now, with xrandr, it will fail gracefully
17:23
so, we don't limit the modelines to achieve a desired mode
17:23
we simply request a mode
17:23
and add it if need-be
17:23
in my implementation, the digits now refer to displays
17:23
so, multiheads would be: 0, 1, 2
17:23
etc
17:23
<vagrantc>
right...
17:23
<Gadi>
and you could customize modes for each head
17:23
as needed
17:24
<vagrantc>
so the X_MODE_1 != XRANDR_NEWMODE_1
17:24
<Gadi>
exactly
17:24
but, for backwards compat, I make X_MODE_0 do the right thing on display 0
17:24
<vagrantc>
got it
17:25
Gadi: any reason why you implemented that as a for loop ?
17:25
<stgraber>
Gadi: I haven't looked at that script yet but what do you do for multi-head and the Virtual thing ?
17:25
<Gadi>
well, the for loop is to loop through settings for each head
17:25
if specified
17:25
as for "Virtual"
17:25
<vagrantc>
for i in ${X_MODE_0}; do
17:25
??
17:25
<Gadi>
turns out only some drivers need it
17:26
like "via" AFAIK
17:26
but, some, like nvidia, do not need Virtual at all
17:26
<stgraber>
I had to use it on intel and ati
17:26
<Gadi>
and can be done completely with xrandr
17:26
<stgraber>
fglrx and nvidia shouldn't need it
17:26
<Gadi>
so, I figured, if we detect multiple heads connected, try to make it multihead
17:26
<stgraber>
but you need the binary driver then
17:27
<Gadi>
if we need VIRTUAL, then we will need configure-x for that
17:27
vagrantc: ah, that for loop was to loop through the modeline params
17:27
<stgraber>
right, let's hope they'll find a way not to need Virtual soon, at least on intel
17:27
<Gadi>
I wanted an easy way to break it up without using awk/sed
17:28
and t wc
17:28
<vagrantc>
Gadi: which we haven't/didn't support in ltsp5 ...
17:28
<Gadi>
didnt support which?
17:28
modeline?
17:28
right
17:28
<vagrantc>
X_MODE_0 modeline
17:28
<Gadi>
exactly
17:29
but, its easy to loop through and see how many arguments there are
17:29
<vagrantc>
so why not if [ -n "$X_MODE_0" ]; then....
17:29
<Gadi>
if its 1, its a mode name
17:29
more than 1, and its a modeline
17:29
so, as I loop, I count
17:29
<vagrantc>
ok.
17:29
<Gadi>
I can break at 2, I suppose :)
17:29rafaelmf has quit IRC
17:29
<vagrantc>
just want to figure out the logic of it before i go suggesting ways to clean it up :)
17:29
<Gadi>
save a nanosecond or two
17:30
<vagrantc>
about all i think i'll be in a position to test will be MODE stuff
17:30
<Gadi>
stgraber: yeah, I hope all the drivers can adapt
17:31
it seems the ones that have their own ways of doing dualhead, dont need it
17:31
but, the ones that rely on Xorg do
17:31
you know, like some have TwinView(tm), blah blah
17:31
:)
17:31
Via has SAMM
17:31* Gadi wonders why they cannot handle it
17:32
<Gadi>
(or maybe the binary driver can?)
17:32
vagrantc: any testing would be great
17:32
<stgraber>
via doesn't handle randr correctly at all last I checked
17:32
only has a "default" output
17:32
(via = openchrome in this case)
17:33
<vagrantc>
i guess i can actually bother to plug in my i810 laptop, and i've got a disklessworkstations 1220 (i think)
17:33
<Gadi>
stgraber: yeah
17:33
stgraber: I think SAMM is in the binary driver
17:33
but, it is also xorg.conf based
17:33
afaik
17:34
vagrantc: test the one I just pushed, tho
17:35
because the quoting in the first one messed things up
17:35
<vagrantc>
Gadi: sure
17:35
Gadi: pulled already
17:35
<Gadi>
btw: does the ldm login screen in your openbox have the textbox shifted right a bit?
17:36
s/openbox/virtualbox/
17:36
is that on purpose?
17:36
I noticed it only in intrepid
17:36
and I am not sure if it is intentional
17:37
<vagrantc>
i've never measured it exactly, but sometimes it seems off-center, yes.
17:38kharloss has quit IRC
17:43warren has quit IRC
17:51Gadi has left #ltsp
18:16
<Ryan52>
I centered it.
18:16
and I think that's what ogra meant by "breaking" the theme.
18:17
well, my patch that fixes wide logos centers it.
18:17
anyway, after that I decided not to touch the GUI stuff anymore :P
18:20hanthana has joined #ltsp
18:27randra has quit IRC
18:27randra has joined #ltsp
18:27warren has joined #ltsp
18:28staffencasa has quit IRC
18:33CAN-o-SPAM_ has joined #ltsp
18:34
<gbolte>
h,,
18:34
er
18:34
hmm
18:35
nice nick CAN-o-SPAM_
18:35
:/
18:36phantom has joined #ltsp
18:37F-GT has quit IRC
18:48CAN-o-SPAM_ has quit IRC
18:57Q-FUNK has quit IRC
19:19gbolte has quit IRC
19:21japerry has quit IRC
19:23otavio has quit IRC
19:24otavio has joined #ltsp
19:24
<vagrantc>
of course, now that i finally test gadi's xrandr stuff, and it's exploding... nowhere to be found.
19:40
<stgraber>
vagrantc: is that in bzr yet ?
19:41
<vagrantc>
stgraber: not trunk...
19:41
http://bazaar.launchpad.net/~gideon/ltsp/ltsp-trunk-xrandr/
19:42
<stgraber>
ok, looking. I need xrandr support for some computers at the office (only the screen resolution bit, not dual-head yet)
19:43
vagrantc: ok, what lts.conf parameters did you use and what happened (if anything happened) ?
19:45
<vagrantc>
stgraber: XRANDR_MODE_0=640x480 ...
19:45
stgraber: ldm just kept restarting
19:45shogunx has quit IRC
19:46
<vagrantc>
stgraber: it was running "xrandr --output default --mode 640x480"
19:46shogunx has joined #ltsp
19:46
<vagrantc>
i tweaked the script so it would run with --output 0 instead, and it doesn't work, but it doesn't cause ldm to fail either
19:47
<stgraber>
vagrantc: what's the result of "xrandr" on your system ?
19:49
<ltsppbot>
"vagrantc" pasted "xrandr on virtualbox" (5 lines) at http://pastebot.ltsp.org/112
19:50
<vagrantc>
i'll try with real hardware some other time
19:50
<stgraber>
ok, so xrandr --output default --mode 640x480 should have worked
19:50
but I have a theory of why ldm was restarting
19:51
maybe X isn't start with -noreset
19:51
so when the last client exists (xrandr in this case) X exists and is then respawned
19:51
*exits
19:52
<vagrantc>
running it from a logged in session causes X to reset
19:52
<stgraber>
oh, so that's a VB problem then
19:52
or a bug in the video driver
19:56
<vagrantc>
i could also try with qemu
19:57
although i think i'm done for the night...
19:57
<stgraber>
let's just hope it's not the same video driver used by both :)
19:58
I did a quick try on an intel based thin client and "xrandr --output LVDS --mode 640x480" worked well (on Intel the default output is LVDS, not default)
19:58
<vagrantc>
i can fairly easily test virtualbox, qemu, my old laptop and a disklessworkstations 1220 (in order of ease)
19:59
i'll give them all a spin maybe wednesday or thursday
19:59
but definitely wouldn't want to merge it without more testing :)
20:00vagrantc has quit IRC
20:00
<stgraber>
I'll add it to the chroot I have in the office and if I get a minute tomorrow I'll do some test on it with intel, ati, matrox and via video cards
20:00
doh, too late
20:00rafaelmf has joined #ltsp
20:02phantom has quit IRC
20:02F-GT has joined #ltsp
20:25randra has quit IRC
20:35otavio has quit IRC
20:48Hyperbyte has quit IRC
20:51petre has joined #ltsp
21:11
<petre>
warren, ping
21:12
<warren>
?
21:20
<petre>
in the RC1, is the document called "Readme..." or "Quick Start Guide"?
21:21
I can't remember when we decided to change the name
21:28
<warren>
petre: you decided to rename it after RC1
21:28
petre: you are now hereby ordained Product Manager, I'm just the tech
21:29
petre: anyhow, we have to figure out exactly how jetpipe broke before pushing a F9 final
21:29
petre: and immediately get started on a F10 version
21:30
petre: or maybe we should rename RC1 to final and just focus on F10?
21:30
RC1 is pretty good except for the text of that readme
21:32
<petre>
I agree; how hard would it be to make an rc2 with the Quick Start Guide?
21:33
<warren>
petre: not hard, just time consuming
21:33
petre: I'm swamped with other things right now, do you want to be trained?
21:33
petre: but if spinning a new image, might as well fix jetpipe first
21:33
petre: do you have a client and a local printer to plug into it?
21:33
<petre>
I'd be delighted, with the qualifier that my spare time comes in spurts
21:33
<warren>
(That is what jetpipe is for right?)
21:34
<petre>
right
21:34
<warren>
we have to fix jetpipe for both F9 and F10 anyway
21:34
I can probably get to jetpipe on Friday
21:34
<petre>
yes, I've got some laptops I use for clients, and a printer I can test with
21:34
<warren>
at the soonest
21:34
but i don't have a local printer
21:35
<petre>
I don't know if I can get to it before Thursday, but I'll try; perhaps tomorrow evening or Wed. evening
21:35
depends heavily on the plans of the rest of my family
21:36
was jetpipe just not working? or what problem did you hear about?
21:36
<warren>
k12linux-devel-list
21:36
someone complains that the latest F9 update broke his working jetpipe
21:36
but failed to provide any more info
21:36
it is plausible that it broke, possibly for a simple stupid reason
21:37
like the changes that went in related to jetpipe have a Debianism
21:37
I've never used jetpipe before so I can't easily test it
21:38
<petre>
ah, yes, that was William Fragakis; I know him
21:39
there may be more to his config that he mentioned
21:39
<warren>
could you please talk to him? He failed to respond.
21:40
<petre>
he had some systems that he had upgraded from previous Fedora versions
21:40
and then he recently posted his notes about moving from ltsp 4.2 to 5
21:40
<warren>
petre: is the latest version of your docs posted to the bugzilla ticket?
21:40
<petre>
(I was impressed, as I didn't think it would work)
21:40
yes
21:40
<warren>
k
21:41* petre goes to check to make sure he didn't dream posting it
21:41
<warren>
petre: you might also want to put those docs at a URL again and ask k12osn and k12linux lists to review them.
21:41
petre: I am also asking a few people to translate that doc into a few key languages.
21:42
<petre>
yes, it's there
21:42
ok. I've got the HTML version, i.e., not the rpm, on my web server, as I've had to point a few people to it
21:43
I'll post a message to the lists for feedback/review
21:43
<warren>
ok, please post the URL on the list and ask for reviews and opinions
21:43
thanks
21:44
<petre>
and yes, I'll check with William
21:44
<warren>
petre: if you have any time this long Thanksgiving weekend, I'd like to train you on building the live image. It is not hard at all.
21:44
petre: and I would like to restart my attendance at the weekly meetings beginning this Sunday
21:45
<petre>
I've re-built ISOs before, I assume this is somewhat similar
21:45
yeah, I've been on every Sun. morning, but I haven't seen anyone else for a few weeks
21:45
<warren>
the old K12LTSP ISO's?
21:46
the tools are completely different from that
21:46
<petre>
no, ubuntu, and Centos 4.x & 5.x
21:46
<warren>
ok, anyway it isn't hard
21:46
i'll show you
21:46
<petre>
ubuntu for my purposes, centos for my dayjob
21:46
Sounds good
21:46
<warren>
you running F9 or F10 on your main machine now?
21:46
<petre>
F9
21:46
<warren>
ok, that's better than me actually
21:47
I'm not 100% sure F10 can safely make a F9 image
21:47
<petre>
and nfs is broken such that I can't get it to serve the vmclient
21:47
or any other clients for that matter
21:48
<warren>
/etc/exports?
21:48
checked iptables?
21:48
<petre>
but I've got an Athlon64 desktop that I also test on
21:48
<warren>
sometimes you have to restart the nfs service
21:48
<petre>
iptables is off, exports looks good
21:49
I get this when I try to start it:
21:49
Starting NFS quotas: Cannot register service: RPC: Unable to receive; errno = Connection refused
21:49
rpc.rquotad: unable to register (RQUOTAPROG, RQUOTAVERS, udp)
21:49
and yet the portmap-replacement (forget the name ATM) is running
21:50
<warren>
did you restart rpcbind?
21:50
same problem after rebooting?
21:50
<petre>
rpcbind, that's the name; yes, it seems to be running
21:50
oh, yes, I noticed it had stopped working about a week ago
21:50
and haven't been able to figure out why
21:51
so I just switch to the Athlon box and do my work there
21:51
not sure what I did to break it, either
21:53
anyway, do you want to show me building the ISO on Sunday morning?
21:53
I'm pretty sure I can be free then, perhaps even any of the mornings Thur-Sun
21:53
<warren>
Sunday probably good
21:53
<petre>
as I get up pretty early
21:54
<warren>
we'll talk before then
22:04CaScAdE^1arAway has joined #ltsp
22:07CaScAdE^FarAway has quit IRC
22:09hanthana has quit IRC
22:10vagrantc has joined #ltsp
22:12petre has quit IRC
22:36vagrantc has quit IRC
22:41alkisg has joined #ltsp
22:45rafaelmf has left #ltsp
22:58Loto_ has quit IRC
22:59Loto_ has joined #ltsp
23:07japerry has joined #ltsp
23:22warren_ has joined #ltsp
23:23warren has quit IRC
23:23siki has quit IRC
23:24warren_ is now known as warren
23:52alkisg has quit IRC