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


Channel log from 17 April 2008   (all times are UTC)

00:06Egyptian[Home1 has quit IRC
00:31asac_ has joined #ltsp
00:39mccann has quit IRC
00:39ace_suares has joined #ltsp
00:42asac has quit IRC
00:42asac_ is now known as asac
00:50subir has joined #ltsp
01:16deavid has joined #ltsp
01:28mikkel has joined #ltsp
01:36ashes_to_ashes has joined #ltsp
01:53toscalix has joined #ltsp
02:53toscalix has quit IRC
02:53ogra has quit IRC
02:54ogra has joined #ltsp
02:55exodos has joined #ltsp
03:13Q-FUNK has joined #ltsp
04:17toscalix has joined #ltsp
04:20fd has joined #ltsp
04:24fd has quit IRC
04:24fdevienne has joined #ltsp
04:28fdevienne has quit IRC
04:29Q-FUNK has quit IRC
04:40viking-ice has joined #ltsp
05:08Pascal_1 has joined #ltsp
05:26Pascal_Debian has joined #ltsp
05:47Pascal_Debian has quit IRC
05:49Pascal_1 has quit IRC
05:55uwe has joined #ltsp
05:57schregge has joined #ltsp
05:58
<uwe>
hello, im trying to play with ltsp-5, havent played with ltsp for a while, apparently its using nbd root instead of NFS, but i cant find where i define the nbd server ip and port for example before the client can read the rootfs ! since apparently it uses scripts in /etc/init.d to start the client ... and i cant find a document to describe how LTSP-5 works other than details on how packaging and stuff works, any good document on how ltsp+nbd works rather than how
05:58
to configure it?
05:59
<schregge>
hello, is someone running LTSP as a domU in xen? And how is the speed?
06:00indradg has quit IRC
06:09
<exodos>
uwe: you can specify it using nbdroot=ip.ip.ip.ip:port
06:09
<uwe>
as boot option exodos ?
06:10
<exodos>
uwe: if you're not scared of reading bash scripts check /opt/ltsp/i386/usr/share/initramfs-tools/scripts/ltsp-nbd
06:10
yes
06:10
<uwe>
all right! thank you exodos , ill take a look at that
06:10
<exodos>
schregge: we're running it under xen and works ok
06:11
<schregge>
exodus: how well is the 2D performance? (Maybe small flash-videos etc.)
06:12
i just installes ltsp in xen on an old 2 GHz Computet, and the performance is quite bad. (I thing the problem is, that the domU has only a vesa-card)...
06:13
<exodos>
graphic card of the server doesnt matter at all
06:13
have you enabled LDM_DIRECTX?
06:14
you can check this by typing 'echo $DISPLAY'
06:14
<schregge>
i think so. i switched the ssh encryption off
06:15
but i will chek it later, i am not on the server yet
06:15
btw, i read somewhere when you have a good graphic card installed into the server, it can speed the display perfomance....
06:18
And another question: Is it possible to run GoogleEarth or maybe Compiz with LTSP? And how well should be the Server/Clients? (With max. 7 Clients)
06:18muh2000 has joined #ltsp
06:19
<uwe>
exodos, so nbd is built in the kernel, and tries default values of the dhcp server or something like that! and all those scripts are used again in a later stage ... right ?
06:20
<laga>
uwe: $ip_of_server:2000 is the default nbd address
06:20
uwe: you can override that in /var/lib/tftpboot/ltsp/...
06:24
<uwe>
thank you laga
06:25
ill be looking further into that
06:26TelnetManta has quit IRC
06:30subir has quit IRC
06:30
<schregge>
Or how does GLX work? I have not found many information about that. Do i hav speciale 'need' (like Graphic card) on server side or only on client side? (and which is the best low-budget card for the clients(with xgl and xvideo support)? At the moment i have a via chipset onboard...
06:49mikkel has quit IRC
06:55Q-FUNK has joined #ltsp
07:08deavid has quit IRC
07:08TelnetManta has joined #ltsp
07:12schregge has quit IRC
07:38indradg has joined #ltsp
07:41ashes_to_ashes has quit IRC
07:44vagrantc has joined #ltsp
07:45deavid has joined #ltsp
07:50Mikaku has joined #ltsp
07:50
<Mikaku>
hi all
07:51
I'm using ltsp-4.2 with the 2.6.20.9-ltsp-1 kernel version, but now I need to add a new driver for a special network card, and I've never compiled the LTSP kernel and i'm new in this process
07:52cliebow has joined #ltsp
07:52
<Mikaku>
I've followed the steps in http://wiki.ltsp.org/twiki/bin/view/Ltsp/Build-LTSP-42 but once downloaded the lbe tree I've found that the kernel on there is still 2.6.9
07:52
<vagrantc>
Mikaku: which linux distro and release?
07:52
<Mikaku>
how can obtain the same lbe tree but using the latest kernel version used in LTSP-4.2?
07:53
<ogra>
vagrantc, ltsp 4.2 he said :)
07:53
<Mikaku>
vagrantc: well, AFAIK the linux distro is not important for LTSP-4.x versions, just for 5.0
07:53
<vagrantc>
ok, host distro.
07:53
<Mikaku>
vagrantc: anyway, I'm using CentOS 4 and 5
07:53
<ogra>
Mikaku, i'd suggest using ltsp5 if your distro has that
07:53
compiling a kernel on 4.2 is a beat of a task
07:54
<Mikaku>
ogra: I dont like the new way of LTSP project, making a linux distro dependant
07:54
ogra: i still prefer to use LTSP 4.2
07:54slidesinger has joined #ltsp
07:55
<vagrantc>
well, it will only get harder to support 4.2 as time goes on.
07:55
<ogra>
vagrantc, it isnt supported
07:55
neither maintained
07:55
if you are lucky scott or jim come around to ask them ... but thats about ti
07:57
<pscheie_>
Isn't that the point of ltsp5, to allow you to do things like rebuild a newer kernel via your distro?
07:57
<Mikaku>
ok, so no one know where may be the sources for 2.6.20.9-ltsp-1 kernel?
07:57
<ogra>
pscheie_, exactly
07:58
its trivial in ltsp5 (if you even need to, usually newer HW is also better suported by teh newer kernels, needing to compile stuff is less likely)
07:58
<pscheie_>
I like 4.2 as well, and it was GREAT at the time, but it's kinda stuck in its time, not evolving for newer hardware
07:58
<ogra>
the point is as well to have security updates
07:58
<Mikaku>
ogra: I tried ltsp5 but it's all for Debian-based distros
07:58
<ogra>
the 4.2 kernel wasnt touched for 2 years
07:59
<Mikaku>
ogra: no, just only one year
07:59
<ogra>
Mikaku, http://wtogami.livejournal.com/24147.html
08:00
Mikaku, and http://fedoraproject.org/wiki/K12Linux/
08:00
its in its early stages but i'm sure warren would appreciate more testers and help
08:00
<pscheie_>
Mikaku, what you *want* is 4.2, but what you *need* is 5
08:00
but it's not ready on CentOS/Fedora/RedHat yet
08:01
5, that is
08:01
<Mikaku>
yep
08:02
<pscheie_>
Mikaku, ltsp5 can be built on Fedora 8/9 although it's not production ready
08:03* ogra dances
08:03toscalix has quit IRC
08:03
<pscheie_>
I've got it running on F8, but only for testing, sound not working yet, etc.
08:03* ogra finally got suspend/resume on the classmate working
08:04
<Mikaku>
well ok, i'll take a look to this recent ltsp5 for Fedora, but it probably will take me time that I dont have right now, hence I only wanted to compile the kernel with a new driver
08:05
<pscheie_>
Mikaku, unfortunately, the answer you want isn't available today, but is coming soon
08:05
;-)
08:06
<Mikaku>
:)
08:06
hmm
08:06
I'm seeing ...
08:06
the src.rpm is just 100KB size
08:07
too small if it contains all the ltsp/ directory tree :/
08:07
<cyberorg>
Mikaku, you dont need kernel compiling, you need to mkinitrd using warren's modification and create initrd with the modules you need
08:07
<ogra>
thats the prupose of ltsp5 :)
08:08
Mikaku, it contains a build script that builds the tree from redhat packages, later you can just chroot into the tree and install i.e. a compiler and build anew kerel within minutes
08:09
<Mikaku>
hmm ok
08:10
<ogra>
or install a full desktop and configure it to authenticate to your directory server if your clients are powerful enough ...
08:10
or install a mythtv client package in the chroot and have a fanless mediacenter client in your living room
08:10
etc etc
08:12
<Mikaku>
it works under NFS as in 4.x, i guess
08:12
<ogra>
in ubuntu it uses nbd by default (while nfs is still an option) i think the fedora version does only nfs atm
08:13
<Mikaku>
nbd?
08:13
<ogra>
network block devices
08:13
<Mikaku>
ok
08:13
<ogra>
you export an image from the server and on the networked client it appears like a disk
08:14
<Mikaku>
well, it seems it has been compiled under my CentOS 5
08:15
now I have ltsp-client, ltsp-server and ltsp-vmclient RPM packages
08:15
<ogra>
not sure you need the last one
08:15
<Mikaku>
i can see inside the server packages something related to ndbrootd and nbdswapd
08:15
<ogra>
warren would be the best person to help here
08:16
<Mikaku>
it is really lightweight :)
08:16
<pscheie_>
ltsp-vmclient is handy for testing
08:16
<ogra>
you should have the ltsp-build-client script somewhere now
08:16
running that as root should build the cliet tree
08:16
(but dont quote me on that, i have never seen teh fedora implementation)
08:17
<Mikaku>
yes, i can see it with its own .conf
08:17
well
08:17
<ogra>
you shouldt install ltsp-client on the server directly
08:17
that will be ulled into the tree by the script
08:17
it contains the initscripts of the client
08:18
<Mikaku>
it's an rpm so the server probably will need it
08:18mraudsepp has joined #ltsp
08:18
<Mikaku>
even if its internal packages will only be used by the client
08:18
anyway
08:19
now I have a lot of hard work switching all my LTSP4 "infrastructure" to LTSP5
08:19
<ogra>
try the script, it should be done withig 20-30min
08:19tux_440volt has joined #ltsp
08:19
<ogra>
for most of the stuff you shouldnt need to do anything unless yu own a lot serial mice or something the autoconfiguration cant catch
08:20
<Mikaku>
i'm still thinking that just knowing how to add a driver to a ltsp4-based kernel will safe me a lot of time :/
08:20
<ogra>
well, then start reading up about LBE
08:20
learn how to set up a build environment ...
08:20
<Mikaku>
yeah, i'm doing it
08:21
the problem is that the kernel it comes is and old 2.6.9, not the current one 2.6.20
08:21Blinny has joined #ltsp
08:21
<Mikaku>
s/and/an/
08:21
anyway
08:21
<Blinny>
ogra: ping
08:22
<ogra>
Blinny, hey
08:22
<Mikaku>
thanks for all guys, i'll follow the ltsp5 project for fedora
08:22
<ogra>
have fun and good luck
08:22
<Mikaku>
cya
08:22
thanks again
08:22Mikaku has left #ltsp
08:23
<Blinny>
ogra: I recall you saying that the Hardy i386 and x86_64 kernels are combined into one -generic package. I have a situation where when booting from the Hardy i386 LiveCD Desktop, my server sees 8GB RAM. When booting from the latest update of the Hardy kernel, my server sees 3GB RAM, like it's booting i386. I posted a bug and they said it's because I'm using i386 -desktop, and not i386 -server or x86_64 -desktop.
08:24
I wanted to make sure I understood what you were saying before pressing the #ubuntu-kernel guys that this really is a bug.
08:26
My interpretation was that, even when installing Hardy Desktop -i386, if I had a 64-bit system I'd be running a 64-bit kernel, on a i386 userspace, and the 8GB would be recognized just fine.
08:26
<ogra>
the standard -generic kernel is built with the highest level of compatibility in mind
08:26
so the highmem stuff is disabled there
08:27
so you need indeed a -server kernel or an image thats natively built on amd64
08:27
<Blinny>
Ah. I understand now. Thanks - sorry to bother you w/ non-LTSP stuff.
08:27
I appreciate it.
08:27
<ogra>
oh, its fine
08:28
i'm testing CD images today so all issues (even non ltsp questions ;) ) might matter :)
08:28
<Blinny>
Right on. I hope it goes well!
08:29
<ogra>
so far it looks good ... even though i have one ugly ltsp bug i'm not sure i get done before release
08:36tux_440volt has quit IRC
08:39mraudsepp has quit IRC
08:46
<vagrantc>
anyone happen to have a usb floppy drive? still wanting to fix: http://bugs.debian.org/415460
08:46schregge has joined #ltsp
08:46
<ogra>
i have one, it doesnt work with the current kernel
08:47
when it worked it used to register as a plain /dev/sgX device
08:47
<vagrantc>
it's not like a major issue, but it's just one of those lingering bugs i want to get fixed someday
08:47
<ogra>
well, if my thingie works i can try it, but current 2.6.24 doesnt like it
08:52
<vagrantc>
i'm guessing a few usb floppy drives have come through freegeek, but i suspect they're just getting recycled
08:52
<warren>
the main thing needing backporting to get centos working with LTSP5 is a few tiny mkinitrd patches
08:52
<vagrantc>
i've tried to put the word out to look for them
08:53nantes_geek has joined #ltsp
08:54
<nantes_geek>
have anyone around here have patch xorg to support geode thin client and can explain to me how i can do that for a mandriva rpm ?
08:55
<ogra>
nantes_geek, wait for Q-FUNK or Gadi to show up
08:55
<nantes_geek>
ok
08:56
<Q-FUNK>
?
08:56
<ogra>
see above
08:56
<Q-FUNK>
the current -geode 2.8.0 already supports that
08:56
<nantes_geek>
hi martin-eric
08:56
<Q-FUNK>
and it's in hardy
08:56
hi :)
08:56
<ogra>
Q-FUNK, hardy != mandriva rpm :)
08:57
<nantes_geek>
Q-FUNK: you know me, i 'm from Tranquil IT
08:57
<Q-FUNK>
tarball is in X.org FTP
08:57
<ogra>
Q-FUNK, he wants to build a mandriva package
08:57
<Q-FUNK>
ah :)
08:57
yes
08:57
<nantes_geek>
Q-FUNK: it's for running yours
08:57
<ogra>
:)
08:57
<Q-FUNK>
you mentioned that in your last e-mail
08:57
lemme find you the URL to the upstream tarball
08:57
http://xorg.freedesktop.org/releases/individual/driver/
08:58
<nantes_geek>
Q-FUNK: ok
08:58
<Q-FUNK>
http://xorg.freedesktop.org/releases/individual/driver/xf86-video-geode-2.8.0.tar.gz
08:58
<nantes_geek>
Q-FUNK: we was looking at your picture in your website ^^
08:58
<Q-FUNK>
you'd need to upgrade the SPECS file from the last -amd RPM to build this one
08:58
<nantes_geek>
ok
08:58
<Q-FUNK>
oh? what about my picture? :)
08:59
<warren>
Q-FUNK: have you seen any signs that the libddc patch causes a regression?
08:59
<Q-FUNK>
or use the RPM that Fedora just made
08:59
<warren>
Q-FUNK: (after folks ran ./autogen.sh)
08:59
<Q-FUNK>
warren: do you have a URL to your -geode RPM for these nice Mandriva folks?
09:00
no, after I regenerated the files, it built fine and it works.
09:00
<nantes_geek>
Q-FUNK: i don't have to patch the x86emu stuff ?
09:00
<warren>
http://koji.fedoraproject.org/scratch/wtogami/task_568943/
09:00
<Q-FUNK>
nantes_geek: it depends whether the X in Mandriva has merged them or not.
09:01
warren: in fact, this libDDC works so well that we might be in a position to stop licensing commercial BIOSes altogether and build everythng with coreboot.
09:01
<ogra>
does it work with the binary drivers ?
09:02
<nantes_geek>
mmh perhaps i'll try to remake my nfs tree for ltsp with a fedora
09:02
if fedora have ever patch the Xorg
09:02
<Q-FUNK>
ogra: which binary drivers?
09:02
<ogra>
Q-FUNK, nvidia, fglrx
09:02
<warren>
Q-FUNK: I don't have any boxes with coreboot in order to make sure it works out-of-the-box here.
09:02
<Q-FUNK>
nantes_geek: warren would know what is the current status of Fedora.
09:02
<ogra>
Q-FUNK, or did i misunderstand and is libDDC a geode only thing
09:03
<nantes_geek>
warren: what version of fedora i can use to make my tree ?
09:03
<warren>
nantes_geek: fedora 9 is best currently
09:03
<nantes_geek>
warren: ok, i try and back
09:04
<Q-FUNK>
ogra: libDDC is a library offered by X.org core that can be used by drivers to probe DDC pins without being dependant upon BIOS environment answers.
09:06
<ogra>
Q-FUNK, right, and the binary drivers have hardwired their access to the bios, so i wonder what libDDC does in such a case
09:07
<Q-FUNK>
well, what happened is that upstream added libDDC support to the -geode driver. there's test packages in my PPA.
09:11
<warren>
weren't the x86emu patches related to bios querying of ddc?
09:11
if that's bypassed now, those patches might not be needed?
09:12
<daduke>
vagrantc: hola. Has anything changed with local USB storage? They're not mounted any more... will debug it tomorrow.
09:12nantes_geek has quit IRC
09:13
<Q-FUNK>
warren: good question. I'd be curious to try this patched driver on top of e.g. a vanilla Gutsy without the X core patches
09:13
<ogra>
daduke, https://bugs.launchpad.net/ubuntu/+source/ltspfs/+bug/218231
09:13
<vagrantc>
daduke: not with LDM_DIRECTX ... hoping to put some time into fixing it this weekend.
09:14nantes_geek has joined #ltsp
09:14
<vagrantc>
daduke: we pretty much know what needs to be done to fix it ...
09:14
<daduke>
ogra: oh, thanks. I didn't make that connection...
09:14
vagrantc: cool. I'll be your beta tester if you need one.
09:14
<vagrantc>
daduke: just need the various ltspfs hooks to know that LDM_DIRECTX=true is set and handle the display differently...
09:15
<daduke>
vagrantc: I see. thanks already. some ppl have asked...
09:16
<vagrantc>
daduke: i had sworn i had tested that it still worked, but i must have made some mistake or something.
09:16
because it sure doesn't work now.
09:17
<daduke>
vagrantc: I also couldn't figure it out properly, because it did work for some ppl but not for others. But now it seems that some clients haven't been restarted in quite some time and still had directx=false
09:17
<Q-FUNK>
nantes_geek: does it work any better?
09:18
<nantes_geek>
so i download the fedora 9 netinst ^^
09:18
<Q-FUNK>
warren: mind you, the patches are still needed overall, because x86emu really _is_ buggy code.
09:19
warren: just that this libddc patch might somehow make things more predictable in the case of a bad bios.
09:19
<ogra>
vagrantc, the only way you could make it work in the current state would be to add -a to ltspfsd
09:20
but that drops all authentication
09:20
<Q-FUNK>
but with x86emu being horrible regardless, there's no telling what other ugly bugs are hidden in there that will only surface later.
09:21
<daduke>
vagrantc: your question about the dual x86/ppc LTSP boot cd: tarzeau said this system doesn't work atm, but I'll have a short look anyway
09:21
<warren>
Q-FUNK: so are we aware of ANY regressions with the libddc patch?
09:21bengoa has joined #ltsp
09:22
<Q-FUNK>
warren: none so far
09:22
<warren>
I wonder if I should just stuff it into Fedora 9 right now.
09:22
<Q-FUNK>
it in fact seems to make operation more consistent, being independant from BIOS info
09:22
warren: I'd say monitor the list for any possible regression and if nobody reports any major catastrophe, let's do it.
09:23Pascal_1 has joined #ltsp
09:23
<Q-FUNK>
warren: when is your target release date, again?
09:23
<warren>
Q-FUNK: uh... next week
09:23
the thing is, geode is pretty hosed without this patch on all 2 of my geode hardware
09:23
<vagrantc>
daduke: if you didn't care about security, do as ogra says and add -a to ltspfsd in /etc/init.d/ltsp*
09:23schregge has quit IRC
09:23Pascal_1 has quit IRC
09:24
<daduke>
vagrantc: thanks. If you manage to fix it soon, I'll wait, otherwise I might do that.
09:25
<vagrantc>
daduke: well, i'm hoping to fix it in sid this weekend. don't know how long till i can test backports.
09:25
daduke: i've already committed to a couple other things this week
09:25
<daduke>
vagrantc: exciting stuff coming all the time, hm?
09:27schregge has joined #ltsp
09:27* daduke is outta here for today
09:29
<Q-FUNK>
warren: for libddc, I wouldn't be surprised if we end up with a -geode 2.9.0 by Monday or so, if nobody reports anhy major breakage using the current patch.
09:30
warren: but I would still let this sleep over the weekend, to give people time to catch up with the patch and report any possible regression.
09:31
<warren>
the autogen.sh thing might be confusing a lot of people
09:31
it builds successfully without doing that
09:33
<vagrantc>
hmmm... well, my virtualbox thin client finished it's bootchart in about 70 seconds ... don't quite know what to make of the chart.
09:35uwe has quit IRC
09:36mccann has joined #ltsp
09:38Arauto has joined #ltsp
09:39
<Q-FUNK>
warren: it builds successfully, but then ddc doesn't work.
09:39
<warren>
Q-FUNK: yes, which makes people think it is a regression
09:39
<Q-FUNK>
warren: you need to autogen.sh to rebuild Makefile.in
09:39
<warren>
*I KNOW THAT*
09:39
<Q-FUNK>
it's not. it's a patch. it requires doing steps that are normally already done by upstream when they release a tarball.
09:40
steps which, in my hasty build, I had forgotten to take, myself.
09:40
after Jordan pointed this out, it suddenly worked
09:57FLOLO has joined #ltsp
10:02bengoa has left #ltsp
10:02abengoa has joined #ltsp
10:03schregge has quit IRC
10:08
<warren>
Q-FUNK: what version of X server is your testing with libddc patch?
10:10
<ogra>
hardy has 7.3+10ubuntu10
10:10
<Q-FUNK>
yup
10:12FLOLO has quit IRC
10:13FLOLO has joined #ltsp
10:16
<FLOLO>
hi, is it possible to log in via ldm with a shell like /bin/false ?
10:17
<vagrantc>
shouldn't be
10:17
<FLOLO>
ok thanks
10:21
<warren>
kernel-2.6.25 in F9!
10:21staffencasa has joined #ltsp
10:33FLOLO has left #ltsp
10:56Pascal_1 has joined #ltsp
10:59
<ogra>
warren, lucky you
10:59
"Linux and open source software giant Red Hat has abandoned plans to develop a consumer desktop product because it cannot compete with the might of Microsoft."
10:59
o_O
11:01
<warren>
ogra: the headlines are wildly inaccurate
11:01
<ogra>
seems your PR team needs training then
11:01
thats really bad press
11:02
<vagrantc>
happens to the best
11:02
bad press, that is
11:04
<ogra>
yeah
11:04
but its also a matter how you sell it
11:04
<pscheie_>
ogra, where'd you see that?
11:04
<ogra>
http://www.channelregister.co.uk/2008/04/17/red_hat_abandons_consumer_desktop_market/
11:04
<Blinny>
slashdot
11:06
<Q-FUNK>
ogra: btw, does the patched driver I uploaded to my PPA work with the Etherboot model you have?
11:06
<warren>
http://koji.fedoraproject.org/koji/getfile?taskID=570744&name=build.log
11:06
hmm... ltsp build fails on ppc due to flex/bison
11:07
I've seen this before, I have no idea what causes it.
11:07
and a second try built successfully
11:07
This might be a makefile race condition
11:09abengoa has quit IRC
11:17
<ltsppbot>
"vagrantc" pasted "add_fstab_entry patch for LDM_DIRECTX" (25 lines) at http://pastebot.ltsp.org/504
11:18
<johnny>
vagrantc, how many of your feature reposotories haven't been merged?
11:18
<vagrantc>
ogra: i think that basically will do it. requires us to modify ldm to insert the DISPLAY name in the socket path before or instead of the vtN stuff
11:19
johnny: there's the ltspfsd/cdpinger from udev branch ... there's the ltsp_chroot function branch ... not sure if there's another
11:19
<johnny>
you have a quite a few in your dir, it's hard to know which has or hasn't been integrated without manually inspecting them
11:19
and diffing..
11:19
<vagrantc>
johnny: which dir are you talking about?
11:20
<ogra>
vagrantc, geez, i'll try that later tonight
11:20
<johnny>
http://llama.freegeek.org/~vagrant/bzr/ltsp/features/
11:20
<vagrantc>
ogra: and we'll have to do something similar to the rc.d stuff
11:20
ogra: it's kind of ugly and hackish ...
11:21
johnny: most have either been merged or abandoned
11:21
<ogra>
but looks like something i can get past mr. langasek :)
11:22gar0t0 has joined #ltsp
11:22gar0t0 has left #ltsp
11:24
<vagrantc>
ogra: this makes me think both add_fstab_entry and delayed_mounted should be calling different stuff
11:25
ogra: oh, and remove_fstab_entry ... yes, i think we need a common function or binary.
11:25
ogra: they all basically do the same things ...
11:30exodos has quit IRC
11:31
<ogra>
yeah
11:39Gadi has joined #ltsp
11:44nantes_geek has quit IRC
11:44nantes_geek has joined #ltsp
11:44tux_440volt has joined #ltsp
11:45nantes_geek has quit IRC
11:45nantes_geek has joined #ltsp
11:49nantes_geek has quit IRC
11:49nantes_geek has joined #ltsp
11:56riggs498 has joined #ltsp
11:57
<riggs498>
Is there anyone around to answer an scanner question for me?
11:58
<nantes_geek>
riggs498: i can try
12:01
<riggs498>
I have done really good. I got everything going on the client with printing and cdroms etc. I have the scanner hooked to the client. I can not get the scaning program to see the device. If I hook it to the server it works.
12:01
I have installed sane on the client and that didn't work. Do I have to do something to lts.conf?
12:02
<pscheie_>
Gadi, ping
12:04Q-FUNK has quit IRC
12:13BUcky has joined #ltsp
12:17
<Gadi>
pscheie_: pong
12:18
riggs498: LTSP 4.2 or 5?
12:18
<pscheie_>
you still with symbio?
12:18
<Gadi>
but, of course
12:18
:)
12:19* pscheie_ was off #ltsp for a while
12:19* Gadi has my head buried a lot too
12:19
<Gadi>
:)
12:19
<pscheie_>
Alan Hodson, from K12ltsp list mentioned he's been trying to contact symbio-technologies via email but getting no response
12:19
<Gadi>
really?
12:20* Gadi has not been following the listserv
12:20
<Gadi>
I have seen some of alan's emails but not all
12:20
it seems one of his email addys may be being blocked
12:20
<pscheie_>
looking for clusters of server & 3/4 clients, supporting Windows apps
12:20
<Gadi>
or sent to spam
12:20
<pscheie_>
message came through last night
12:21
<Gadi>
hmm..
12:21
ok
12:21
weird
12:21
did he email the listserv?
12:22
im seeing nothin
12:22
<pscheie_>
http://www.redhat.com/archives/k12osn/2008-April/msg00175.html
12:22
<Gadi>
can you please let him know that we are not getting the emails from whatever addy that is
12:22
?
12:22
I do not have the listserv setup atm
12:22
<pscheie_>
will do
12:22
<Gadi>
thanks so much
12:22
and apologize from me, too pls
12:23* Gadi hates it when emails don't get thru
12:23
<Gadi>
its like those dropped cell phone call commercials
12:23
everybody gets pissed
12:23
<pscheie_>
Should I tell him you'll try to track him down?
12:23
<Gadi>
yeah - I'll send him an email and see if he can respond to it
12:25
<riggs498>
LTSP 5
12:25
Gadi LTSP 5
12:26
<Gadi>
riggs498: too bad, in 4.2 it was a simple lts.conf param - in 5 there is no OOTB support
12:26
you need to do a few things:
12:27
1. Install openbsd-inetd in the chroot
12:27
2. Setup network scanning in sane backends
12:27
3. configure inetd for network scanning
12:27
<johnny>
hmm.. that's kinda sad
12:28
<Gadi>
4. On the server, add an environment variable to the session that directs scanning to the thin client IP
12:28
<ogra>
yeah, we should ahve a package for that
12:28
<joel_>
Hello, everyone! I am lucky to have gotten an LTSP server set up, however, when I do a PXE client boot, the result is a "(initramfs)" command prompt. Any suggestions?
12:28
<vagrantc>
doesn't sound difficult to implement.
12:28
<Gadi>
johnny: yeah, there are a few regressions from 4 to 5 - since 5 was a complete rewrite
12:28
<johnny>
would be nice to have a list
12:28
<ogra>
vagrantc, apt-get install ltsp-scanner-client :)
12:28
<johnny>
launchpad blueprint?
12:29
<ogra>
johnny, if you like :)
12:29
<vagrantc>
i'm guessing "2. Setup network scanning in sane backends" is the tricky part.
12:29
<Gadi>
nah thats easy
12:29
<ogra>
yeah
12:29elisboa has quit IRC
12:29
<ogra>
its all easy
12:29
<johnny>
just convoluted..
12:30
<ogra>
so easy that its a shame we dont have it
12:30
<Gadi>
the tricky part is the session env var - which is prolly best done with an rc.d script
12:30
<vagrantc>
is sane big enough that it would be bad to install by default?
12:30
<Gadi>
sane is optional enough that it is
12:30
:)
12:30
<johnny>
it includes a bunch of scanner drivers..
12:30
<ogra>
Gadi, needs to go to the session with IP of the client, right ?
12:30
<Gadi>
ogra: yeah - SANE_NET_HOST or some such
12:30elisboa has joined #ltsp
12:30
<Gadi>
its been a while
12:30
<vagrantc>
Gadi: can either stick it in /etc/X11/Xsession.d/ or you would probably have to modify ldm
12:30
<ogra>
yup
12:30
for me too
12:31
<riggs498>
I might just have to leave that one alone ad just do that one on the server
12:31
<ogra>
vagrantc, rc.d script :)
12:31
<vagrantc>
ogra: how do you set an environment variable in an rc.d snippet ?
12:31
<ogra>
hmm, right, export doesnt work
12:31
<vagrantc>
ogra: needs to be in the same place that DISPLAY is set for LDM_DIRECTX
12:32
<ogra>
yup
12:32
well, no biggie
12:32
two lines to change in ldm
12:32
<Gadi>
anyway, I think it should be an optional add-on
12:32
<vagrantc>
are the rc.d scripts run before the Xsession script is called?
12:32elisboa has quit IRC
12:32
<Gadi>
vagrantc: yes
12:33
<johnny>
hmm.. wouldn't it be nice ot have those vars come from a shell script?
12:33
<vagrantc>
maybe we need a mechanism for rc.d scripts to set environment variables for the initial login
12:33
<Gadi>
you can always populate a ~/.profile
12:33
<johnny>
yes..
12:33
<Gadi>
(yuck)
12:33
<riggs498>
Gadi thanks for the information. I think that one will just have to wait till I have days to figure it out
12:33
<ogra>
Gadi, addon but installing something like ltsp-scanner-client should get it up and running rght away
12:33
<Gadi>
riggs498: np
12:33
<johnny>
/etc/ltsp/loginvars (bad name)
12:33
<Gadi>
ogra: right
12:33
<vagrantc>
Gadi: why not just supported by default?
12:34elisboa has joined #ltsp
12:34
<Gadi>
vagrantc: bec some of us may implement it differently
12:34
and there will for sure be more addons
12:35* Gadi thinks of packages for video cameras, cd burning, etc
12:35
<Gadi>
all of which I may not want in my default chroot
12:35
<ogra>
hal will solve a lot of the headdache there
12:35
<vagrantc>
Gadi: well, we can add the code to ltsp-client-core, and add the requisite deps to ltsp-client ...
12:35
<Gadi>
or I would prefer the distro to decide what is in my default chroot
12:35
so, perhaps debian has a bigger chroot by default than ubuntu
12:35
<ogra>
if we have hal its just a matter of apps being installed or not
12:36
<vagrantc>
the ltsp-client/ltsp-client-core split allows us to have defaults with everything, but makes it possible to install other stuff.
12:36
er, makes it possible to install less if needed
12:37
<johnny>
bzr question for ya'll.. what is the appropriate way for me to merge recent stuff from ltsp-trunk into my ltsp-gentoo branch ?
12:37
<Gadi>
sounds good
12:37
<johnny>
in a way that will best promote you pulling it in later
12:37
<Gadi>
I, personally, and quite happy that ltsp-server and -standalone are two different pkgs
12:37
(by analogy)
12:37
<johnny>
that's different..
12:37
one installs stuff that could break your setup if youre not careful..
12:38
ie: dhcp server
12:38
<vagrantc>
not on debian :)
12:38
<johnny>
your ltsp-server-standalone doesn't depend on a dhcp server?
12:38
<vagrantc>
johnny: well, how big is the diff so far?
12:39
johnny: it depends on one, but doesn't take over the configuration of the dhcp server
12:39
<johnny>
yes, neither does ubuntu's
12:39
<vagrantc>
johnny: default on debian is for the dhcp server to do nothing
12:39
<johnny>
sure.. but it's there to be started, even if you might break your dhcp server
12:39
<vagrantc>
johnny: ubuntu has a patch which uses /etc/ltsp/dhcpd.conf if present instead of /etc/dhcp3/dhcpd.conf
12:39
<johnny>
other dhcp server on the network*
12:40
<ogra>
vagrantc, well, it will only do anything if you configured a static interface the right way for it
12:40
<johnny>
that makes it more safe..
12:40
<ogra>
else it will just silently not start
12:40
<vagrantc>
ogra: or if you configured a static interface close enough for dhcp to start and break your network
12:40
<johnny>
fedora's method is interesting btw..
12:40
me.. no ltsp-server-standalone
12:40
<vagrantc>
debian just requires administrator intervention in order to configure dhcp.
12:41
<johnny>
emerge -pv ltsp-server
12:41
ltsp-server -dhcp +nfs
12:41
<vagrantc>
johnny: regarding bzr merging ...
12:42
johnny: you should just be able to merge and it shouldn't interfere with merging you back upstream.
12:42
johnny: but in any case, ideally get merges in somewhat regularly to keep the diff small
12:42
<johnny>
there are no mods to files in ltsp-trunk outside the ltsp-build-client, server/configs/Gentoo, client/initscripts/Gentoo
12:43
<vagrantc>
johnny: also, make a new branch off trunk once your changes have been merged.
12:43
<johnny>
well.. i tried to run merge.. and now it wants me to commit those changes
12:43
as a seperate change
12:43
<vagrantc>
well, you don't have to make a new branch, but it looks a little cleaner
12:43
<johnny>
that didn't make sense to me
12:43
<vagrantc>
that's how it works.
12:43
<johnny>
really?
12:43
what kinda design is that/
12:43
<vagrantc>
no, i'm telling you lies.
12:43
:P
12:44
that's how it works.
12:44
<johnny>
sorry.. it's just confusing
12:44
i'm trying to reconcile it with my vcs world view
12:44
<vagrantc>
you commit is basically saying "this is how i resolved the merge of this collection of patches"
12:44
<johnny>
which is mostly monotone
12:44elisboa has quit IRC
12:44elisboa has joined #ltsp
12:44
<johnny>
ok.. it sounded like a history breaking operation
12:45
that's why i was confused
12:45
<vagrantc>
johnny: do you have anything worth pushing upstream yet?
12:45
sounded like you have something working
12:45
<johnny>
it mostly builds a client..
12:45
95% of one
12:45
works iwth nfs root
12:46
the main blocker for total workingness, is figuring out the initramfs situation
12:46
<ogra>
johnny, http://blogs.gnome.org/jamesh/2006/04/23/repositories-in-bzr/ look at the pic on the right
12:46
<johnny>
i'm taking another approach atm
12:46
<ogra>
the blue lne is -trunk upstream
12:47tux_440volt has quit IRC
12:47
<ogra>
for new features you make a branch ... that gets merged back at some point (one of teh black tracks)
12:48Q-FUNK has joined #ltsp
12:48
<johnny>
i'm used to that style..
12:48
<ogra>
if i have a separate branch for my release or so ... i can just grab your branched feature and merge it into mine
12:48
<johnny>
i've worked with distributed cvs before, but mostly bitkeeper and monotone
12:49
<ogra>
without having -trunk inbetween ...
12:49
<johnny>
bzr seems kinda half assed to me at this point :(
12:49
<ogra>
its far beyond everything i know personally
12:49
<johnny>
i read their docs, and their blueprints.. and it looks like the ydon't learn lessons from other vcs.
12:49
didn't seen any mention of.. here's the approach these other guys have taken
12:50
<ogra>
s/learn lessons/make the same mistakes/ ?
12:50
<johnny>
not how i see it
12:51
most especially the bzrarchives blueprint..
12:51
but i digress..
12:51
ok.. i'll do what vagrantc said
12:51riggs498 has quit IRC
12:52
<vagrantc>
pull/push can be used in a history-breaking way.
12:53
<johnny>
i was pretty happy with monotone .. but being written in C++ makes it harder for other folks to get involved , so it isn't evolving as fast as these others
12:54
<vagrantc>
if i ever wrap my head around git...
12:54
<johnny>
git isn't too difficult
12:54
i'm using it right nwo for our equivalent of your ltsp-debian-packaging branch
12:54
<vagrantc>
every single time i've tried to learn it, i've come away frustrated.
12:54
<ogra>
yeah, thats gits problem
12:54
<johnny>
vagrantc, i got a link for yoU!!!
12:54
read this..
12:55
http://tomayko.com/writings/the-thing-about-git
12:55
<ogra>
it might even be technically advanced over bzr, but that doesnt really weight out usability
12:55
<johnny>
i haven't found bzr to be usable for me at all
12:55
had to read the entire user manual to figure it out
12:55
it didn't jive with all my version control readings
12:56
<vagrantc>
bzr may be a bit slow, and has that obnoxious pull/push bug that can break history, but it makes total sense to me and is so easy to use.
12:56
<johnny>
monotone didn't?
12:56
<ogra>
well, hwen i started with bzr the userguide was probably 50 lines :)
12:56
one wikipage ...
12:57
<vagrantc>
johnny: your git url is not working for me. figures. :)
12:57
502 error
12:57
<johnny>
that's just a webpage.. :(
12:57
<vagrantc>
"502 Bad Gateway" is all i see.
12:57
<johnny>
http://64.233.169.104/search?q=cache:nHF0vXFwiCAJ:tomayko.com/writings/the-thing-about-git+http://tomayko.com/writings/the-thing-about-git&hl=en&ct=clnk&cd=1&gl=us
12:58joel_ has quit IRC
12:58
<johnny>
this feels odd to me to b e promoting git.. as i barely even use it.. but i found that read insightful
13:01
bzr does have the shelve and rebase plugins now obviously
13:01
that page was amended to include them
13:01
<vagrantc>
i use shelve extensively
13:04Q-FUNK has quit IRC
13:04Q-FUN1 has joined #ltsp
13:04Q-FUN1 is now known as Q-FUNK
13:08
<vagrantc>
johnny: that just sounds like what i do with bzr regularly with a lot more steps
13:09
<johnny>
ther'es one part where he reiterates the process i can see
13:09
<ogra>
johnny, for me personally the most compelling reason for bzr is that ubuntus masterplan is to drop all source packagng in favore of bzr branches :)
13:09K_O-Gnom has joined #ltsp
13:11leio_ has joined #ltsp
13:13
<johnny>
ogra, i know that
13:14
it's one of the reasons i am looking at it in the first place
13:14
<ogra>
i'm very much looking forward to that
13:14
<johnny>
i would probably choose bzr over git..
13:14
for my users
13:15
but.. i don't know about choosing bzr over monotone
13:15
having trouble with that decision right now for our other projects
13:15viking-ice has quit IRC
13:19
<johnny>
maybe i'm crazy.. but i like my repo data to be in sqlite.. :)
13:20
vagrantc, i want my branch to be acceptable by dberkholz before it is requested to be merged in ltsp-trunk
13:21
i'm almost there i think
13:21
<vagrantc>
johnny: ah, ok.
13:21
good to have peer review, for sure.
13:21
<ogra>
great habit !
13:21
<johnny>
it is definitely to my advantage to have deployed ltsp in RL before doing this port
13:21
thanks ogra/vagrantnc :)
13:22
now i just understand it more
13:22
ubuntu at work.. gentoo at home.. so that's how it goes :)
13:23
oh.. and warren for removing the debianisms..
13:23
i'm glad he started the ltsp port before i did..
13:23
<warren>
I didn't remove any
13:23
I just pushed them aside and used crippled scripts I wrote myself
13:23
<johnny>
blog post proof?
13:24
i just quote you. that's all
13:24
<warren>
don't trust him
13:24
that guy in the blog
13:24
<johnny>
it's your blog.. :)
13:24leio has quit IRC
13:24
<johnny>
so.. newest meme on planet.gnome.org... be luis villa
13:24
this is my favorite meme
13:27
<ogra>
haha
13:31
<warren>
fixed race condition
13:31
it was in client/getltscfg/Makefile
13:35* vagrantc looks forward to 5.1.5
13:35
<vagrantc>
:)
13:37jammcq has joined #ltsp
13:37
<jammcq>
bom tardes LTSP friends
13:37
err
13:37
boa terdes
13:37Pascal_1 has joined #ltsp
13:38
<ogra>
oh, wow, encrypted filesystem install works :)
13:38
<jammcq>
cool
13:38* ogra is testing ubuntu cds
13:39
<ogra>
seems we have encrypted lvm2 as option in the FS defaults now
13:43
<warren>
vagrantc: perhaps some of the options should be removed from lts-parameters.txt
13:43
vagrantc: ?
13:43nantes_geek has quit IRC
13:43
<warren>
vagrantc: XF86CONFIG_FILE
13:44bengoa has joined #ltsp
13:44
<warren>
vagrantc: and LDM_REMOTECMD is the only thing mentioned LDM related
13:45
<Pascal_1>
hello
13:45
hello vagrantc any news about ldm bug ?
13:46
<vagrantc>
Pascal_1: like i've said every time, i'd update the bug report.
13:46
Pascal_1: i'm not actively working on it, as i don't have an environment to test it on.
13:50
<Blinny>
Yuck. My hardy box's pam suddenly is broken. PAM unable to dlopen(/lib/security/pam_smbpass.so) then sshd[7075]: pam_unix(sshd:session): session opened for user ppi by (uid=0) and finally sshd[7075]: error: Failed to allocate internet-domain X11 display socket.
13:52
<vagrantc>
warren: well, people often come across XF86CONFIG_FILE online, and i figured it would be better to explicitly mention it as deprecated.
13:53
<warren>
vagrantc: it isn't mentioned as deprecated
13:53
<ogra>
vagrantc, erm, your fix cant work
13:53johnny has quit IRC
13:53
<ogra>
case ${LDM_SOCKET} in
13:53
<vagrantc>
warren: well, then we should mention it as deprecated
13:53
<ogra>
ldm socket always exists
13:53
<vagrantc>
ogra: and?
13:54
i'm not comparing weather it exists or not, i'm comparing the name of the socket
13:54
<ogra>
/var/run/ldm_socket_vt* is always true
13:55
<vagrantc>
ogra: we have to modify ldm to create /var/run/ldm_socket_$DISPLAY_vtN_$SERVER
13:55
<ogra>
meh
13:55
why dont you just match LDM_DIRECTX ?
13:55
<vagrantc>
then we'd have to source ltsp_config from the udev scripts
13:56BadMagic has quit IRC
13:56BadMagic has joined #ltsp
13:56
<vagrantc>
and implement boolean_is_true
13:57
which means that ltspfsd would have to depend on ltsp-client-core ...
13:58
which i'd like to avoid upstream. but it might be ok for a hardy bugfix
13:59
<ogra>
i really wonder if all that effort is needed
13:59
<vagrantc>
all what effort?
14:00BUcky has quit IRC
14:00
<ogra>
i'm looking fr a lighter fix
14:00
<vagrantc>
that's the best i've come up with thus far ...
14:02
need a way to distinguish LDM_DIRECTX while not necessarily having access to the variables.
14:04Pascal_Debian has joined #ltsp
14:04Pascal_Debian has joined #ltsp
14:05
<vlt>
Hello. Any idea where to put a command I want to run on every client start when putting it in the chroot's /etc/rc.local doesn't work?
14:06
<vagrantc>
vlt: RCFILE_01=/path/to/file, RCFILE_02=/path/to/other/file, etc. in lts.conf
14:07
<ogra>
vlt, /etc/rc.local needs to be executable
14:07
it isnt by default
14:07
<vagrantc>
well, actually, that depends on which version of LTSP you're using ...
14:07
i think 4.2 used relative paths to /etc
14:07
<ogra>
right
14:07
<vlt>
ogra: Thank you.
14:09
ogra: hmm, it *is* executable ... maybe I'll try the lts.conf way ...
14:14
<ogra>
vagrantc, DISPLAY=$(usr/bin/ssh -S ${LDM_SOCKET} ${SERVER} "xauth -f ~/.Xauthority list|grep -v unix|cut -d" " -f1")
14:14
how about that ?
14:15
<warren>
how often would it have to run that?
14:15
<ogra>
at mount of a device
14:15
<vlt>
vagrantc: RCFILE_01 works, thanks.
14:16
<ogra>
warren, not often
14:16viking-ice has joined #ltsp
14:16
<vagrantc>
ogra: so a second ssh call for every device insertion, removal, or delayed mount?
14:17
<ogra>
i'm not sure we need it at removal
14:17
<warren>
Is it really a good thing to add a ssh call (especially the latency) in order to get something that is already elsewhere local?
14:17
<ogra>
thats the point, it insnt
14:17
<vagrantc>
ogra: i think ltspfsmounter checks for the DISPLAY
14:18
<ogra>
vagrantc, for a display
14:18
for unmounting you dont need ltspfsd access
14:18
<vagrantc>
warren: we can get the information locally, but it requires some convoluted steps to do so.
14:19
ogra: i think ltspfsmounter bails if it can't connect to DISPLAY
14:19
weather it should is another story.
14:19
<ogra>
warren, i currently just look for the last intrusive fix with the smallest code footprint
14:19
to get inot hardy
14:19
into
14:19
vagrantc, to *a* display
14:20
sh -X gives us one in any case
14:20
ssh
14:21
<warren>
vagrantc: where's your branch that does this again?
14:21
DISPLAY=$(usr/bin/ssh -S ${LDM_SOCKET} ${SERVER} "xauth -f ~/.Xauthority list|grep -v unix|cut -d" " -f1")
14:21
<ogra>
there is no branch
14:21
<warren>
~/.Xauthority is already contained in /root/.Xauthority of the client
14:21
<ogra>
we're discussing a fix
14:21
<warren>
in the LDM_DIRECTX case
14:21
<vagrantc>
warren: does what?
14:21
<ogra>
does not have the needed data
14:22
(look yourself :) )
14:22Pascal_1 has quit IRC
14:23Pascal_Debian has quit IRC
14:23
<ogra>
warren, i want a fix in the release that works and doesnt break security or much code
14:23
<warren>
vagrantc: what are examples of convolunted steps?
14:23
<ogra>
the proper fix we develop upstream can go into ubuntu 8.04.1
14:23
<vagrantc>
warren: well, currently, you have to use some magic to figure out the DISPLAY from the socket name.
14:23
<ogra>
for now all i want is an easy way
14:24
<warren>
vagrantc: oh, crap.
14:24
I see
14:24
<vagrantc>
warren: basically, you don't have the LDM_DIRECTX environment variable available in all cases.
14:24
warren: so you can't just change behavior based on that.
14:24
it's started from udev...
14:25
<warren>
LDM_DIRECTX present or not is not the only problem right?
14:25
<vagrantc>
although, the only case where it really matters is when ldm is actually connected ... so you do have an ssh socket available.
14:25
<ogra>
LDM_DIRECTX is in env on the client
14:25
<warren>
I know
14:25
<ogra>
its exported by getltscfg
14:25
<vagrantc>
but not in udev's environment.
14:25
<ogra>
i dont know whch cases vagrantc talks about
14:26
hmm, it should be though
14:26
<vagrantc>
ogra: log in, plug in USB stick, add_fstab_entry is called and doesn't have ltsp environment variables available.
14:26
<warren>
So LDM_DIRECTX availability is a separate problem from DISPLAY name
14:26
<vagrantc>
warren: you need to tell it which display to connect to.
14:26
<ogra>
yeah
14:27
<vagrantc>
ogra: since cdpinger is started from ltsp-client's init script, it actually should have LDM_DIRECTX available ...
14:27
ogra: but USB sticks plugged in after login won't.
14:27
and floppy devices won't.
14:27
<ogra>
i guess udev runs to early :(
14:29
<vagrantc>
so, it's trivial to get some devices working ok
14:30
<ogra>
pfft
14:31
IS_DIRECTX=$(ps ax|grep -q "ssh -X -t -S")
14:31
its so trivial :)
14:35
<warren>
err
14:35
please don't rely on screen scraping
14:35
<ogra>
well or grep DISPLAY
14:36
<vagrantc>
warren: these are just ogra's quick and dirty hacks for hardy
14:36joel_ has joined #ltsp
14:36
<vagrantc>
ogra: though you could probably just use: pgrep -f -l "ssh -X -t -S"
14:37
<ogra>
well, iÄd rather match for DISPLAY
14:37
<vagrantc>
because then you get DISPLAY
14:37
<warren>
where is ssh -X from?
14:37
<ogra>
its in the ssh command if LDM_DIRECTX is there
14:37
ldm's second call
14:37
<warren>
there's no instance of the string "-X" in ldm-trunk
14:37
<vagrantc>
warren: ltspfs
14:37
<ogra>
it was when i checked out the code :)
14:38
vagrantc, no, iÄm grepping for the ldm comand, warren is right
14:38
but my ldm upstream checkout is from mid feb.
14:38
<vagrantc>
ah, right.
14:38
<warren>
grepping for that is quite a bit fragile
14:38
<ogra>
you changed X to Y sometime in march
14:38
warren, DISPLAY should be safe
14:38
<joel_>
Interrupting for a sec.... How do you add user accounts to the LTSP images?
14:39
<vagrantc>
with a single ldm instance
14:39
<ogra>
right
14:39
i'm counting on the fact that we have a better solution for 8.04.1
14:39
<vagrantc>
for your typical use case, it's fine ... but as warren says, fragile.
14:39
<ogra>
but i dont want to end up with floppy only or so for directx users
14:40
its used way to widely
14:40
<Q-FUNK>
ogra: I have a sense that many people are gonna try getting their pet fix into 8.04.1, which is turnig this whole 8.04 into a pathertic dressed rehearsal.
14:41
<warren>
Do you need to ssh into the server in order to get DISPLAY?
14:41
<ogra>
no
14:41
thas all on a client
14:41
<warren>
how do you get it then?
14:41
<ogra>
vagrantc, pgrep -f -e DISPLAY |grep ${LDM_SOCKET}
14:41
HA !
14:42
<vagrantc>
nice!
14:42
<ogra>
warren, from the processlist
14:42
<vagrantc>
that's considerably less fragile.
14:42
<ogra>
yeah
14:42
will give me only the display associated with the socket
14:43
<warren>
I guess that can be improved in the future by using a pid
14:43
<ogra>
indeed
14:43
<warren>
we *DO* keep the pid of ldm somewhere right?
14:43
<ogra>
nope
14:43
<vagrantc>
warren: what do you think this is, unix?
14:43* vagrantc smiles
14:43
<ogra>
heh
14:43
<warren>
LSB initscripts keep them
14:44
<ogra>
only if you call the functions
14:44
<vagrantc>
we actually start it from the screen scripts on debian
14:44
<ogra>
its moot to cean up after yourself if all you can do is shut down/reboot
14:44
*clean
14:45
(on a tmpfs /)
14:45
<vagrantc>
resume-from-nbd-swap!
14:45
<ogra>
heh
14:45
i think i'll look into X a bit more in the future
14:45
for session migration ;)
14:46
i have some ideas ...
14:46
<vagrantc>
at least i have warren and others to help keep you in check now :)
14:46
<warren>
resume-from-encrypted-nbd-swap
14:46
<ogra>
lol
14:47
yeah someone has to watch my freaky ideas :)
14:47
<warren>
pgrep -f -e DISPLAY |grep ${LDM_SOCKET}
14:47
this doesn't make me puke so much
14:48
vagrantc: so this fix requires changes to both ldm and ltspfs?
14:48
<ogra>
i like it ... and it gives me both infos i need
14:48
<warren>
does it get the xauth key from local?
14:49
<ogra>
it doesnt need any xauth
14:49
ltspfsmounter in the session cares for auth
14:49
but it needs to know the display
14:49
<warren>
so this is only a change in ltpsfs?
14:50
<ogra>
the way we have it above atm, yes
14:51
(that needs more code, its just to get the info "are we in directx mode or not")
14:52
<vagrantc>
then you'll need to parse out the DISPLAY from that ... and set -X or not, depending.
14:52
<ogra>
IS_DIRECTX=$(pgrep -f -e DISPLAY |grep ${LDM_SOCKET})
14:53
then i can loop over IS_DIRECTX subsequently and match for DISPLAY to get the line i need
14:53
<vagrantc>
ogra: not -e ... -l
14:54
<ogra>
er, right :)
14:55
<vagrantc>
i don't know if this is a good idea or a bad one, but ldm could leave a file with a few variables somewhere for the ltspfs scripts and hooks to source ...
14:55
/var/run/ldm_socket_$FOO.conf or some such
14:55jammcq has quit IRC
14:55
<ogra>
depends how many we need
14:56bengoa has quit IRC
14:56
<ogra>
if its only display we can easily ut that in the socket name
14:56
if its more i agree
14:56Blinny has quit IRC
14:56
<vagrantc>
ogra: it keeps more future-flexible without much complexity.
14:58indradg is now known as idg|dinner
15:02cliebow has quit IRC
15:07TelnetManta has quit IRC
15:07
<ltsppbot>
"ogra" pasted "improved add_fstab_entry fix" (30 lines) at http://pastebot.ltsp.org/505
15:10Q-FUNK has quit IRC
15:10Gadi has left #ltsp
15:20
<vagrantc>
ogra: looks pretty good to me.
15:20DonSilver has joined #ltsp
15:20K_O-Gnom has quit IRC
15:20
<ogra>
yep
15:20joel_ has quit IRC
15:21
<vagrantc>
ogra: i'm not sure, but you might be able to incorporate the 'grep ${LDM_SOCKET}' into the pgrep
15:21
<ogra>
i'll test that later tonight, i need a break, if it works i'll try to get it in
15:22
<vagrantc>
maybe pgrep -f -l ${LDM_SOCKET}.*DISPLAY ....
15:22
ogra: oh, i'd use a case statement to get the DISPLAY_INFO ...
15:22Arauto has quit IRC
15:23
<vagrantc>
for line in $IS_DIRECTX ; do case $line in DISPLAY*) DISPLAY_INFO="${line}" ;; esac ; done
15:23
<ogra>
sweet
15:24
<vagrantc>
it's probably minutely faster than echo + grep
15:25
SSH_OPTS (and SERVER, for that matter) could probably go inside the if [ -S ${LDM_SOCKET} ] ...
15:25idg|dinner is now known as indradg
15:26
<vagrantc>
sure hope it works!
15:27
<ogra>
yeah, we could use that upstream, looks really elegant now
15:28
<vagrantc>
might have to do the same for the rc.d script ... and i'm not sure on remove_fstab_entry ...
15:29* vagrantc would like to rename *_fstab_entry to *_ltspfs_entry
15:31
<vagrantc>
i think this works, too: IS_DIRECTX=$(pgrep -f -l "${LDM_SOCKET}.*DISPLAY")
15:34
if i wasn't leaving town in a few hours, i'd test it now!
15:41
<ogra>
vagrantc, why do we have a dpatch in the ltspfs pacage ?
15:42leio_ is now known as leio
15:44
<vagrantc>
ogra: i never got any feedback on weather it was needed on other distros
15:44
ogra: it's upstream now
15:44
<ogra>
vagrantc, works :)
15:44
i got a cdrom on my vbox client desktop :)
15:45
<vagrantc>
ogra: did you just use your patchc, or with my suggested changes?
15:45
<ogra>
ll, actually an usb disk icon saying scsicd-cdrom
15:45
<vagrantc>
heh
15:45
<ogra>
not the last IS_DIRECTX one
15:45
but the former one
15:45
and yes, with the changes as we discussed them
15:46
i'll drop a debdiff on the bug if i'm done with nasty dpatch
15:46
<vagrantc>
"done with nasty dpatch" ?
15:47
<ogra>
well, i prefer to just make changes and keep the source packages around
15:47
so i can debdiff them
15:47
instead of adding an additional patch system
15:47
i mean ...
15:47
dpkg is half a pach system :)
15:47
<vagrantc>
modifying the source directly?
15:48
<ogra>
so that it goes into the diff.gz
15:48
<vagrantc>
right. everything i've ever heard recommends against that
15:48
<ogra>
needs more package revisions though
15:49
ubuntu recommends is where there is no patch system in a debian package yet
15:49
<vagrantc>
i'll test it and push it upstream, unless you'd rather do so
15:49
<ogra>
s/is/it
15:49
no, feel free
15:49
<vagrantc>
i probably won't be able to get to it till tomorrow night or the next day ... we'll see.
15:50
<ogra>
iÄll do the change in the package anyway
15:50
<vagrantc>
yea, that's what i figured.
15:50
also needs to modify rc.d similarly ...
15:51
<ogra>
only delayed mounter
15:51
<vagrantc>
right
15:51
<ogra>
thats what iÄm fiddling with atm
15:51
<vagrantc>
remove_fstab_entry works without changes?
15:52
<ogra>
no, i copied the same changes in
15:52
actually i just logged out, i didnt remove the cd, wait a se
15:52
c
15:53
yup
15:53
works
15:54
<warren>
could you please submit that change to the list now?
15:54
I'm not exactly sure what you are doing
15:55
I'd like to review it as well.
15:55
<ltsppbot>
"ogra" pasted "add/remove_fstab_entry final" (68 lines) at http://pastebot.ltsp.org/506
15:55
<warren>
ogra: this is the only necessary change from ltspfs-trunk?
15:55
<ogra>
plud the one iÄm working on for delayed mounter
15:55
*plus
15:56
warren, right, no further changes needed
15:56
<warren>
what is the purpose of delayed mounter?
15:57
<ogra>
mounting floppies and devices that were plugged before login
15:59
if there is no ldm socket the device and mountpoint info get written to a file, on login the delayed mounter rc.d script loops over the entries and mounts them
16:00
<warren>
ok
16:02bricode has joined #ltsp
16:11
<vagrantc>
ogra: trailing ; on your esac ...
16:11
ogra: and if you want to minimize your diff, leave SERVER where it is ...
16:11
even though in the long term it should be there
16:12
<ogra>
nah, thats fine, it lives iide the if in remove_ as well
16:12
*inside
16:14
<vagrantc>
ogra: in remove_fstab_entry, looks like you put the if [ -S ${LDM_SOCKET} ] ... after all the IS_DIRECTX stuff ... if there's no socket, no sense checking for DIRECTX
16:14
or am i missing something?
16:15
<ogra>
http://launchpadlibrarian.net/13543330/udev-scripts-security-regression.dpatch
16:15
oh, right
16:16
<vagrantc>
ogra: otherwise, looks great :)
16:17
<ogra>
i think so as well
16:17
i need to test the delayed mounter changes still
16:17
<vagrantc>
only thing i'd do different is make a common function for them all to use
16:17muh2000 has quit IRC
16:18
<vagrantc>
for upstream
16:18
<ogra>
yeah
16:21Q-FUNK has joined #ltsp
16:21
<warren>
I'm interested in this going into upstream
16:21
after more testing
16:23
<vagrantc>
ogra: you've made sure it doesn't break LDM_DIRECTX=false ?
16:24
<ogra>
vagrantc, as long as LDM_DIRECTX=false doesnt put DISPLAY into the ssh command
16:24
<vagrantc>
ogra: yeah, i mean it looks good. but i'm a sucker for testing :)
16:25
<ogra>
which would be news for me
16:25DonSilver has quit IRC
16:26
<ogra>
hmm, delayed mounter doesnt work
16:26
<vagrantc>
hrm.
16:26
ogra: DISPLAY isn't set up server-side till after delayed-mounter runs ?
16:27
<ogra>
hmm. that may be
16:27davidj has joined #ltsp
16:27
<vagrantc>
gah, this is difficult.
16:27
<davidj>
vagrantc: Hey, it wouldn't be fun if it weren't. ;-)
16:27
What are you working on at the moment?
16:30deavid has quit IRC
16:33
<vagrantc>
davidj: localdevices with LDM_DIRECTX=true after ldm security patches
16:34
<bricode>
Q-FUNK: What was the final verdict of that patch from Jordan?
16:39
<Q-FUNK>
bricode: works for me. what about you? tried my test hardy package?
16:43slipttees has joined #ltsp
16:43cpunches has quit IRC
16:43
<davidj>
vagrantc: I'd like to look at it. Where are you keeping the code these days?
16:43Q-FUNK has left #ltsp
16:44slidesinger has quit IRC
16:45
<vagrantc>
davidj: https://code.launchpad.net/~ltsp-upstream/
16:46
davidj: are you familiar with ltsp5 ?
16:46bricode has quit IRC
16:46
<vagrantc>
basically, it looks like ogra's got it ... but delayed-mounter is looking rather tricky.
16:47
ogra: any way for delayed-mounter to queue things up to be started later ?
16:47
kind of ugly and intrustive ...
16:48
<ogra>
thats what i'm playing ith atm
16:48Q-FUNK has joined #ltsp
16:48
<ogra>
basically adding a delayed option to ltspfsmounter
16:48
that lets it sleep for 10 secs before mountin
16:48
the ssh command needs to background it for that, not sure thats possible
16:49
<vagrantc>
what i don't get is why the display isn't available ... because the xauth stuff runs before that.
16:52abadger1999 has quit IRC
16:53
<vagrantc>
ogra: AH!
16:53
<ogra>
ah ?
16:53
<vagrantc>
ogra: it's because the pgrep doesn't work yet
16:53
<ogra>
meh
16:54
indeed
16:54
<vagrantc>
almost back where we started...
16:54
<ogra>
not reallly
16:54
i have working usb devices at least
16:54
<vagrantc>
back to figuring out how to get the DISPLAY value, though
16:54
<ogra>
nah
16:55
back to fix up delayed mounting :)
16:55
<vagrantc>
you might have more variable names available in the rc.d scripts
16:58
<ogra>
i have LDM_DIRECTX
16:59
but i dont have my ip
16:59
nor DISPLAY
16:59
<Q-FUNK>
where did brian go?
16:59
<ogra>
erm
17:00
<vagrantc>
ogra: oh, you don't have LDMINFO_IPADDR
17:00
<ogra>
i have display i think
17:00
one sec
17:00
hmm
17:01
:6
17:01
why the heck dont we put the ip into display
17:02
ok, got all i need :)
17:14
hey
17:14
a floppy
17:14
!
17:14
a floppy AND a CD !
17:15
sadly it needs is_boolean_true() :(
17:15
sicne it uses LDM_DIRECTX
17:26
<warren>
is_boolean_true() isn't huge
17:30
<vagrantc>
boolean_is_true ... :)
17:30
<warren>
whatever
17:31
=)
17:31
<vagrantc>
well, when making patches near to a release, can be good to point out small details :)
17:33mccann has quit IRC
17:45
<ltsppbot>
"ogra" pasted "patch with working delayed-mounter" (94 lines) at http://pastebot.ltsp.org/507
17:49
<warren>
were there any changes to your previous paste?
17:50
<ogra>
delayed-mounter is completely different
17:50
<warren>
I know
17:50
<ogra>
that were the changes
18:03
<vagrantc>
not too bad ...
18:05
ogra: should still test for the socket in remove_*
18:06
"for foo in /var/run/ldm_socket_* " returns "/var/run/ldm_socket_*" if there's nothing there.
18:07
and i'd still use a case statement like boolean_is_true, to get an exact match ...
18:08
for the test on LDM_DIRECTX ...
18:13
<ltsppbot>
"vagrantc" pasted "delayed-mounter with simpler LDM_DIRECTX test" (20 lines) at http://pastebot.ltsp.org/508
18:14
<vagrantc>
ogra: like?
18:14
<ogra>
heh
18:14petre has joined #ltsp
18:15
<ogra>
yes, looks weird though
18:15
<vagrantc>
it looks weird, but it's more like the rest of our boolean tests
18:16
<ogra>
changed the dpatch
18:16
<vagrantc>
ogra: that works with or without modifications to ltspfsmounter ?
18:17
<ogra>
without
18:17
<vagrantc>
cool.
18:17
<ogra>
the pgrep was the issue, not xauth
18:17
<vagrantc>
right
18:19abadger1999 has joined #ltsp
18:20
<vagrantc>
ogra: very cool to have this figured out.
18:20
it was getting on my nerves.
18:22
<ogra>
yeah
18:23
on mine as well
18:23bricode has joined #ltsp
18:23
<ogra>
now to figure out where i et the time back i inveseted :P
18:23
<vagrantc>
eeyk
18:25* ogra gets his amd64 for install tests and goes on to shuffle CDs
18:29elisboa has quit IRC
18:31elisboa has joined #ltsp
18:34viking-ice has quit IRC
18:39
<Q-FUNK>
bricode: did you get a chance to try the test package i made for hardy with the libddc patch?
18:40
<bricode>
Q-FUNK: Not as of yet. I've got it slated for tomorrow morning.
18:40
<Q-FUNK>
bricode: ok
18:40
<ogra>
vagrantc, adding [ "${BASEDEV}" = "cdrom" ] && LABEL="cdrom" under the floppy line in add_fstab_entry fixes the cdrom issue
18:41Q-FUNK has quit IRC
18:42
<davidj>
vagrantc: Sorry for long time no answer, had to get the kids fed. Yes, I am familiar with ltsp5
18:42
nuts
18:42* davidj is away: I'm busy
18:43johnny has joined #ltsp
18:55bricode has quit IRC
18:58staffencasa has quit IRC
19:07elisboa has quit IRC
19:20
<johnny>
ogra, i'm testing hardy right now.. and it has the same problem as my 2.6.24 running genkernel box
19:20
any batched operaton like installing packages.. slows down the entire thing
19:20
it's really noticable
19:31
<warren>
oh good, it isn't just us
19:31
=)
19:31
<johnny>
try 2.6.25 yet?
19:31
i was trying it.. but i can't make it work
19:31
not sure what i did to make it not boot
19:32
folks in gentoo are using ported versions of the old ck patches
19:32
dev friends i know that is
19:32
it's like HORRIBLE i don't know why it is so bad :(
19:32
<warren>
yeah, 2.6.25 hit F9 today
19:33
<johnny>
how did that even make it out????
19:33
<warren>
we were using git snapshots of 2.6.25 before this
19:33
<laga>
https://bugs.launchpad.net/bugs/131094 <- maybe it's this bug? johnny
19:33
<johnny>
how were the snaps?
19:33
recent ones..
19:34
<dberkholz>
i've been using .25 rc's forever
19:35
they've all been fine
19:36
<ogra>
johnny, i dont see such stuff here (using 2.6.24)
19:37
neither on the core duo nor on the classmatePC .... is yours x86 ?
19:47
vagrantc, <slangasek> ogra: 218231 looks ok to me for post-RC
19:48
and uploaded ... sits in the queue
19:48
<vagrantc>
ogra: cool!
19:50elisboa has joined #ltsp
19:50
<ogra>
wow, what a successfull day .... suspend/resume on the classmate works, all the ltspfs uglyness sorted :)
19:51johnny has quit IRC
19:51
<ogra>
theer is so much oddity in the ldm ui i so would love to fix as well
19:52
(like:being able to type something in the input field while its verifying the pw ... etc)
19:53ikonia has quit IRC
19:55
<vagrantc>
well, off on another crazy vagrant adventure...
19:56
probably drop in later tomorrow
19:56vagrantc has quit IRC
20:01praveer_cool has joined #ltsp
20:02praveer_cool has left #ltsp
20:04ikonia_ has joined #ltsp
20:05ikonia_ is now known as ikonia
20:25ltspbot has joined #ltsp
20:45elisboa has quit IRC
21:27joebaker has joined #ltsp
21:49petre has quit IRC
21:53joebake1 has joined #ltsp
21:54joebaker has quit IRC
21:57joebake1 has left #ltsp
21:58* davidj is away: sleeping
22:01gonzaloaf has quit IRC
22:13mccann_ has joined #ltsp
22:14captain_magnus has quit IRC
22:17captain_magnus has joined #ltsp
22:26mccann has quit IRC
22:37johnny has joined #ltsp
22:45vlt has quit IRC
22:45Lumiere has quit IRC
22:45ikonia has quit IRC
22:45asac has quit IRC
22:45mistik1 has quit IRC
22:45loather-work has quit IRC
22:45gonzaloaf_work has quit IRC
22:45tdiehl has quit IRC
22:45ikonia_ has joined #ltsp
22:45asac_ has joined #ltsp
22:45Lumiere has joined #ltsp
22:45mistik1 has joined #ltsp
22:45asac_ is now known as asac
22:45gonzaloaf_work has joined #ltsp
22:47
<johnny>
hi..
22:47captain_magnus has quit IRC
22:48tdiehl has joined #ltsp
22:49loather-work has joined #ltsp
22:50vlt has joined #ltsp
22:51captain_magnus has joined #ltsp
23:10cpunches has joined #ltsp
23:14captain_magnus has quit IRC
23:15johnny has quit IRC
23:17captain_magnus has joined #ltsp
23:17johnny has joined #ltsp
23:33
<johnny>
uggh.. my bridge almost works again
23:47
<cpunches>
johnny: why wouldn't it?
23:49
<johnny>
when i upgraded my init system to the next gen gentoo rc scripts, i screwed up when merging the config file :(
23:50
and then i went on a wild goose chase trying to upgrade to 2.6.25 :(
23:51
but now.. my tftp file isn't downloading , and i'm not sure why