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


Channel log from 7 January 2008   (all times are UTC)

00:22open_lad has joined #ltsp
00:41Egyptian[Home1 has joined #ltsp
00:42Egyptian[Home1 has left #ltsp
01:11basanta has joined #ltsp
01:13sep has joined #ltsp
01:13sep has quit IRC
01:14sepski has joined #ltsp
01:17sepski has quit IRC
01:18sepski has joined #ltsp
01:21sepski has quit IRC
01:21sep has joined #ltsp
01:22tux_440volt has quit IRC
01:23vagrantc has quit IRC
01:31ogra has quit IRC
01:32ogra has joined #ltsp
01:37cyberorg has quit IRC
01:41mikkel has joined #ltsp
01:59basanta has quit IRC
02:10ari_stress has quit IRC
02:11ari_stress has joined #ltsp
02:17roseman has joined #ltsp
02:33cyberorg has joined #ltsp
02:37ari_stress has left #ltsp
02:47roseman has quit IRC
03:04tux_440volt has joined #ltsp
03:21makghosh has joined #ltsp
03:41aboo0ood has quit IRC
04:13makghosh|afk has joined #ltsp
04:14makghosh has quit IRC
04:14makghosh|afk is now known as makghosh
04:36tux_440volt has quit IRC
04:46putty_thing has quit IRC
04:53elisboa has joined #ltsp
04:53putty_thing has joined #ltsp
05:25otavio has joined #ltsp
05:51open_lad has quit IRC
06:08Joe90 has joined #ltsp
06:08
<Joe90>
hi there, quick question - default X manager on ltsp5 (gutsy) - GDM?
06:09
I want to put an auto login line in ltsconf. i.e GDM user= etc
06:09vagrantc has joined #ltsp
06:10
<vagrantc>
otavio: so, there were 3 lintian errors from your build that didn't happen on mine.
06:10
otavio: ltsp build, that is. the two i don't understand are the .po errors
06:14
!localdev
06:14
<ltspbot>
vagrantc: "localdev" is (#1) Information regarding using pluggable devices and cdroms with LTSP can be found at http://wiki.ltsp.org/twiki/bin/view/Ltsp/LTSP-42-LocalDev, or (#2) for ubuntu: https://wiki.ubuntu.com/EnableLTSP5LocalDevices, or (#3) for debian-edu: http://wiki.debian.org/DebianEdu/HowTo/LocalDeviceLtspfs, or (#4) http://wiki.ubuntu.com/DebugLocalDev
06:15
<ogra>
vagrantc, !
06:15
happy 2008
06:15
<vagrantc>
ogra: same to you
06:16
ogra: i can't sleep... brain just started thinking about ltspfs :)
06:16
<ogra>
heh
06:16
just grab my hal stuff and finish it :P
06:16
<vagrantc>
might be out of my league
06:16
where is your hal stuff?
06:17
<ogra>
we need a *lot* of changes anyway, nothing in newer gnome or kde releases will work without policykit and consolekit
06:17
which makes everythig harder for us :(
06:17
<vagrantc>
indeed
06:17
especially those who want it to be feasible for a backport... and other distros which may or may not implement all this new stuff
06:18
that's why the simple mounting approach is attractive to me
06:18
though i finally hit a wall with the direct mounting to /media/USERNAME ...
06:18
<ogra>
all distros *have* to use it
06:18
as long as they use hal
06:19
which means all gnome/kde desktops at least
06:19
<vagrantc>
only way to make /media/USERNAME writeable and not actually be on the same filesystem is to mount something there... which causes an icon to show up in nautilus
06:19
<ogra>
nothing in gnome or kde will work without approval through one of the kits
06:20
ldm will be the hardest part though
06:20
<vagrantc>
yes, but older versions of gnome or kde don't use it, and i would hate to make it impossible to support those
06:21
<ogra>
hrmpf ...
06:21
why does sed -i not replace tabs properly :/
06:21* ogra just notices that XSERVER doesnt work anymore
06:21
<vagrantc>
what's your sed?
06:21
<ogra>
the default one we use in the image
06:22
<vagrantc>
the x configuration
06:22
er, i meant "sed line"
06:22
hrm. maybe i'm not really awake
06:22
<ogra>
if [ -n "$XSERVER" ] && [ "$XSERVER" != "auto" ]; then
06:22
SERVERLINE="Driver\t\"$XSERVER\""
06:22
sed -i /'Section "Device"'/,/'EndSection'/s/'Driver.*'/$SERVERLINE/g $INPUT_FILE
06:22
<otavio>
vagrantc: strange
06:22
vagrantc: have you used up to date sid for building?
06:22
<ogra>
the \t isnt parsed as tab
06:23
<vagrantc>
otavio: it was updated within a day of when you uploaded
06:23
<otavio>
vagrantc: it's strange. Wants me to do another build/upload?
06:23
vagrantc: I've build it in amd64 arch
06:24
<vagrantc>
otavio: well, i've got some other changes first.
06:24
otavio: ah, my builds were i386 ...
06:24
otavio: i think the script not executable error was because the executable bit isn't represented in the .diff.gz
06:25
otavio: i don't think any of them are urgent enough to upload before it hits testing
06:26
otavio: which should only be a couple days
06:26
<otavio>
vagrantc: ok
06:27
<vagrantc>
ogra: any objection to moving the ldm rc.d scripts related to ltspfs into ltspfsd ?
06:27
<ogra>
nope
06:27
and i think we had that dscussion already :)
06:28
<vagrantc>
yeah...
06:28
just wanted to check before i started making commits
06:28
<ogra>
i hope scott returns soon
06:28
<vagrantc>
been vacationing in a cabin or something, i hear
06:29
<ogra>
tere is so much ldm and ltspfs bugs we need him for (at least i dont see the time on my schedule to work into all that code atm)
06:29
yeah, he bought that cottage with a lot of property around it
06:30
<vagrantc>
yeah, ldm needs some work and is largely over my head
06:31
<ogra>
not over mine, but i need a certain amount of time ot understand the new stuff
06:31
and classmate keeps me busy this release
06:32
i just started ltsp tests with the -generic image ...
06:32
thats cross arch, so we wont need separate amd64 setups anymore :)
06:32
i was always suspecting the missing 486 instruction set would break the booting ... or make it slow or so ...
06:33
but apparently it even runs on my old 300MHz cyrix chip here
06:34
<vagrantc>
hm.
06:35
ogra: cross-arch, as in a single environment will have some i386 and some amd64 binaries?
06:35
and it'll use whichever are appropriate?
06:36
<ogra>
no as in amd64 capable kernel with x86 binaries by default
06:36
the ubuntu -generic kernel runs fine on both arches
06:37
binaries are all x86 but run onder a kernel that actually boots on amd64
06:37
<vagrantc>
ah
06:37
the -386 didn't run on amd64?
06:37
<ogra>
argh
06:37* ogra did just noice that he edits in the wrong chroot since 30min
06:38
<ogra>
*sigh*
06:38
right, the -386 kernel is built with the most minimal instruction set -M486
06:38
and doesnt run on amd64 that way
06:38
<vagrantc>
hrm.
06:40
that's probably the same on debian
06:41
<ogra>
we used to have an -amd64 kernel ...
06:41
that was merged into -generic with the -686 one
06:43
<vagrantc>
otavio: if you haven't already seen it: http://lintian.debian.org/reports/maintainer/pkg-ltsp-devel@lists.alioth.debian.org.html#ltsp
06:44
otavio: i can get the script-not-executable when i build it, but i can't get the invalid-po-file warnings
06:47
<ogra>
gah
06:47
<wizzy__>
The new ltsp client root is a combination of nbd/unionfs/squashfs. A squashfs is built, compressed, and mounted via nbd. Then there is supposedly a unionfs overlay (that I don't think works). How do I tell if unionfs is overlaying a file (besides looking at its content) and how do I debug this ?
06:47
<ogra>
looks like all the calls to sed -i need to become sed -ie now
06:47
<vagrantc>
ogra: -i -e
06:47
<ogra>
yeah
06:47
still
06:47
thats silly
06:47
<vagrantc>
ogra: -ie does something weird
06:48* vagrantc likes -e anyways
06:48
<ogra>
well
06:48
it will berak tons of scripts
06:48
*break
06:48
wizzy__, SCREEN_02=shell in lts.conf and then see if anything is writable
06:49
wizzy__, what makes you think unionfs doesnt work
06:49
<vagrantc>
wizzy__: what distro ?
06:49
<wizzy__>
ogra: edubuntu gutsy - because the message at the top of lts.conf says I can put it elsewhere
06:50
I tried that, but it doesn't change from SCREEN_02=shell
06:51
<ogra>
if you set SCREEN_02=shell it wont start anything else
06:51
<wizzy__>
Do I need to remove it from the squashfs for it to appear ?
06:51
<ogra>
you will have a shell on tty2
06:51
<wizzy__>
yes, but I log in there, and look at the lts.conf, and it is the squashfs version, not the tftp/* version
06:52
<ogra>
where exactly lies your tft version ?
06:52
*tftp
06:52
<wizzy__>
It says put it at /var/lib/tftpboot/ltsp/i386/lts.conf
06:52
<ogra>
and thats where you got it ?
06:53
<wizzy__>
I put something there, but it does not overlay squashfs
06:53Joe90 has quit IRC
06:53
<ogra>
and /var/lib/tftpboot/ltsp/i386/lts.conf exists ?
06:54
with that name ?
06:54
<wizzy__>
ogra: yes, yes
06:55
and I rebooted
06:55
(the client)
06:55
<vagrantc>
ogra: when i had nbdroot working on debian, the lts.conf from tftp never worked...
06:55
<ogra>
right
06:55
i currently use it on a feshly built hardy
06:55
works perfectly here ... on my gutsy as well
06:56
<wizzy__>
ogra: What can I do on the client to debug it ?
06:56
<ogra>
note that it needs to be in ltsp/i386/ ....
06:56
<wizzy__>
I have it at exactly the path above - I cut/paste
06:57
<ogra>
well, you could try to get the file manually with a tftp client
06:57
to see the server side works as excpected first
06:58
<wizzy__>
Does it /get/ it via tftp ? I thought it is unionfs magic
06:58* vagrantc upgrades vagrantc's virtualbox debian ltsp install...
06:58
<ogra>
it gets it via tftp and then copies it over the one in the image
06:59
this all happens in initramfs
06:59
<wizzy__>
Ok - I will give that a bash
06:59
<ogra>
vagrantc, -e doesnt help :(
07:00
<vagrantc>
ogra: you sure the file still contains a single tab ?
07:00
<ogra>
i replace whatever is there with a tab
07:00
<vagrantc>
ogra: might want to use something that catches any amount of whitespace
07:00
<ogra>
i dont check if there is one already
07:01
its the writing, not the reading thats broken
07:01
<vagrantc>
still works on debian sid ...
07:02
<ogra>
hmm
07:02
there were no changes in ubuntus sed
07:03
and the above code makes 'Driver "cyrix"' to be 'Drivert"vesa"'
07:03
<vagrantc>
bash/dash differences?
07:04
<wizzy__>
"tftp 172.20.1.1 -c get ltsp/i386/lts.conf" works
07:04
<ogra>
i couldnt imagine where that should have any effect on sed
07:04
<vagrantc>
they tend to handle backslash characters differently ...
07:04
<ogra>
i know
07:04
<vagrantc>
well, if the shell is interpreting the tab somehow
07:05
<ogra>
but does sed take that behavior from the shell ? couldnt imagine
07:05
<vagrantc>
depends on how it's passed...
07:06
ogra: have you modified configure-x.sh ?
07:06
<ogra>
no
07:06
its the identical one we use in gutsy
07:06
its the first time i try XSERVER=vesa in hardy though
07:06
so it might be there longer
07:12chupa has quit IRC
07:12chupa has joined #ltsp
07:13
<ogra>
grr
07:14* ogra wonders why it works fine with other variables but not XSERVER
07:14
<ogra>
X_MODE has no probs with tabs for example
07:15
<wizzy__>
Jan 7 15:14:19 zululand in.tftpd[23920]: RRQ from 172.20.1.158 filename /lts.conf // sending NAK (1, File not found) to 172.20.1.158
07:15
<ogra>
filename /lts.conf is surely wrong
07:15
<wizzy__>
You say it is in the initramfs
07:15
<ogra>
should be /ltsp/$arch/lts.conf
07:16
the path is assembled from $(basename filename)
07:16
where filename is the name of the kernel image
07:16
<wizzy__>
Jan 7 15:13:52 zululand in.tftpd[23913]: RRQ from 172.20.1.158 filename /ltsp/i386/nbi.img
07:18
<ogra>
ogra@ceron:~$ filename=/ltsp/i386/nbi.img
07:18
ogra@ceron:~$ tftpath=$(echo ${filename%/*}|tr -d '"')
07:18
<wizzy__>
ltsp-server 5.0.39
07:18
<ogra>
ogra@ceron:~$ echo $tftpath
07:18
/ltsp/i386
07:19
are you using etherboot ?
07:19
<wizzy__>
Is that a greedy * in busybox or something ?
07:20cliebow_ has joined #ltsp
07:20
<ogra>
or PXE ?
07:20
<wizzy__>
It is RTL8139 with rom-o-matic from a long time ago
07:20
<ogra>
hmm
07:20
that might be it
07:20
can you try a rom-o-matic image with PXE emulation ? (the recet ones all can do PXE)
07:21
seems the filename gets wiped in etherboot images or something like that
07:21
<wizzy__>
ogra: I do not have all my rom burner stuff here ATM - stuck in the bush. No matter, I will move the file
07:21
<ogra>
you could also just move lts.conf up two levels ;)
07:21ariane has joined #ltsp
07:21
<wizzy__>
Ja, that is what I meant
07:22
<ogra>
/opt/ltsp/i386/usr/share/initramfs-tools/scripts/ltsp_nbd
07:22
<wizzy__>
Thanks - I didn't realise unionfs used tftp - now I can fix it.
07:22
<ogra>
thats the responsible initramfs snippet btw
07:23
<wizzy__>
I don't have the dev stuff here - that file doesn't exist. Should I update a wiki somewhere ?
07:23
<ogra>
it mounts the image readonly via nbd, then mounts a tmpfs in the clients ram and then merges both in a unionfs ...
07:23
*then* it copies over lts.conf
07:23
<wizzy__>
ok
07:25
<ogra>
well, there is surely a bug,filing that one would help me to not forget about it :)
07:25
no need for wikipages atm
07:25
<wizzy__>
where do I file ?
07:25
<ogra>
launchpad.net
07:25
<wizzy__>
k
07:25
<ogra>
https://bugs.launchpad.net/ubuntu/+source/ltsp/+filebug
07:26
actually :)
07:32
<vagrantc>
ogra: so, i'm guessing what you were saying earlier about ltspfs requiring hal recently hit debian sid ... no icons anymore
07:32
<ogra>
no, currently we dont use hal at all
07:32
and gnomevfs works arund the icon stuff atm
07:33
but soon the access rights wil only be handled by polkit
07:33
currently the access rights for the session are already handled by consolekit ....
07:34
which means no administarative tasks at all anyore, everything that goes beyond /home is blocked by default
07:34
in the near future polkit and conkit will work hand in hand like an ACL mechanism
07:34
and dbus wont let anything through thats not autorized by them
07:35
since 80% of the gnome apps and good amount of KDE apps as well only communicate through dbus nowadays that makes the desktop barely usable
07:35
s/barely/hardly
07:35
<vagrantc>
hm. well, at the moment, ltspfs don't create icons in nautilus from sid
07:35
<ogra>
oh
07:35
it does in hardy
07:36
i tested that some mins ago
07:36
<vagrantc>
yeah, guess i'll just switch :P
07:36
it was working a couple days ago in sid ... no idea what changed
07:36
<ogra>
and it did in gutsy ...
07:36
gnomevfs likely
07:36
the mounts happen i bet
07:37
there is also a gconf key in nautilus that rules if icons are shown at all
07:38
you might want to check that with gconf-editor
07:40
<vagrantc>
no new libgnomevfs* uploads since november
07:40
<ogra>
nautilus ?
07:41
<vagrantc>
ditto
07:41
yeah, the mounts happen...
07:41
user can read them and stuff
07:42
and this is without all my crazy patches
07:42
<ogra>
are you sure its not your playing ?
07:42
hmm
07:42
<vagrantc>
straight from debian sid
07:42
<ogra>
did debian enable consolekit already ?
07:42
<vagrantc>
how would i check?
07:42
<ogra>
(even though it shouldnt have any effect yet on ltspfs)
07:43
well, is console-kit-daemon running
07:43subir has quit IRC
07:43
<vagrantc>
no
07:44
<ogra>
dpkg -l consolekit
07:44
<vagrantc>
although, this isn't the exact same setup ... this is an upgrade from etch
07:44
<ogra>
we have 0.2.3-3ubuntu1
07:45
<vagrantc>
it's not installed ...
07:45
<ogra>
2.22 will require it
07:45
(gnome 2.22
07:45
)
07:45
<vagrantc>
0.2.3-3 available
07:46
<ogra>
what do you have in sid atm ?
07:46
(gnome)
07:46
<vagrantc>
on 2.20 for most gnome related stuff
07:46
<ogra>
thats what we have in gutsy
07:46
<vagrantc>
this is perplexing
07:46
<ogra>
yeah
07:46
i dont see a reason why it shuldnt work
07:47
<vagrantc>
i should head down to freegeek, test it on the setup that was working a couple days ago, upgrade, and see if it's still working
07:47
<ogra>
yeah
07:47
or try a fresh virtbox setup
07:47
<vagrantc>
i need to scrounge more drive space for that :)
07:48* ogra built a raid0 machine with 4G and 300M/s disk tranfser rates over christmas ....
07:48
<vagrantc>
big difference is one is doing full-blown (or mostly full) gnome, while this is icewm + nautilus
07:48
<ogra>
ltsp-build-client (with an apt-proxy in the lan) takes less than 5 min :)
07:48
<vagrantc>
which was working with etch as the server and a sid ltsp
07:51* ogra goes for a coffeebreak
08:17ariane has quit IRC
08:19sean_csve has joined #ltsp
08:19Egyptian[Home] has quit IRC
08:19
<sean_csve>
hi , how do I share afolder between all users so that they all have read write access to all the files in it
08:21Egyptian[Home] has joined #ltsp
08:23mikkel has quit IRC
08:25
<PerfDave>
ogra: Hmm, what's better? apt-proxy, or just using a cacheing HTTP proxy?
08:26
<vagrantc>
PerfDave: the answer is no
08:28
well, maybe progress has been made in the last few years, but i've been disappointed by all of the dynamic proxying alternatives ...
08:29
<otavio>
hi
08:29
<vagrantc>
best luck i've had is using squid, and then parsing the logs to determine which packages to mirror
08:35Gadi has joined #ltsp
08:37sean_csve has quit IRC
08:47makghosh has quit IRC
08:55K_O-Gnom has joined #ltsp
08:56slidesinger has joined #ltsp
08:57K_O-Gnom has quit IRC
08:57K_O-Gnom has joined #ltsp
08:58Guaraldo has joined #ltsp
08:59makghosh has joined #ltsp
09:23spectra has joined #ltsp
09:24cliebow__ has joined #ltsp
09:28sid has joined #LTSP
09:39Guaraldo has quit IRC
09:39Guaraldo has joined #ltsp
09:48cliebow__ has quit IRC
10:16sep has quit IRC
10:16sep has joined #ltsp
10:18BadMagic has joined #ltsp
10:18slidesinger has quit IRC
10:18makghosh has quit IRC
10:22Guaraldo has quit IRC
10:23Guaraldo has joined #ltsp
10:24
<vagrantc>
hrm.
10:24* ogra agrees
10:25
<vagrantc>
so there appears to be something going on running gnome that doesn't happen when running nautilus
10:25
ltspfs is working fine with gnome, just not with icewm + nautilus
10:25
<ogra>
so there are services missin i guess
10:26
did you try to start gnome-settings-daemon with nautilus ?
10:26
or at least gconfd
10:27* ogra pokes around in consolekit ...
10:27
<ogra>
there seems to be a pam module, i wonder if we could just use that
10:30
<vagrantc>
ogra: i didn't try starting gnome-settings-daemon ... pretty sure gconfd was running
10:32
<ogra>
ogra@ceron:~/devel/packages/consolekit-0.2.3$ ps ax|grep vfs
10:32
28635 ? S 0:00 /usr/lib/gnome-vfs-2.0/gnome-vfs-daemon
10:32
30105 ? S 0:00 /usr/lib/gnome-vfs-2.0/gnome-vfs-daemon
10:32
seems thats started as well for me here
10:32
no idea by what but it probably helps to have it running
10:33
<vagrantc>
ok
10:38
hmm... doesn't seem to run manually
10:38staffencasa has joined #ltsp
10:39
<vagrantc>
fails to connect to dbus ... dbus is running
10:39
<ogra>
likely somehow started by gnome-session
10:39
i guess you dont have the right dbus privs if you start it from a terminal
10:39
<vagrantc>
yeah...
10:40
now i've got a sid virtualbox ltsp server and a virtualbox sid ltsp client that has gnome-desktop
10:40
i'm pretty happy with that :)
10:40
<ogra>
on your laptop ?
10:40
<vagrantc>
yeah
10:41
<ogra>
and thats usable
10:41
i mean ...
10:41
does the mouse move more than every 20min ?
10:41
<vagrantc>
upgraded to a 866MHz with 512MB of ram
10:41
a while back ...
10:41
<ogra>
oh
10:41
congrats :)
10:41
<vagrantc>
heh
10:41
6-8 hours of battery life :)
10:42
<ogra>
wow, more than i get out of the classmate
10:42
<vagrantc>
virtualbox works great on this
10:44aboo0ood has joined #ltsp
10:45
<aboo0ood>
hi everybody, i have a question about LTSP-loadbalancer .. is there anything about balancing per application ?
10:46
<ogra>
aboo0ood, you will have to wait for francis (fgiraldeau), the loadbalancer stuff isnt part of the default ltsp
10:46
<vagrantc>
aboo0ood: not yet, really.
10:47tux_440volt has joined #ltsp
10:47
<aboo0ood>
ogra, vagrantc : What kind of balancing does LTSP-5 offer ?
10:48
<ogra>
per session ... you can provide a list of different servers the user can select from
10:48
<tux_440volt>
hi ! while trying to boot my clients i get /etc/rc.early_sysinit not found
10:48DawnLight has joined #ltsp
10:49
<aboo0ood>
ogra, vagrantc : Is Francis working on a per application solution ? I read about LTSP-loadbalancer and if i didn't miss anything i didn't see anything about balancing per application
10:49
<vagrantc>
you can also control which server gets logged in by default with a script...
10:49
aboo0ood: that's an order of magnitude more complicated
10:50
<DawnLight>
what can i use for network sound in the debian etch packages?
10:50
<vagrantc>
if you really want to reduce the load on the server, use diskless workstations
10:50
DawnLight: i *think* esound should work, but it's been so long since i've tried.
10:51
DawnLight: the etch ltsp backports should work with pulseaudio+alsa
10:51
<aboo0ood>
vagrantc, ogra : i am thinking about developing such a solution and trying to figure out the possibilities
10:51
<DawnLight>
backports? hmm...
10:52
<aboo0ood>
vagrantc, ogra : why did u say it's orders of magnitude more complicated ?
10:52
<ogra>
aboo0ood, the prob is that you need a network auth mechanism in place (ldap, nis etc) which is simply beyond ltsp
10:52
<aboo0ood>
an order *
10:53
<ogra>
its rather a "network design and setup" than an application :)
10:53
<aboo0ood>
ogra, can't this be done over ssh ?
10:53prpplague has joined #ltsp
10:53
<aboo0ood>
the idea of "ssh -X" .. can't this be used ?
10:53prpplague has left #ltsp
10:53
<ogra>
sure, but you still need to make sure that auth is provided and in sync on all servers ... DISPLAY variables are set etc
10:54
its trivial to set that up on a single network as a specific solution in a company or so ...
10:55
but its very hard to build something like that in a manner to have it set up everything automatically for you
10:55
in a functional basic state at least and integrated with different desktop environments etc
10:56
<vagrantc>
whereas, it's fairly simple to allow multiple login servers
10:56
aboo0ood: why do you need per-application load balancing?
10:57
<aboo0ood>
vagrantc, i thought it would be more beneficial in massive LTSP solutions
10:57
<vagrantc>
but no specific implementation?
10:57
<aboo0ood>
vagrantc, not yet
10:58
vagrantc, don't u think it would be better than per session balancing ?
10:59
vagrantc, more dynamic ..
10:59
<vagrantc>
more headache
10:59
<aboo0ood>
lol
11:00
yes, but for the developer maybe, not for the user
11:00
<vagrantc>
per session is simple to understand and troubleshoot
11:00
yes, as long as everything works perfectly
11:00
but when things break, it will break much more badly
11:01
i think the simplest form of load balancing would be to get local applications working
11:02
<ogra>
that still needs a new sshfs
11:02
<vagrantc>
that seems more feasible to me
11:02
<ogra>
which scottie hasnt done yet
11:02
<vagrantc>
indeed
11:02
<ogra>
and session side integration
11:02
<vagrantc>
details...
11:02
<ogra>
(menu integration etc)
11:02
indeed
11:02* ogra doesnt count on localapps for hardy
11:03
<vagrantc>
i'm hoping to finally fix the double-mount issue for ltspfs... and finally have a working setup where i can test it
11:03
<ogra>
feature freeze is in five weeks ...
11:03
right i'm concentrating on bugs as well
11:04
and most of my time is eaten by classmate anyway
11:04
<vagrantc>
i actually have a working fix i like ...but it still causes an extra mount to show up... which is better than 2 mounts for each device... but still
11:04DawnLight has left #ltsp
11:04
<ogra>
hmm
11:04
i think we have a fix in the ubuntu gnomevfs pacage
11:04
<vagrantc>
that's a workaround, not a fix :P
11:05
i've been thinking about this so much i feel near to fixing it
11:06
<ogra>
did you have a look at ignore_ltspfs_volumes.patch in the ubuntu gnomevfs package ?
11:06
<aboo0ood>
what is required to configure the network for ssh tunneling ? and what's hard about it ?
11:07
<ogra>
vagrantc, that patch is in ubuntu since edgy at least with this reason you could probably push it through in debian :)
11:14
<vagrantc>
ogra: i'd still prefer to fix it outside of gnome-vfs
11:15primeministerp has quit IRC
11:21
<vagrantc>
ogra: well, i came up with a one-line patch to fix the double-mount issue :)
11:21
change --bind to --move
11:23
well... maybe a little more patching ... but it's basically working
11:25
i seem to recall etch still behaving a little oddly...
11:25
with --move
11:26tux_440volt has quit IRC
11:36
<vagrantc>
ok, so 2 lines of patches has got it doing a single mount and unmounting the ltspfs fuse mount ... it leaves the directories behind, though...
11:51slidesinger has joined #ltsp
11:54Guaraldo has quit IRC
12:01mikkel has joined #ltsp
12:05Guaraldo has joined #ltsp
12:08Egyptian[Home1 has joined #ltsp
12:25Q-FUNK has joined #ltsp
12:34primeministerp has joined #ltsp
12:39
<Spliffster>
hi all. can any one tell me who the guy was which is knowledgeable with ibm netvista thin clients ?
12:40zodman has joined #ltsp
12:43
<sutula>
vagrantc: IIRC, there's been a couple of bugs that people are working around, at least in the etch backports version: a) In /opt/ltsp/i386/usr/share/ldm/rc.d/S10-delayed-mounter, change "ssh -S" to "ssh -X -S", b) delete env['DISPLAY'] = 'localhost:10.0' from /usr/sbin/ltspfsmounter, line 21
12:44
vagrantc: Just wondering if a fix is imminent or should this be added to http://wiki.debian.org/LTSP/Howto
12:46
<vagrantc>
sutula: yeah
12:46
sutula: a fix is in progress, but add it to the wiki anyways
12:46* sutula does that to his kids too, when they ask an "or" question :)
12:46
<vagrantc>
sutula: we can always remove it when it's no longer relevent
12:47
<sutula>
vagrantc: K, I'll add it
12:47
<vagrantc>
ldm in sid already has the fix
12:47
although i may muck things up a bit more
12:48
sutula: how's your C code?
12:49
think i'm on the verge of fixing http://bugs.debian.org/443200
12:49
otavio: i need some help with C code in ltspfs... lbmount.c, specifically
12:49
<sutula>
vagrantc: It's good...what do you need?
12:49
<vagrantc>
sutula: oh, excellent ...
12:50
sutula: well, i'm trying to fix lbmount.c to only do a single mount ... i've almost got it, just can't figure out how the is_mounted function is working...
12:51
sutula: i'll paste what i have to pastebot
12:51
<sutula>
vagrantc: K
12:53
<ltsppbot>
"vagrantc" pasted "lbmount.c patched to use mount --move instead of mount --bind" (302 lines) at http://pastebot.ltsp.org/408
12:55
<vagrantc>
sutula: i think there's something in is_mounted function ... probably in the while loop ... which is looking at the wrong directory ...
12:55
<sutula>
vagrantc: Looking
12:55chupa has quit IRC
12:55
<vagrantc>
sutula: for umount, lbmount doesn't need to touch the /tmp files
12:55
well, doesn't need to anymore
12:56chupa has joined #ltsp
12:56
<cliebow_>
Spliffster, i know more than i want to know about netvista..
12:57* vagrantc wonders what "realpath" does
12:57
<vagrantc>
a path sanitizer?
12:57
ah, it actually has a man page
12:59* vagrantc wonders what strcmp is doing ...
12:59* vagrantc wonders a lot
13:00Egyptian[Home1 has quit IRC
13:00
<sutula>
vagrantc: strcmp is comparing strings, returns 0 if they compare exactly
13:00
<vagrantc>
sutula: right ... but the code is ... if (!strcmp(targetpath1, targetpath2)) {
13:01
ah!
13:02
it's making sure you're not checking the same directory that it's bind-mounted to ...
13:02
but in this case, that's the only one we want to check
13:03* sutula doesn't seem to have enough of vagrantc's context, so continues to just answer vagrantc's questions
13:03
<vagrantc>
:)
13:03
<otavio>
vagrantc: say
13:03
<vagrantc>
patching all this C is really stretching my limits
13:04
i'm amazed i've gotten this far
13:04* Gadi hugs perlcc :)
13:05
<otavio>
vagrantc: this if runs only if both variable are _different_
13:05
vagrantc: strcmp returns 0 if they're iqual
13:05
_e_qual
13:06
<vagrantc>
sutula: ok... context ... currently lbmount does a few things... it makes /media/USERNAME/DEVICE ... it checks if /tmp/.USERNAME-ltspfs/DEVICE is fuse-mounted and owned by USERNAME, and then bind-mounts /tmp/.USERNAME-ltspfs/DEVICE to /media/USERNAME/DEVICE
13:06
sutula: that's the mount mode ...
13:07
sutula: then it basically does the opposite .... umounts /media/USERNAME/DEVICE, checks if /tmp/.UERNAME-ltspfs/DEVICE is owned by USERNAME and umounts it, and then removes /media/USERNAME/DEVICE and /media/USERNAME (if empty)
13:09
my patch is different in that instead of --bind mounting /tmp/.USERNAME-ltspfs/DEVICE it bind --move's it ... so at umount time, /tmp/.USERNAME-ltspfs/DEVICE doesn't exist anymore
13:09primeministerp has quit IRC
13:10primeministerp has joined #ltsp
13:10
<vagrantc>
otavio: yeah, i figured that out from the man page, i just couldn't figure out how it was used in the code at first
13:10
otavio: i.e. why it was used
13:10
but now i get it
13:11
and realpath sets a variable based on an argument, more or less... ?
13:12
<sutula>
vagrantc: Yes...get's rid of symbolic links, extra "./", etc, in the pathname
13:13
vagrantc: Think of it as sanitizing the path, but should not affect the logic
13:14
<vagrantc>
i just was having a hard time figuring out where the targetpath* variables were coming from
13:15
<otavio>
vagrantc: wants me to take a look on the source?
13:15
<sutula>
otavio: vagrantc posted something in a pastebot above
13:16
<vagrantc>
otavio: if you could
13:16ltsp has joined #ltsp
13:17
<ltsp>
hi all
13:18
anyone know how i'm increase nbd swap ? the default value is 32 MB
13:18
<vagrantc>
/etc/ltsp/nbdswapd.conf
13:19
SIZE=64
13:19
or whatever number you want
13:19
<ltsp>
this file dont exist
13:19
:(
13:19
<vagrantc>
create it
13:19
echo SIZE=64 >> /etc/ltsp/nbdswapd.conf
13:19
<ltsp>
ok ! i try it thank you vagrantc
13:22
vagrant i having issues wich users pendrivers, the option in nautilus for safe remove dont umount it
13:22
<vagrantc>
you can just unplug them after 2 seconds of inactivity
13:24
<ltsp>
this "bug" its normal on ltsp terminals ?
13:26
<vagrantc>
ltsp: i've heard others report it
13:27
ltsp: the way ltspfs works is that after 2 seconds of inactivity on the client, in un-mounts the filesystem ... the server still sees the ltspfs mount, but the media is not actually mounted ...
13:27
so unless you write to it and immediately yank it out, your media should be fine.
13:28
<ltsp>
great !
13:28
sorry but my english its sooo poor
13:28
:(
13:28
<vagrantc>
what nautilus does or doesn't do... i've used very little nautilus
13:28
not a problem
13:28
<ltsp>
i need other question for you
13:29
in my laboratory , the users contantly remove/delete my desktop icons on terminals
13:29
exist one way for i'm lock the desktop icons ?
13:33
<vagrantc>
no idea
13:33
i don't really use desktops much
13:34
maybe someone else
13:34
<sep>
ltsp, make the files owned by root ? :)
13:34
<vagrantc>
that would probably work
13:35
<ltsp>
chown root.root /home/$USER/Desktop/* ?
13:35
<vagrantc>
they'd probably still have permission to remove them
13:35
ltsp: try it out...
13:36
ltsp: well, try with one user first
13:36
<ltsp>
i try it thats works but the icons have no pictures
13:36
only X rede on icons
13:36
<sutula>
ltsp: Make a copy of what you want on each desktop in a separate directory, and have a login script to copy it to the desktop on login?
13:37
<vagrantc>
ltsp: you may need to give them read permission...
13:37
<ltsp>
exempla = openoffice icons ---> X (red)
13:37
its the good idea sutula
13:37* sutula doesn't know enough gnome to do that, but suspects that there's an IRC channel somewhere (else) where people would know
13:40
<ltsp>
for example i know lock the bckground proprierty , the command is:
13:40
sudo chmod 700 /usr/bin/gnome-background-properties
13:40
this command lock the background options
13:42
<johnny>
i always use sabayon for this..
13:42
isn't there a way to use gconf to mark a property as mandatory?
13:42
<ltsp>
sabayon dont work for me :(
13:43
<vagrantc>
sutula: ok ... i'm going crazy... i'm getting an error message ... it's printing using fprintf ... i'm adding a line immediately before it with the exact same syntax, yet i never see the output
13:43
<johnny>
it will be soon :)
13:43
i've been submitting some patches for it
13:43
or rather.. committing some patches
13:43
first some simple ones.. and then i'll be working on the heavier issues
13:44
<sutula>
vagrantc: Can you pastebot it again?
13:45
<ltsppbot>
"vagrantc" pasted "lbmount.c patched to use mount --move instead of mount --bind" (308 lines) at http://pastebot.ltsp.org/409
13:46
<vagrantc>
sutula: the error i'm getting is right below the UGH2 line ...
13:46
sutula: but UGH, HEY and UGH2 do not display anything
13:47
<ltsp>
guys i can use LDM2 (gutsy) on feisty (ldm python) ?
13:47
<sutula>
vagrantc: That almost sounds like you missed compiling it or are executing the old version. Is that possible?
13:48
<vagrantc>
sutula: m5sums match with the freshly compiled version
13:48
no warnings or errors at compile time
13:49
<sutula>
vagrantc: There's nothing wrong with the "C" syntax in that area
13:50
<johnny>
ltsp, i wouldn't try
13:51* vagrantc was putting the file in the wrong directory
13:51
<sutula>
vagrantc: That sounds more like it
13:51
<ltsp>
humm LDM2 is more faster than LDM
13:51
<johnny>
then use gutsy
13:51* vagrantc wonders how many times
13:52
<ltsp>
gutsy ltsp version dont boot wich nbi mages on my terminals
13:52
only pxelinux.0
13:52
<vagrantc>
ltsp: what version of etherboot are you using?
13:53
<ltsp>
5.4.3
13:53
<johnny>
mine boots fine with nbi
13:53
<ltsp>
on feisty nbi image works , but in gutsy dont work
13:54
<johnny>
you should generate a new chroot
13:55
<ltsp>
yes
13:55
but dont work wich nbi images
13:56
<johnny>
must be some small issue for that
13:56
<ltsp>
and i listen in ltsp br discuss list ( i'm from brazil) ltsp gutsys version is soo unstable
13:57
<johnny>
hmm.. my ltsp is stable.. i just had a problem with autologn
13:57
and firefox
13:57
but that isn't really related to ltsp tho
13:59
<ltsp>
humm i try again use ltsp wich ubuntu 7.10
14:02
ltsp in gutsy is more faster than feisty version ?
14:02
i have 20 terminals celeron 333+64/128MB, ltsp 5 boot is slow :(
14:03
<johnny>
333+64 ?
14:03bobby_C has joined #ltsp
14:03
<ltsp>
celeron 333 Mhz wich 64 MB SDR
14:03
or 128 MB SDR
14:03
<johnny>
64mb ram is always going to be a bit slow
14:03
if you run gnoe
14:03
gnome*
14:03
or firefox/openoffice
14:03sep has quit IRC
14:04
<ltsp>
yes ! and firefox and openoffice
14:04
the terminals freezes any times on powerpoint apresentations
14:04
<johnny>
they aren't powerful enough..
14:04
<ltsp>
:(
14:04
64 mb is no enough ?
14:04
<johnny>
nope
14:05
<ltsp>
"thinclients" :(
14:05
<johnny>
not for openoffice at all
14:05
hmm.. what did the docs say..
14:05* johnny looks
14:05
<ltsp>
i not understand sorry
14:06
my english is so terribly
14:06
¬¬
14:06
<johnny>
how powerful is the server?
14:06
what are the specs?
14:06
maybe you want to use xfce on the clients, and then upgrade the server
14:06
<ltsp>
my server is the Atlhon 64 3700+ wich 2 GB DDR + 80 GB HDD SATA II
14:06
<johnny>
ldm in gutsy will make it boot faster, but that's all that will change
14:06
<ltsp>
:(
14:07
<johnny>
it won't make much else go any faster
14:07
hmm how many clients/
14:07
<ltsp>
20
14:07
2 GB is enough ?
14:08
<johnny>
http://doc.ubuntu.com/edubuntu/handbook/C/server-hw.html
14:09iMacGyver has quit IRC
14:09iMacGyver has joined #ltsp
14:09
<ltsp>
256 + (60 * 20) = 256 + 1200 = 1456 MB
14:09
is enough :D
14:09
<johnny>
do you use the server as a desktop?
14:10
<ltsp>
yes :(
14:10
for professor
14:10
professor pc
14:10
<johnny>
add that in too
14:11
and those numbers don't take firefox or openoffice into account
14:11
that's for basic functionality
14:11
those two apps are notoriously memory hungry
14:11
:)
14:11
try with firefox beta 2
14:11
firefox3 beta2
14:11
you might do a bit better with that
14:11
i guess you'll be fine with 128mb on the client
14:11
but 64 is kinda tight with gnome
14:12
even for what happens to run locally on the server
14:12
<otavio>
vagrantc: sorry, was on phone
14:12
<johnny>
whihc isn't much.. but enough :)
14:12
<ltsp>
firefox3 beta is hungry less memory ?
14:12
<otavio>
vagrantc: wants me to take a look on it, now?
14:12
<johnny>
yes
14:12
firefox2 is horrendous to run
14:12
<ltsp>
go go apt-get install !
14:12
<johnny>
on thin clients :)
14:12
or anything reallly :)
14:13
many people were saying they were being treated as second classs citizens to windows
14:13
in relation to firefox
14:13
but firefox3 is much better so far
14:13
<ltsp>
humm i try install firefox3 beta
14:13
<johnny>
i've been running it on my box for a few weeks now
14:13
it's not in apt-get yet
14:14
only alpha8
14:14
<ltsp>
wich xfce
14:14
dou know where im find feisty deb package ?
14:14
<johnny>
if you are ever interested in getting sabayon working tho, talk to me, as i am now committing various patches
14:14
no
14:14Blinny has joined #ltsp
14:14
<johnny>
you're going to have to wait on that i think, unless it's already in hardy-backports..
14:15
err feisty-backports
14:15
<vagrantc>
otavio: well, i've made good progress ...
14:15
<ltsp>
hauuha
14:15
"hardy-backports"
14:15
<vagrantc>
otavio: now i'm just perplexed why it works for floppy, but doesn't quite work for cdrom
14:17
<ltsp>
its normal saves on floppy have a small delay ? 10 20 seconds to save
14:17
<johnny>
yes
14:17
seems to be
14:17
seen that in feisty and gutsy
14:17
<vagrantc>
otavio: maybe i spoke too soon.
14:17
<otavio>
?
14:18
vagrantc: it works?
14:18
<ltsp>
feisty
14:18
<vagrantc>
otavio: mostly
14:18
<Blinny>
...mostly...
14:19
<vagrantc>
otavio: i'm totally confused why it doesn't work for cd but works for floppy
14:19
otavio: and it doesn't remove /media/USERNAME/DEVICE when it's done ...
14:19
<otavio>
where I can grab this code?
14:19
<ltsp>
johnny for better performace i need increase my terminals memorys ?
14:19
<vagrantc>
otavio: i'll push a branch
14:20
<otavio>
vagrantc: and how I can test it?
14:20
<ltsp>
i'm downloading firefox beta2
14:20
<vagrantc>
otavio: you'd need a sid LTSP server and a thin client ...
14:20
otavio: i've been using virtualbox quite happily
14:20
<otavio>
vagrantc: oh! it'll take a long time to setup it
14:20
<johnny>
firefox3 will prolly help
14:20
<vagrantc>
otavio: yes...
14:20
<johnny>
abiword would be better for you than openoffice for documents
14:21
and gnumeric better for spreadsheets
14:21
<ltsp>
jonny i have other issues wich pendrive users
14:21
<otavio>
vagrantc: I'll have dinner and will be back later. (~20min)
14:21
<johnny>
but there is no replacement for presentations yet
14:21
afaik
14:21
<otavio>
vagrantc: then I'll get your branch and we look at it together. ok?
14:21
<ltsp>
abiword is soo poor :(
14:21
<johnny>
by better. i mean lower memory usage and faster
14:21
<vagrantc>
otavio: ok
14:22
<ltsp>
johnny any pendrivers like a mp3players the ltsp can't mount them
14:23
<johnny>
in feisty?
14:23
<ltsp>
yes
14:24
<johnny>
never had that problem
14:24
<ltsp>
i dont test in gutsy yet
14:24
<johnny>
gutsy there is a known problem, and some fixes
14:24
i just lack the unmount option
14:24
as mentioned earlier
14:24
<ltsp>
umount option dont work in nautilus but vagrant says that is normal
14:25
<johnny>
yeah i saw that
14:25
but they did mount for me in feisty
14:25
i don't know what else to say about that as i don't have any feisty installs around anymore
14:25
<ltsp>
hum
14:26
johnny how 'im install the firefox3 beta 2 ?
14:26
<johnny>
unpack it somewhere
14:26
in /usr/local
14:26
/usr/local/apps/ maybe
14:26
i don't know what is preferred
14:26
<ltsp>
hum ok
14:27
and creat a sym link to executable
14:27
<johnny>
yes
14:29
<ltsp>
ltsp@ltsp:~$ /home/ltsp/Desktop/firefox/firefox-bin /home/ltsp/Desktop/firefox/firefox-bin: error while loading shared libraries: libxul.so: cannot open shared object file: No such file or directory
14:29
:(
14:30
<johnny>
not over the firefox2 tho
14:30
make a firefox3
14:30
and not directly to the bin
14:30npman has joined #ltsp
14:31
<ltsp>
ltsp@ltsp:~$ /home/ltsp/Desktop/firefox/firefox and open my firefox 2.0.11 O_o
14:31
i need remove the ff 2.0.11 ?
14:31
<johnny>
no
14:31
do you have firefox2 open tho?
14:31
<ltsp>
yes
14:31
<johnny>
if you have firefox2 open, it will open firefox2
14:32
and not firefox3
14:32
<ltsp>
works now !
14:32
i close firefox 2 :P
14:33
<johnny>
i'm going to be upgrading my clients to firefox3 later on tonight
14:33
keeping the old one tho
14:34
gotta shower..bbiab
14:36
<ltsp>
firefox 3 beta 2 is soo cute haahhaa
14:38Blinny has quit IRC
14:43Gadi has left #ltsp
14:47
<otavio>
vagrantc: back
14:47
vagrantc: where I get the source?
14:52
<vagrantc>
otavio: i've been to busy working on it to push it :)
14:54
too busy
14:54
ok ...
14:54
otavio: do you just want to look at the code, or are you preparing an LTSP environment?
14:55
<otavio>
vagrantc: want to take a look so I could try to help you to hunt it
14:57K_O-Gnom has quit IRC
14:59
<vagrantc>
otavio: http://llama.freegeek.org/~vagrant/bzr/ltspfs/bugs/ltspfs-debian-bts-443200
15:01
so, what i don't understand is why when cdpinger is called ... it doesn't seem to do the "mount --move" part ...
15:05
<otavio>
vagrantc: what file I should take a look?
15:07jammcq has joined #ltsp
15:07
<vagrantc>
otavio: main things to look at are src/lbmount.c
15:08
otavio: scripts/ltspfsmounter
15:08
otavio: src/cdpinger.c
15:08
otavio: i think i've got ltspfsmounter working pretty good
15:13zodman has quit IRC
15:13cliebow_ has quit IRC
15:14
<vagrantc>
otavio: on the server-side, you initiate an ltspfs mount by calling... ltspfsmounter ... so the udev rules and ldm login scripts call that
15:15
otavio: ltspfsmounter creates a dir in /tmp, fuse mounts ltspfs, and calls lbmount to mkdir /media/USERNAME/DEVICE, and then bind-mounts (or moves) to /media/USERNAME/DEVICE
15:16Q-FUNK has quit IRC
15:24* vagrantc notes the upload of ldm and ltsp for debian GNU/Hurd
15:25
<vagrantc>
DGHTSP
15:25
<otavio>
vagrantc: humm
15:26
<vagrantc>
otavio: any ideas ... ?
15:26
<otavio>
vagrantc: ltspmounter is called by udeb with add, right?
15:26
<vagrantc>
otavio: no
15:26
<otavio>
?
15:26
<vagrantc>
otavio: udev, you mean?
15:26
<otavio>
yes yes
15:26
udev
15:26
sorry
15:26
<vagrantc>
ah, yes...
15:27
<otavio>
too used to write udeb ;-)
15:27
<vagrantc>
scripts/*_fstab_entry is what calls ltspfsmounter
15:27
<otavio>
right
15:27
<vagrantc>
as well as ldm's /usr/share/ldm/rc.d/*
15:27
<otavio>
good
15:27
so it then calls it for the CDROM ... right?
15:27
<vagrantc>
well, cdpinger calls *_fstab_entry
15:28
<otavio>
cdpinger?
15:28
<vagrantc>
since there's no udev events for cd insertion or removal
15:28
<otavio>
openin git
15:28
<vagrantc>
otavio: src/cdpinger
15:28
<otavio>
ahh ok
15:28
<vagrantc>
otavio: src/cdpinger.c
15:28
<otavio>
well .. there's hall for it
15:28
hal does have cd pinger like
15:29
vagrantc: good. It basically pings CD every 3 seconds and check its status
15:29
<vagrantc>
everything works with cdpinger, except it doesn't mount to /media/USERNAME
15:30
i.e. the /tmp/.USERNAME-ltspfs/cdrom mount kicks in
15:30
<otavio>
vagrantc: so it _does_ calls add_fstab_...?
15:30
<vagrantc>
otavio: yes, cdpinger calls add_fstab_entry and remove_fstab_entry ...
15:30
you know what ...
15:30
the floppy never gets called through add_fstab_entry or remove_fstab_entry
15:30
it's always done through the LDM scripts
15:30
<otavio>
humm
15:31
<vagrantc>
that's the difference ... though both do call ltspfsmounter
15:31
<otavio>
if _you_ call add_fstab byhand for the CD, does it work too?
15:32
<vagrantc>
ah, good thinking
15:33
ldm calls .... ssh -X -S ${LDM_SOCKET} ${LDM_SERVER} "/usr/sbin/ltspfsmounter ${DIR} add"
15:33
and add_fstab_entry: /usr/bin/ssh -X -S ${LDM_SOCKET} ${SERVER} \ "/usr/sbin/ltspfsmounter ${MOUNTPOINT} add"
15:35
<otavio>
humm
15:35
and are their equivalent?
15:35
SERVER == LDM_SERVER?
15:35
DIR == MOUNTPOINT?
15:36
SERVER looks to be
15:36
but not sure since I cannot try it here
15:38Guaraldo has left #ltsp
15:39
<otavio>
vagrantc: checked the values?
15:40
vagrantc: try to call ssh by hand and see if it works or not
15:40
<vagrantc>
add_fstab_entry gets an appropriate value for $SERVER
15:40
it's a little tricky to check each thing
15:40
and i got to eat some food :)
15:40
<otavio>
ahh ok
15:40
np, ping me when back
15:41
vagrantc: ah, my mom sent you a "hello"
15:41
<vagrantc>
otavio: hola a tu madre
15:41
<otavio>
vagrantc: :-) We're looking forward to meet you again on Argentina, later this year! .-)
15:42
<vagrantc>
otavio: indeed!
15:42
hope i can get enough work in the next few months to stay a while
15:44
<otavio>
vagrantc: I'll try to stay too. Dunno if it'll be possible or not. Let's see
15:46Guaraldo has joined #ltsp
15:46mikkel has quit IRC
15:47
<vagrantc>
otavio: ok, calling add_fstab_entry by hand seems to get it mounted to /media/USERNAME/cdrom
15:48
otavio: and now, running it manually for floppy0 only gets as far as mounting to /tmp/.USERNAME-ltspfs/floppy0
15:48
so it seems the second mount is what causes it to fail
15:48
or, partially fail
15:48
<otavio>
yes
15:49
but it's better since you can test it byhand now
15:49primeministerp has quit IRC
15:50primeministerp has joined #ltsp
15:53
<vagrantc>
ok ... now i'm getting some error messages :))
15:56
"Error: can't bindmount under %s: already mounted\n",
15:56
in mkdir_safe
15:56
otavio: so it's trying to check if /media/USERNAME is mounted, and is_mounted is returning true ...
15:57
<otavio>
yes.
15:57
and is it?
15:57
<vagrantc>
no, but /media/USERNAME/FOO is mounted ... so something is prabably messed up there ...
15:58Guaraldo has quit IRC
15:58
<vagrantc>
ah, it's probably my || (mode = "umount")
15:58
<otavio>
and _whhy_ it doesn/ t use same media place? wasn't it suppose to be the same?
15:58
ahh ehhe
15:59
<vagrantc>
otavio: well, lbmount needs to basically: mkdir /media/USERNAME, and then mkdir /media/USERNAME/DEVICE
16:00Guaraldo has joined #ltsp
16:00
<otavio>
what I means is: why not in /tmp/.USERNAME-ltspfs/ too?
16:02
<vagrantc>
because it causes debian bug 443200
16:02
otavio: oh...
16:02
otavio: ltspfsmounter handles /tmp/.USERNAME-ltspfs/*
16:02
otavio: that part can be handled in python as the user- lbmount is a setuid root binary.
16:05
ltspfsmounter can create /tmp/.USERNAME-ltspfs/, do the fuse mounts ... and then calls lbmount to "mount --move/--bind" into /media/USERNAME/
16:05
<otavio>
yes
16:05
<vagrantc>
did i answer your question, or do i not understand your question?
16:08
<otavio>
I think you answered my question
16:08
<vagrantc>
it would be a lot easier if we could just mkdir -p /media/USERNAME ; chmod g+w /media/USERNAME
16:09
then we can just mount directly to /media/USERNAME ... but people objected to that on the groups that someone could fill up / with files (where /media is most likely mounted)
16:09
<otavio>
and why we can't?
16:09
ahh
16:09
<vagrantc>
whereas, a good sysadmin will split /tmp off of /
16:10
<otavio>
yes or leave it as tmpfs
16:10
<vagrantc>
though, arguably, you could just as easily split /media off of / if you were worried
16:11
so i'm trying to do better than what we have now... and it's so close...
16:12outofrange has left #ltsp
16:12
<otavio>
it should be a really stupid mistake
16:13
is floopy also being --bind?
16:13
<vagrantc>
well, the current implementation uses --bind, but i'm trying to switch to using --move
16:13
because then only a single mount shows up
16:14
and it just looks nicer
16:14
:)
16:15
<otavio>
yes
16:16
<vagrantc>
it's a little confusing how "lbmount device" and "lbmount --umount device" are distinguished in the code...
16:16
big if statements
16:16
<otavio>
it probably deserves a refactoring ;-)
16:17staffencasa has quit IRC
16:18
<vagrantc>
otavio: i'm having a hard time not refactoring it... but i want to keep the diff understandable
16:18
<otavio>
vagrantc: so invert it.
16:18
<vagrantc>
?
16:18
<otavio>
vagrantc: refactor it and then go to your improvement .-)
16:18
<vagrantc>
hmmm...
16:22Guaraldo has quit IRC
16:23ltsp has quit IRC
16:26
<vagrantc>
otavio: ok, so i switched back to the single-line patch
16:27
well, still patched ltspfsmounter
16:27
but only the change from --bind to --move
16:27
in lbmount.c
16:27
works pretty good
16:28
but it leaves behind the /media/USERNAME/* directories
16:31
hmmm...
16:44
i think i have it
16:44* vagrantc dances a bit
16:45jammcq has quit IRC
16:47
<vagrantc>
otavio: having worked on it all day, i just messed it up.
16:47
otavio: but i started over and got it working :)
16:50ccherret1 is now known as ccherrett
16:51slidesinger has quit IRC
17:10
<vagrantc>
whew!
17:15
<otavio>
works?
17:15
<vagrantc>
otavio: seems to :)
17:15
<otavio>
awesome!
17:16
<vagrantc>
otavio: much less invasive patch than my previous attempts
17:17
<otavio>
that's even better
17:17
<vagrantc>
well, they're also not as good ... i.e. we're still using predictable paths in /tmp
17:18
<otavio>
vagrantc: but now you can improve it, step be step
17:18
<vagrantc>
otavio: exactly
17:18
<otavio>
vagrantc: that's the best way to work
17:18
<vagrantc>
otavio: i woke up at 3am thinking about this, and have been working on it for 11-12 hours :)
17:18
otavio: eventually i decided i had to start over on the actual code, and poof, it was easy :)
17:18
<otavio>
vagrantc: wow! this makes me remember Mexico! ;-)
17:19bobby_C has quit IRC
17:19
<vagrantc>
the 1-line patch switching --bind to --move was really impressive ... that almost worked without change
17:21
so i switched back to that, and then tried to figure out what else was broken...
17:50* vagrantc commits upstream
18:06ari_stress has joined #ltsp
18:06
<ari_stress>
morning all :D
18:33
<aboo0ood>
morning ari_stress
18:34
guys, i have a question about LTSP boot process
18:35
Does the client request the IP address of the server where the kernel resides, then requests the IP address of the application server ?
18:35
2 requests ?
18:36
<ari_stress>
what is application server?
18:37
<aboo0ood>
ari_stress, i mean the server where the chroot environment resides
18:37
<ari_stress>
oh
18:37
as far as i know it only request the kernel
18:38
<aboo0ood>
http://en.wikipedia.org/wiki/Linux_Terminal_Server_Project
18:38
ari_stress, look at this
18:38
it says there'a a (new) DHCP request after the first one
18:38
<ari_stress>
we can specify the X server and log server to on another ip
18:38
<vagrantc>
typically, you get one DHCP request for the bootloader, and one DHCP request from the initramfs whre to mount it's root filesystem
18:40
<aboo0ood>
thanx guys :)
18:41Q-FUNK has joined #ltsp
18:41
<ari_stress>
aboo0ood: you can see the details in lts-parameter.gz
18:46spectra has quit IRC
18:47
<aboo0ood>
vagrantc, when we say chroot environment we are talking about /opt/ltsp .. then we say that the client mounts its root filesystem, where does it mount it ?
18:47
<vagrantc>
on "/"
18:48
<aboo0ood>
vagrantc, what is the full path on the server ?
18:48
<vagrantc>
typically, /opt/ltsp/ARCH
18:48
i.e. /opt/ltsp/i386
18:48
<aboo0ood>
mmm
18:48
<vagrantc>
the most common
18:48
<aboo0ood>
i am lost
18:48
chroot env. is already there
18:48
why is it remounting it
18:49
?
18:49
<vagrantc>
already where?
18:49
<aboo0ood>
/opt/ltsp/ARCH
18:49
<vagrantc>
the thin-client does something kind of like this
18:49
:
18:50
mount -t nfs 192.168.1.255:/opt/ltsp/i386 /
18:50
er
18:50
.255 is a bad example
18:50
mount -t nfs 192.168.1.254:/opt/ltsp/i386 /
18:50
where 192.168.1.254 is the ip address of the server
18:50
<aboo0ood>
aha
18:51
it mounts something on the server .. on the server , right ?
18:51
<vagrantc>
from the server
18:51
<aboo0ood>
yes
18:51
<vagrantc>
the server provides the files to the thin-client
18:51
<aboo0ood>
and the thin-client mounts it on the server ? on "/" ? but where is this "/" relative to the root of the server ?
18:54
vagrantc, does it mount it back to /opt/ltsp/i386 ? which is "/" relative to the client ??
19:04
<vagrantc>
aboo0ood: i really don't understand what you're asking.
19:04
it mounts it at "/", just like you would typically mount a hard drive at "/"
19:05
instead of /dev/hda1 or whatever, you have 192.168.1.254:/opt/ltsp/i386
19:05
aboo0ood: i think you're trying to make this more complicated than it is
19:05
<aboo0ood>
it looks to me like mount /dev/hda1 /dev/hda1 , that's why i am confused ..
19:05
<vagrantc>
no.
19:06
<aboo0ood>
since "/" is really the chroot evn. path
19:06
<vagrantc>
it's mount -t FILESYSTEM DEVICE PATH
19:06
in the case of a hard disk, FILESYSTEM might be ext3, DEVICE might be /dev/hda1 and PATH is /
19:07
in the case of NFS Root, Filesystem is nfs, DEVICE is 192.168.1.254:/opt/ltsp/i386 and PATH is /
19:07
<aboo0ood>
ok, let me ask it in another way
19:08
does issuing the mount command on the server (say manually ) works the same as when its issued by the client ?
19:08
work *
19:10
<vagrantc>
i just don't understand your question.
19:10
<aboo0ood>
mmm
19:10
<vagrantc>
there are files on the server at /opt/ltsp/i386.
19:10
<aboo0ood>
i think i got it :)
19:10J45p3r has joined #ltsp
19:10
<vagrantc>
the thin-client mounts those files, and sees them as /
19:10
<aboo0ood>
i was confused since this "/" resides in the end on the server
19:11
<vagrantc>
the server is just a file server
19:11
aboo0ood: chroot /opt/ltsp/i386 will see the same "/" as the thin-client (for the most part)
19:11
they're just files, not magic beans
19:12
<aboo0ood>
lol
19:12
vagrantc, thanx :)
19:14
<vagrantc>
i guess i've just been doing this too long
19:19Q-FUNK has quit IRC
19:27jammcq has joined #ltsp
19:37daniQ has joined #ltsp
19:47npman has quit IRC
19:53nobios has joined #ltsp
19:56nobios has quit IRC
19:58nobios has joined #ltsp
20:10vagrantc has quit IRC
20:12daniQ has quit IRC
20:15RiXtEr has joined #ltsp
20:27
<RiXtEr>
Hey all anyone around?
20:29
<ari_stress>
wazzup
20:29
<RiXtEr>
just saying hi
20:30
rjune_: you around? Long time no c
20:44RiXtEr_ has joined #ltsp
20:46RiXtEr has quit IRC
20:46RiXtEr_ is now known as RiXtEr
20:57
<rjune_>
RiXtEr: yup
21:04
<RiXtEr>
rjune_: hows it going
21:08
<rjune_>
ok
21:08
http://www.bravegnuworld.com/~rjune/mov005.mod
21:17J45p3r has quit IRC
21:46nobios has quit IRC
22:16
<RiXtEr>
rjune_: what opens a mod file?
22:17
<rjune_>
mplayer
22:17
it's just an mpeg1 file
22:17
or maybe mpeg2
22:17
<RiXtEr>
gotcha
22:17
<rjune_>
still, should be most players
22:18* RiXtEr is in dows
22:18
<RiXtEr>
win dows that is ;)
22:18
downloading that file at a blazing 17.5 KB/s
22:19
Much better than the 3-4 KB/s i was getting with my dialup
22:19
cellphone internet is great!
22:25
that looks fun.. what kinda heli is that and how much does it cost
22:29
<rjune_>
heh
22:29
5G6
22:29
it was a christmas present, they run about $130
22:29
http://www.bphobbies.com/ sells them
22:30
<RiXtEr>
eh... too much for me i'd have it broken in about 10 mins
22:30
or less :)
22:31open_lad has joined #ltsp
22:31
<rjune_>
start with a Mosquito
22:32
or a cheap foamie
22:32
though this is a pretty robust heli, I had my worst crash today
22:32
fell out of the air at about 18'
22:32
<RiXtEr>
did you 'total' it
22:33
<rjune_>
nope
22:34
I did knock one of the gears off though
22:34
<RiXtEr>
nice
22:34
ow
22:34
<rjune_>
and bent the hell out of my landing gear
22:34
<RiXtEr>
i'll bet
22:34
<rjune_>
that tail you see isn't stock, the stock one looks like a truss
22:34
<RiXtEr>
I see that in the picture
22:35
does it handle different with the tail you have?
22:35
<rjune_>
currently the tail is a carbon fiber job we did as a proof of concept
22:35
yeah, the one currently on is about 2" at the end and about 1" towards the body
22:36
and it's covered in duct tape, so it's an actual fin instead of a skeleton
22:36
<RiXtEr>
you build carbon fiber stuff ?
22:36
<rjune_>
got rod from the hobby shop
22:36
<RiXtEr>
ah
22:36
:)
22:36
<rjune_>
cut it with a dremel and hot melt glued it together
22:37
we have a roughly seven minute video of me flying around earlier.
22:37
it ends by me falling out of the air
22:38
<RiXtEr>
run out of rf range?
22:38
<rjune_>
no, tried a manouvre that is outside my, or the helicopter's ability
22:39
<RiXtEr>
barrel roll?"
22:39
<rjune_>
heh, no sudden stop
22:39
see that rod with two weights up top?
22:39
that keeps you from stopping too quickly
22:39
you stop fast, you fall fast.
22:40
<RiXtEr>
ah... i thought that was another propeler
22:40
propeller
22:40elisboa has quit IRC
22:41elisboa has joined #ltsp
22:41
<rjune_>
nope.
22:41
it's called a flybar
22:41
the top blades are attached to it
22:41
<RiXtEr>
i c
22:41
<rjune_>
the weights tend to keep the top blades perpendicular to the shaft.
22:42
it provides stabilization
22:42
<RiXtEr>
gonna change the subject.... when do you see beryl / compiz ran with ltsp :)
22:42
<rjune_>
no clue
22:42
<RiXtEr>
lol
22:43
should ask jim
22:43
<rjune_>
probably
22:43
should I have another piece of cake?
22:43
Morgan made a My Melody cake for Jen's birthday
22:44
<RiXtEr>
that'd be great... could you just see the x server running 20 clients with xgl .... WOW CPU USAGE!!
22:44
My Melody cake?
22:44
<rjune_>
heh
22:44
yeah
22:44
google images for "My Melody"
22:44
it's a hello kitty character
22:44
<RiXtEr>
gotcha
22:44Joris has joined #ltsp
22:45
<RiXtEr>
kiddo is 3 now?
22:45
or 4?
22:45
mine turned 1 last month
22:47
<rjune_>
Morgan is 12
22:47
Merlin is 4
22:47
(in Feb)
22:48
<RiXtEr>
wh
22:48
<rjune_>
Connor is 2
22:48
<RiXtEr>
ah
22:48
I thought (for some dumb reason) you only had 2
22:48
<rjune_>
two boys and one girl
22:48
brb
22:50* RiXtEr thinks rjune is getting some cake ;)
22:53
<rjune_>
no
22:53
putting my kid to bed, *again*
22:54
<RiXtEr>
Holy cow... have you seen what you can do with a wii remote ?
22:55
<rjune_>
if it goes along some of our earlier conversations, no. but I sure want to
22:55
<RiXtEr>
get on youtube and search for jcl5m
22:55
This guy is a friggen genius
22:57
<rjune_>
cool
22:58* RiXtEr owns a wii
23:00
<rjune_>
bitch.
23:00* rjune_ wants one
23:01Joris_ has quit IRC
23:02
<rjune_>
so who plays with your wii more, you or the wife?
23:02
<RiXtEr>
wii play equally
23:02
;)
23:02
wii would like to play :)
23:19RiXtEr has quit IRC
23:47zodman has joined #ltsp