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


Channel log from 1 July 2008   (all times are UTC)

00:05mccann has quit IRC
00:19plamengr has joined #ltsp
00:35vagrantc has joined #ltsp
00:40Egyptian[Home] has joined #ltsp
00:42subir has joined #ltsp
00:45basanta has joined #ltsp
00:51Ronin1234 has joined #ltsp
00:51RoninBaka has quit IRC
01:01MacIver has quit IRC
01:18deavid has joined #ltsp
01:21MacIver has joined #ltsp
01:35vagrantc has quit IRC
01:39Ronin1234 has quit IRC
01:39RoninBaka has joined #ltsp
01:40Pascal_1 has joined #ltsp
01:44ace_suares has quit IRC
01:45ace_suares has joined #ltsp
02:22chrisjrob has joined #ltsp
02:38basanta has quit IRC
02:38xachen has quit IRC
02:39basanta has joined #ltsp
02:39xachen has joined #ltsp
02:58alekibango has joined #ltsp
03:20makghosh has quit IRC
03:21makghosh has joined #ltsp
03:27Q-FUNK has joined #ltsp
03:29mikkel has joined #ltsp
04:38Ronin1234 has joined #ltsp
04:38RoninBaka has quit IRC
04:40Nubae has joined #ltsp
04:47Egyptian[Home1 has joined #ltsp
04:51Ronin1234 has quit IRC
04:51RoninBaka has joined #ltsp
04:53Egyptian[Home] has quit IRC
05:14makghosh has quit IRC
05:19makghosh has joined #ltsp
05:36Nubae has quit IRC
05:36Nubae has joined #ltsp
05:48Rucal has joined #LTSP
05:53RoninBaka has quit IRC
05:53RoninBaka has joined #ltsp
06:16cyberorg has quit IRC
06:16ogra_cmpc has quit IRC
06:17ogra_cmpc has joined #ltsp
06:18basanta has quit IRC
06:35elisboa-off has quit IRC
06:41Blinny has joined #ltsp
06:54dmseg has joined #ltsp
06:56
<dmseg>
hello, i would like to know if i can run an Ltsp server in one place and it uese the mirror of my freenas server (ftp , tftp , smb) also i would like to know how much bandwith is requred by a thinclient(using broadband)
07:00dmseg has left #ltsp
07:00
<Blinny>
Huh?
07:02deavid has quit IRC
07:18jammcq has joined #ltsp
07:18K_O-Gnom has joined #ltsp
07:24Q-FUNK has quit IRC
07:27K_O-Gnom has quit IRC
07:29ace_suares has quit IRC
07:30ace_suares has joined #ltsp
07:31DonSilver has joined #ltsp
07:33DonSilver has quit IRC
07:34DonSilver has joined #ltsp
07:37cyberorg has joined #ltsp
07:40otavio has joined #ltsp
07:42slidesinger has joined #ltsp
07:49elisboa has joined #ltsp
07:53mhterres has joined #ltsp
07:58cliebow has joined #ltsp
07:58mikkel has quit IRC
08:01dyn has quit IRC
08:02subir has quit IRC
08:18ogra_cmpc has quit IRC
08:22ogra_cmpc has joined #ltsp
08:28epoxy|w3rk has joined #ltsp
08:48plamengr has quit IRC
08:53DonSilver has quit IRC
08:54ace_suares has quit IRC
08:56elisboa has quit IRC
08:58
<epoxy|w3rk>
hey i stumbled accross this http://www.ltsp.org/twiki/bin/view/Ltsp/PXESplash while hunting for splash screen docs
08:59
what 'kernel dir' are they talking about?
09:00mccann has joined #ltsp
09:04Subhodip has joined #ltsp
09:04elisboa has joined #ltsp
09:06F-GT has quit IRC
09:12Subhodip has quit IRC
09:13K_O-Gnom has joined #ltsp
09:18F-GT has joined #ltsp
09:20vagrantc has joined #ltsp
09:22fie_wr0k has joined #ltsp
09:27Gadi has joined #ltsp
09:27
<Gadi>
Blinny: ping
09:28
<ltsppbot>
"Gadi" pasted "This one should do the trick" (39 lines) at http://pastebot.ltsp.org/10
09:29
<Blinny>
Gadi: pong!
09:30Subhodip has joined #ltsp
09:30Pascal_1 has quit IRC
09:30
<Gadi>
Blinny: so, while I was in the car picking up the kids yesterday, I had a thought
09:30
we dont need LDM_USERNAME at all
09:31
<Blinny>
Kids are good for that sort of thing.
09:31
Why not?
09:31
<Gadi>
just use the $USER from the server's ssh connection
09:31
so, see the above paste
09:32
<Blinny>
And $USER is equivalent to LDM_USERNAME post-login?
09:32
<Gadi>
yup
09:32
bec it is the $USER of the ssh connection
09:32
and you connect as the user you log in as
09:32
QED
09:32
and such
09:32
:)
09:32
<ogra>
you wont need the first command
09:33
<Gadi>
just need to use single quotes instead of doubles
09:33
:)
09:33mikkel has joined #ltsp
09:33
<ogra>
$USER will come from the servers env anyway, no need to explicitly pull it
09:33
<Gadi>
ogra: ?
09:33
<ogra>
userProcs isnt needed
09:33
<Gadi>
sure it is
09:34
<ogra>
oh, sorry
09:34
indeed it is
09:34
<Gadi>
this scriptie should work - i think
09:34* ogra is to distracted by work stuff, sorry
09:34
<Blinny>
This doesn't necessarily need to be chmod +x does it?
09:34
<Gadi>
yeah, that was me yesterday :)
09:34
Blinny: shouldnt - it gets sourced
09:34
<Blinny>
Right-O.
09:35
And where use singles? ~line15?
09:35
the sshdChildren line
09:35
<ogra>
that walks the processtree, right ?
09:35
<Gadi>
copy what I pasted exactly
09:35
<Blinny>
Ah. Ok.
09:36
<Gadi>
ogra: yeah - it will kill all procs except for those related to the newest sshd connection (which is our socket)
09:36
<ogra>
sshdChildren=$(ssh -X -S ${LDM_SOCKET} ${LDM_SERVER} "/usr/bin/pgrep -P ${PPID}")
09:36elisboa is now known as elisboa-off
09:36
<Gadi>
?
09:36
that doesnt pake sense
09:36* vagrantc wonders if it could use pattern negation instead of an ugly for loop
09:36
<Gadi>
*make
09:37
vagrantc: lets get it working first
09:37
:)
09:37
<vagrantc>
heh :)
09:37
<ogra>
Gadi, you want all processes that are childern of the currently running sshd
09:37
/usr/bin/pgrep -P ${PPID} in teh server will get you that
09:38
<Gadi>
ah
09:38
well, same thing
09:38
still would need to grab the value of $PPID
09:38
<ogra>
i'm pretty sure you can merge sshdChildren and sshdProc
09:38
<Gadi>
as well as the value of the kids
09:38
<ogra>
with some thinking
09:38
to save one tunneled request
09:39
<Blinny>
Hrm. Clicking 'Cancel' doesn't return to LDM. Just stays on black screen w/ cursor.
09:39
<Gadi>
yeah - im not prone to thinking
09:39
:)
09:39
yeah, ogra - that I need more help with
09:39
gracefully killing ldm
09:39
we dont store the pid, do we?
09:40
<ogra>
nope, not atm
09:40
<Gadi>
Blinny: does "OK" work?
09:40
<Blinny>
Gadi: It did yesterday.
09:40
Gadi: I'll reboot & test.
09:40
<Gadi>
cool
09:41
(yeah - but yesterday, we were hardcoding LDM_USERNAME for testing - today we use $USER)
09:41
<Blinny>
Right
09:42
*poof* hopefully
09:42Blinny has quit IRC
09:42
<Gadi>
yeah!
09:42
he died!
09:42
:)
09:42
how morbid
09:42Blinny has joined #ltsp
09:43
<Blinny>
Ok works. (;
09:43
<Gadi>
welcome back!
09:43
cool
09:43
so, now we just need to kill ldm well
09:43
on cancel
09:44
but, at least "OK" works
09:44
:)
09:44
<Blinny>
You should still add the -center to your copy before -buttons :P
09:45
<Gadi>
I wonder what a: pkill -P ${PPID} would get us
09:45
<Blinny>
This isn't NeXTStEP ya know (;
09:45
<Gadi>
heh
09:45
adding now...
09:46
hmm...
09:46
<Blinny>
I thought we tried the pkill PPID yesterday.
09:46
<Gadi>
did we?
09:46
damn
09:46
<Blinny>
I can't say for certain.
09:46
<ogra>
with proper brackets ?
09:46
<Blinny>
I'm sure you're using subversion for this behemoth.. can you look back? (;
09:47
<ogra>
it will only work as ${PPID}
09:47
<Blinny>
May as well try it again
09:47
<Gadi>
can you login to the thin client console and do: ps -ale >/tmp/junk
09:47
and paste /tmp/junk
09:47
(Im not on a client atm - and then I can see the process tree)
09:48
<Blinny>
Any ideas how to accomplish such a feat? (pasting from the thin client's terminal)
09:49
<Gadi>
scp the file to the server
09:49
so:
09:49
<Blinny>
Oh. Heh
09:49
<Gadi>
well, u get it
09:49
<ltsppbot>
"Blinny" pasted "beer" (42 lines) at http://pastebot.ltsp.org/11
09:50
<epoxy|w3rk>
if i add usplash, would i need to throw 'splash' in /whatever/tftpboot/boot/pxelinux.cfg or /usr/opt/i386/boot ?
09:50
gah.. ruined the path names.. but you get what i mean :)
09:51
<ogra>
epoxy|w3rk, it should be in there by default (and be set up by default)
09:51
oh, wait, you were the etch user ...
09:51* ogra remembers
09:51
<epoxy|w3rk>
haha
09:51* epoxy|w3rk sighs
09:51
<ogra>
well, yeah, that *might* work
09:51
i never tried usplash in debian
09:51Subhodip has quit IRC
09:51
<epoxy|w3rk>
well it sucks :)
09:52
i threw it in both.. to no avail. i can just stick with quite... my boss just doesnt want users seeing init scroll by
09:52
<ogra>
or use ubuntu :)
09:52
<epoxy|w3rk>
true..true :)
09:53F-GT has quit IRC
09:54Subhodip has joined #ltsp
09:54
<Blinny>
Gadi: pkill'ing the PPID provides different result - back to blinkenkursor
09:54
I lost the mouse & x resolution
09:55
Can't we just throw a -HUP to one of these?
09:55
<Gadi>
blinny did you do: pkill -P ${PPID}
09:55
?
09:55
<Blinny>
Yes.
09:56
<Gadi>
try this:
09:56
<Blinny>
pkill -P ${PPID}
09:57
<Gadi>
hmmm...
09:57
we need to kill ldm, too
09:57
so: pkill -P ${PPID}
09:57
pkill ${PPID}
09:57
<rjune>
!g
09:57
<ltspbot>
rjune: "g" is Gadi!!!!!!!!!!!!!!!!!!!!!!!!
09:57
<Gadi>
rjune!
09:58
<Blinny>
updating now
09:59
<Gadi>
hopefully we dont kill ourselves before we can kill daddy
09:59
if we are, we may need to pgrep everything and then go about killing them all
09:59
:)
10:00
or maybe there's a better way...
10:00
rjune: you're an assassin by trade - how do you make it look so easy?
10:01
oh, wait
10:01
ogra, vagrantc: is ldm called in a loop from the screen script?
10:01
or is the loop maintained from within ldm?
10:01
<rjune>
Gadi: you know I can't talk about that.
10:02* Gadi seems to think that we did it right by killing PPID
10:02
<Gadi>
but, that killed ldm
10:02
and the screen script doesnt respawn it
10:02
<rjune>
killall -9 1
10:02* Gadi checks code...
10:02
<rjune>
killall -9 init
10:02
<ogra>
then you didnt use the server side PPID
10:02
<rjune>
that'll fix all your problems
10:03
<Gadi>
oh... the server side PPID
10:03
that's a good idea
10:03
<ogra>
:)
10:03
<Gadi>
:)
10:03
<Blinny>
pkill ${PPID} didn't kill anything, just logged the user right in.
10:03
<Gadi>
yeah, here
10:03
Ill paste a better one
10:03
<Blinny>
Does it have the -center? (;
10:04
<ltsppbot>
"Gadi" pasted "This one should do the trick - really!" (39 lines) at http://pastebot.ltsp.org/12
10:05
<Gadi>
or alternatively...
10:05
<ltsppbot>
"Gadi" pasted "This one should do the trick - really!" (39 lines) at http://pastebot.ltsp.org/13
10:05twinprism has quit IRC
10:05
<Gadi>
something along either of those lines
10:06
one kills the socket's parents (i think)
10:06
<Blinny>
Is it just that one line that changed?
10:06
<Gadi>
the other should kill the socket's parents' kids
10:06
yeah
10:06
note the single quotes
10:06
that's important
10:07
<Blinny>
Which to try first?
10:07Ronin1234 has joined #ltsp
10:07
<Gadi>
hmmm
10:07RoninBaka has quit IRC
10:07
<Gadi>
try the first one first
10:07
and the second one second
10:08
its only fair
10:08
the first one *did* come first
10:08
<Blinny>
We wouldn't want to make it feel slighted.
10:10
OK. First one no worky. Back to mouse & blackscreen.
10:10F-GT has joined #ltsp
10:13
<Blinny>
OK. Second one no worky. Back to mouse & blackscreen.
10:13
<Gadi>
hmm...
10:14
just make sure on the client that the file is changing
10:14
wouldnt want to chase a rabbit
10:14
or wild goose
10:14
<Blinny>
It is.
10:14
<Gadi>
or any kind of farm animal or undomesticated cousin
10:14
<Blinny>
I had checked that - thinking the same thoughts.
10:15
<Gadi>
are you still at the blackscreen?
10:15
stay there a sec
10:15
and do a ps -efw on the server and look to see how many sshd procs are running as you
10:16
oh, duh!
10:16
we can do better than PPID
10:17
<ltsppbot>
"Gadi" pasted "This one should do the trick - really!" (39 lines) at http://pastebot.ltsp.org/14
10:17
<Gadi>
we have sshdProc!
10:17
<Blinny>
Two ssh processes, no sshd
10:17
<Gadi>
thats the one we need to kill
10:17
on the server?
10:18
you should have sshd procs
10:18
<Blinny>
Sorry. Was on the client.
10:18
For this user, I have 6.
10:18
3 groups of 2
10:18
(rebooting client now w/ sshdProc)
10:24
I'm sorry to say, black screen & cursor again.
10:25
<Gadi>
damn
10:27* Blinny nods.
10:28* Gadi thinking...
10:28
<ogra>
do you kill ldm or also X ?
10:28* ogra would only kill X
10:28
<Gadi>
no, now we are killing the server-side sshd
10:28
as you suggested
10:28
the server side of the socket
10:28
<ogra>
hmm, the right one ?
10:28
here are two
10:29
the first only gets the tunnel up ... the second starts the session
10:29
<Gadi>
right
10:29
<ogra>
the rc scripts sit between the two calls iirc
10:29
<Gadi>
well, there is one sshd, no?
10:29
doesnt the same one control all
10:29
<ogra>
i'm not sure it doesnt spawn a second one
10:30
<Gadi>
because the second ssh call uses the same socket
10:30
<ogra>
right but if you catch the second instead of the first it wont kill ldm
10:30
check that
10:30
<Gadi>
will do
10:30
<ogra>
i'm not sure if there is one or two sshds
10:31
<Blinny>
I see one as root, one as the user.
10:31
In each instance, 2 PIDs separate the two
10:32
sshd: myUser [priv] and sshd: myUser@pts/9
10:33staffencasa has joined #ltsp
10:33
<Gadi>
ogra: it doesnt spawn another sshd
10:34
I did a quick test just ssh'ing into a machine - once with a master socket and the second using that socket
10:34
with both connected, only one sshd
10:39
<ogra>
oki
10:43plamengr has joined #ltsp
10:58mccann has quit IRC
10:58makghosh has quit IRC
11:00F-GT has quit IRC
11:01makghosh has joined #ltsp
11:05vagrantc has quit IRC
11:11vagrantc has joined #ltsp
11:13RoninBaka has joined #ltsp
11:13Ronin1234 has quit IRC
11:15mccann has joined #ltsp
11:17japerry has joined #ltsp
11:19
<Blinny>
Bummer.
11:31mighty-d has joined #ltsp
11:31
<cliebow>
/PRIV jammcq
11:31
<mighty-d>
Hi
11:31
My thin clients are running slow, im on hardy as the ltsp server, i wanted to tweak the lts.conf but it isnt on the tftp root, should i add one there?
11:32alekibango has quit IRC
11:32
<Gadi>
mighty-d: yes
11:33
in: /var/lib/tftpboot/ltsp/i386/lts.conf
11:33
same old structure
11:33dyn has joined #ltsp
11:33
<mighty-d>
Gadi: ok thanks, any hint on what could be causing the slow behaviour?
11:34
<Blinny>
LDM_DIRECTX ?
11:34
<Gadi>
is it slow to boot or slow once you are logged in?
11:34
<mighty-d>
gadi, slow once im logged inç
11:34
<Blinny>
mighty-d: though I've noticed that hardy is much more sluggish than my FC6+LTSP4.2 box was
11:34
<Gadi>
then, try setting LDM_DIRECTX=True
11:34
as blinny says
11:35
<mighty-d>
Blinny: this clients where running on a slackware 11 + ltsp 4.2 and run ok
11:35
<Gadi>
by default, all video is securely tunneled thru ssh
11:35
which tends to be slow on older clients especially
11:35
LDM_DIRECTX turns that off
11:35
<Blinny>
I blame gnome eycandy.
11:36
<mighty-d>
Ok, lets see
11:38
<Blinny>
mighty-d: Don't forget to create the [default] section header in lts.conf.
11:39
<mighty-d>
Blinny: yeah, thanks, it seems to work
11:40F-GT has joined #ltsp
11:40
<Blinny>
mighty-d: Weeha!
11:40
<mighty-d>
Gadi: thanks, Blinny, thanks, ssh its taking too much time to log in, is there a way to tune it up
11:40
?
11:40
Blinny: yeah, i was breaking my head with that one
11:41
<Blinny>
mighty-d: Sounds like name resolution issues.
11:41
mighty-d: You can either add all your hosts to /etc/hosts, or set them up so they resolve correctly (192.168.0.1->ws001 and ws001->192.168.0.1)
11:41
mighty-d: Dig?
11:42
<mighty-d>
Blinny: thanks, i gotta change my chroot password... i dont remember how to build the squashfs on ubuntu, would you give me that last clue?
11:42
<Blinny>
mighty-d: ltsp-update-image --arch i386
11:42
mighty-d: You can leave the --arch i386 if your server is not x86_64 though
11:43
:s;leave;drop
11:43Rucal has left #LTSP
11:43
<mighty-d>
Blinny: thanks a lot, and i mean A LOT! ;)
11:43
<Blinny>
mighty-d: No worries mate.
11:43
mighty-d: http://doc.ubuntu.com/edubuntu/edubuntu/handbook/C/ ain't a bad handbook, even if you're not using the Edubuntu flavor.
11:44
<mighty-d>
ok, gotta read it, thanks Blinny
11:44
<Blinny>
Cheers.
11:47
And on that note, it's off to job #2. Fun!
11:47Blinny has quit IRC
11:56
<mighty-d>
Gadi: any idea of how to speed up boot?
11:56vagrantc has quit IRC
11:58K_O-Gnom has quit IRC
12:01makghosh_ has joined #ltsp
12:03DonSilver has joined #ltsp
12:04makghosh has quit IRC
12:07Pascal_1 has joined #ltsp
12:07makghosh_ has quit IRC
12:07Pascal_1 has quit IRC
12:09LTSP has joined #ltsp
12:12plamengr has joined #ltsp
12:12plamengr has left #ltsp
12:15Subhodip has quit IRC
12:16Pascal_1 has joined #ltsp
12:19warren has joined #ltsp
12:20plamengr has joined #ltsp
12:20plamengr has left #ltsp
12:20
<LTSP>
Hi! A question I had recently, and knew the team here would know the answer... I'm looking for a faster graphical remote access than VNC (road warrior getting into office server scenario). LTSP has much greater speed than any VNC I've tried on the same LAN with the same equipment. Any suggestions where I should look? Is anyone working on something like this?
12:27F-GT has quit IRC
12:28Pascal_1 has quit IRC
12:28makghosh_ has joined #ltsp
12:33mighty-d has quit IRC
12:37chrisjrob has left #ltsp
12:41elisboa-off has quit IRC
12:42ogra has quit IRC
12:43elisboa has joined #ltsp
12:44ogra has joined #ltsp
12:52
<cliebow>
LTSP:you looked at nx?
13:00F-GT has joined #ltsp
13:01
<irule>
LTSP you may also vpn and "X -query server:port"
13:03
<LTSP>
I haven't looked at NX (other than just googling and taking a peek) - is it fast?
13:04
I didn't know LTSP could work across WAN.... how to try out the "X -query server port" (or is that all I'll need to experiment?).
13:05
<fie_wr0k>
I was wondering if LTSP supports sessions like terminal services? Where you can close the RDP window and then re-open it from another computer?
13:05
And your xsession/apps are stored in ram or whatever.
13:06
<rjune>
irule: X -query isn't so great over a wan
13:06
fie_wr0k: LTSP does not as far as I know. You can however do it with VNC and/or NX
13:08
<ogra>
there are some solutions, no sane one though ...
13:08
xmove would be one
13:09
xvfb in combination with vnc would be another (you can "park" you vnc connection in the virtual framebuffer provided by xvfb)
13:10
<fie_wr0k>
hmmm
13:10
<ogra>
the proper solution would be to add such functionallity to Xorg though
13:12
<irule>
LTSP how avout using the TIGHT compression within vnc, 64 colors and a few tweaks? that is what I use with vnc, it works like a charm
13:13
<fie_wr0k>
ogra, so ltsp aims to be an open solution compared to vnc/nx?
13:13
<irule>
LTSP I use vnc within a vpn wan, I have basic adsl service on all sites
13:14
fie_wr0k no
13:14
<ogra>
fie_wr0k, yes
13:14
<fie_wr0k>
:)
13:14
<irule>
fie_wr0k there is no fttp nfs etc in vnc/nx
13:15
fie_wr0k plus, vnc is open as well, as well as the nx client
13:15
<ogra>
right, vnc is free, nx isnt (apart from the client to whetten your taste for the commercial version)
13:16Ryan52 has joined #ltsp
13:17
<ogra>
and sadly the way nx is implemented will never even allow the free client in any distros (it ships its own hacked up copy of xlibs which is a nono in most distros)
13:18
<dberkholz>
we have a nxserver-freenx package...
13:18
<ogra>
dberkholz, youre not shipping binaries on CDs :)
13:19spectra has joined #ltsp
13:19
<dberkholz>
oh btw, you might be interested in Xpra
13:19
<ogra>
there are ubuntu and debin packages and for sure there are rpms for varuious rpm based distros ... but none of these are included
13:19
<dberkholz>
it's an X server that supports disconnects & reconnects
13:19
<ogra>
yay
13:19
<dberkholz>
http://partiwm.org/wiki/xpra
13:19
<ogra>
a properly developed xmove ?
13:19|Ryan52 has quit IRC
13:19
<ogra>
xmove already does the right thing its just a horrible implementation
13:20
<dberkholz>
i haven't actually looked at it yet, just noticed it and bookmarked for later investigation
13:20
<ogra>
(which i wouldnt expect different from a proof of concept implementation for a thesis though)
13:20Ryan52 is now known as |Ryan52
13:26
<LTSP>
Xpra looks interesting
13:27
ogra, is xmove or xpra the one on a thesis project
13:30
<ogra>
xmove was
13:34
<warren>
xmove stopped in 1997
13:34
<LTSP>
Are xpra and/or xmove setup (or other similar projects) so the client locally runs the GUI and only minimal text/data is sent back and forth (I'm thinking how the MMORPGs seem to be executed) vs VNC that is trying to send graphic data back and forth and hence its slowness.
13:34Nubae has quit IRC
13:35
<warren>
reading about how it works
13:36
it sounds like VNC is the most bandwidth efficient
13:36Nubae has joined #ltsp
13:36
<warren>
keep in mind that you have bandwidth usage and latency
13:36
VNC uses far less bandwidth at the expense of higher latency
13:40plamengr has joined #ltsp
13:43
<LTSP>
So how is LTSP so much faster than VNC, across the LAN? LTSP trades bandwidth for latency then?
13:46
<rjune>
LTSP uses X
13:46
which works very well over a low latency, high bandwidth link
13:46
<warren>
VNC has nothing to do with LTSP
13:46
<rjune>
warren: there is some historical context there.
13:46
<warren>
well sure
13:47
<rjune>
anyway, VNC compresses it and works better for low bandwidth high latency link
13:47
<warren>
internet is generally too slow to do X
13:50
<LTSP>
Xpra/others will then be slow from the start (since they use X)?
13:51
<warren>
I've never used xpra
13:51
LTSP: this is generally not a solved problem
13:51
LTSP: if there were a better solution then everyone would be using it
13:55
<LTSP>
Thanks guys, I have to head to an offsite... you gave me some things to think about, some research to do, and maybe improved questions to follow up with. J
14:07vagrantc has joined #ltsp
14:10indradg has joined #ltsp
14:11mikkel has quit IRC
14:15plamengr has joined #ltsp
14:16plamengr has left #ltsp
14:50alekibango has joined #ltsp
14:56Pascal_1 has joined #ltsp
14:58cliebow has quit IRC
15:06indradg has quit IRC
15:09savetheWorld has quit IRC
15:12tata has joined #ltsp
15:13Pascal_1 has quit IRC
15:14tata is now known as Pascal_1
15:14Pascal_1 has left #ltsp
15:14irule has quit IRC
15:14fsgaston has quit IRC
15:16irule has joined #ltsp
15:16fsgaston has joined #ltsp
15:17fsgaston2 has joined #ltsp
15:17fsgaston has quit IRC
15:19indradg has joined #ltsp
15:23savetheWorld has joined #ltsp
15:26indradg_ has joined #ltsp
15:28
<warren>
horray
15:28
I have an ugly version of ldm gtkgreet embedded X keyboard layout setting
15:30
<vagrantc>
wacky.
15:30
you LTSP devs all turn to embedded devs eventually.
15:31indradg_ has quit IRC
15:31
<warren>
uh
15:31indradg_ has joined #ltsp
15:32
<rjune>
?
15:32* vagrantc will stop making off-the-cuff random comments
15:32
<rjune>
why? it just confuses the hell out of us
15:33
<vagrantc>
ah, yes. always forgetting my goals in the end.
15:33
good that i have rjune to turn to to keep faith.
15:35
<rjune>
lol
15:41
<warren>
sourceforge's list server is slow....
15:42
Attached is a preliminary implementation of X keyboard layout setting within ldm gtkgreet. The XKB* environment variables are already present in the greeter process, and it must be done here synchronously in order to guarantee that the keyboard layout is properly set before the user can type anything.
15:42
Any comments on this patch or suggestions of how it can be improved? Otherwise I would like to commit this in ldm-trunk soon.
15:42indradg__ has joined #ltsp
15:42
<warren>
http://fedorapeople.org/~wtogami/temp/ldm-x-keyboard-layout.patch
15:43indradg has quit IRC
15:45indradg_ has quit IRC
15:53
<ogra>
warren, i wonder if there is no glib function like scopy
15:54
beyond that it looks good to me
15:55spectra has quit IRC
16:03epoxy|w3rk has quit IRC
16:06
<fie_wr0k>
Does LTSP do anything with PXE?
16:08
Oh envermind, hmm.
16:14elisboa has quit IRC
16:27vagrantc has quit IRC
16:27ogra_cmpc has quit IRC
16:27cyberorg has quit IRC
16:27fie_wr0k has quit IRC
16:27viking-ice_ has quit IRC
16:27sutula has quit IRC
16:27iMacGyver has quit IRC
16:27vagrantc has joined #ltsp
16:27fie_wr0k has joined #ltsp
16:27ogra_cmpc has joined #ltsp
16:27cyberorg has joined #ltsp
16:27viking-ice_ has joined #ltsp
16:27iMacGyver has joined #ltsp
16:27sutula has joined #ltsp
16:37slidesinger has quit IRC
16:45
<Gadi>
well, now I know why ldm doesnt die...
16:45
*sigh*
16:47
<ogra>
take a gun
16:47
yore ameican you got the right to :P
16:48
<Gadi>
well, if an rc.d script kills the control socket, the x_session ssh call that ldm makes will hang waiting for a password
16:48
thats why he gets stuck at black screen + mouse
16:49
a simple check for the existence of the socket before running x_session would fix it
16:49
<ogra>
kill the local X
16:49
that should clean up everything and properly respawn ldm
16:49
<Gadi>
hmm...
16:50
I think that ssh call would still be executed, no?
16:50
<ogra>
if the parten goes away ?
16:50
*parent
16:50
that would be a bug imho
16:50
<Gadi>
well, its parent is ldm
16:50
not X
16:50
<ogra>
(not sure it does though)
16:50
well, X will tear down ldm
16:50
<Gadi>
X ans ssh are both children of ldm
16:50
<warren>
ogra: I rewrote the scopy parts to standard string.h functions but it became less simple looking, so reverted.
16:50
<Gadi>
I dont think so
16:50
not how it is written
16:51
<ogra>
warren, i'm not saying its wrong, imho just commit it ... i was just wondering if glib doesnt have any code you could just use directly like scopy
16:52
so the function wouldnt be needed and the rest of the code would stay the same
16:52
seems like a common task to make sure to have the string zeroed if there is nothing in the coyp surce
16:52
*copy
16:53
<warren>
I'm tagging this as ldm-2.0.7
16:54
<Gadi>
ogra: ldm is not wrapped in a loop, right?
16:54
<warren>
Gadi: from screen.d it is
16:54
<ogra>
not anymore
16:54
<Gadi>
neither it nor the screen script that calls it
16:54
<warren>
oh wait
16:54
<Gadi>
yeah
16:54
<warren>
I loop it, but maybe you don't.
16:54
<Gadi>
thats why we have no luck killing ldm
16:54|Ryan52 is now known as Ryan52
16:55
<Gadi>
so, I guess the upshot is - something needs to be patched
16:55
:)
16:55
which do you guys think is better - to kill the socket or to kill ldm?
16:55
<ogra>
Gadi, it is in a loop
16:55
while :; do
16:55
...
16:56
<Gadi>
ah - in hardy as well?
16:56
<ogra>
thats from a fresh hardy chroot
16:56* ogra just tested a lot of geode stuff for the 8.04.1 CD
16:58
<Gadi>
hmm.. so killing daddy and not grandpa should work
16:58
daddy should be ldm and grandpa should be the screen script
16:58
<ogra>
yeah
16:59
<Gadi>
kill -9 ${PPID}
16:59
easy enuff
16:59
I wonder if we ever tried that one
16:59
<ogra>
if that doesnt work, worst case lsof | grep $LDM_SOCKET and pull the pid out there to slay the ssh client
17:00
<Gadi>
killing ssh client wont help
17:00
we need to bring down ldm
17:00
so it doesnt proceed to spawn the next ssh client
17:00
<ogra>
i thought thats what ldm was waiting for in your scenario
17:00
<Gadi>
only after rc.d scripts finish
17:00
then it tries to do the x_session ssh
17:01
but without a control socket, it falls back on password
17:01* warren transit
17:04jammcq has quit IRC
17:22
<vagrantc>
warren: look forward to tryinng ldm 2.0.7 on debian :)
17:22
although, i don't really have any way to verify it ...
17:22* ogra looks forward to CONFIGURE_X=False by default :)
17:23
<vagrantc>
indeed.
17:23
<ogra>
its one major step
17:23
<vagrantc>
looks like debian unstable(maybe testing) uses a very skeletal xorg.conf
17:23
<ogra>
we do so since gutsy
17:23
keyboard and mouse sections only
17:24
and its planned that these go away in intrepid
17:24
in favour of hal input
17:24
<vagrantc>
i'm hoping to sit down and get a reasonable Xsession handling one of these days.
17:24
<ogra>
oh, well, then go ahaed :)
17:25* ogra was actually planning to do that on his way to boston ...
17:25
<ogra>
but if you go for it i wont stop you :)
17:25
<vagrantc>
heh.
17:25
i hope someone remembers what we had agreed to implement :)
17:26
<ogra>
roughly, yes
17:26
<vagrantc>
this is where mailing lists beat out irc conversations.
17:26
<ogra>
well, only in channels without logs :)
17:26
i have logging on in xchat
17:26
(i think)
17:26* ogra checks
17:28
<ogra>
booo
17:28
i dont
17:28
damned fresh HW with new installs ...
17:28* ogra fixes
17:29
<ogra>
one of these days i'll find the time ad set up a proper logbot for #ltsp, its annoying to not have searcheable logs like i do in all other channels i'm in
17:39japerry has quit IRC
17:43RiXtEr has joined #ltsp
17:46RiXtEr1 has joined #ltsp
17:51
<ogra>
vagrantc, ldminfo provides the location of Xsession, we have an rc.d script that kicks in and walks the different paths if we dont have the information from ldminfo ... tha actual session manager always gets attached to Xsession and by default we call the "default" session in all cases LDM_SESSION will only override the session manager call
17:52
and distro packagers should reorder the rc.d script to have their default path first so we limit the slowdown to connections to foreign distros
17:53
i think that was merely what we agreed on
17:59mccann has quit IRC
18:09RiXtEr1 has left #ltsp
18:10RiXtEr2 has joined #ltsp
18:11RiXtEr2 has left #ltsp
18:11RiXtEr has quit IRC
18:16DonSilver has quit IRC
18:26mhterres has quit IRC
18:32staffencasa has quit IRC
19:20makghosh_ has quit IRC
20:07ace_suares has joined #ltsp
20:15jammcq has joined #ltsp
20:15elisboa has joined #ltsp
20:19elisboa has quit IRC
20:41MacIver has quit IRC
20:41MacIver has joined #ltsp
20:48johnny2 has joined #ltsp
21:05Gadi has left #ltsp
21:08johnny2 has left #ltsp
21:08johnny2 has joined #ltsp
21:09elisboa has joined #ltsp
21:10vagrantc has quit IRC
21:10johnny2 has left #ltsp
21:25juanpaul has joined #ltsp
21:27epoxy|w3rk has joined #ltsp
21:32Nuba1 has joined #ltsp
21:33Nubae has quit IRC
21:37Faithful has quit IRC
21:38Faithful has joined #ltsp
21:46PMantis has joined #ltsp
21:47
<PMantis>
Sup dawgs
21:47
<rjune>
howdy
21:47
how goes it?
21:49
<PMantis>
Great... really busy
21:52Faithful has quit IRC
21:58vagrantc has joined #ltsp
21:59
<rjune>
good
22:04RiXtEr has joined #ltsp
22:04RiXtEr has left #ltsp
22:10juanpaul has quit IRC
22:15elisboa has quit IRC
22:19indradg_ has joined #ltsp
22:23mccann has joined #ltsp
22:23PMantis has quit IRC
22:26ace_suares has quit IRC
22:37indradg__ has quit IRC
22:37indradg_ is now known as indradg
23:37plamengr has joined #ltsp
23:51cassie has joined #ltsp
23:59Subhodip has joined #ltsp