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


Channel log from 10 October 2008   (all times are UTC)

00:05rcy has joined #ltsp
00:10petre has quit IRC
01:22mikkel has joined #ltsp
01:25Pascal_1 has joined #ltsp
01:31artista_frustrad has quit IRC
01:31Pascal_1 has left #ltsp
01:32artista_frustrad has joined #ltsp
02:25maalsa has joined #ltsp
02:28bobby_C has joined #ltsp
02:47maalsa has quit IRC
02:47maalsa has joined #ltsp
03:30
<nubae1>
moin folks...
03:33siki has joined #ltsp
03:38IsleVegan has joined #ltsp
04:04bobby_C has quit IRC
04:04bobby_C has joined #ltsp
04:08IsleVegan has quit IRC
04:17
<nubae1>
@images teaching
04:19maalsa has quit IRC
04:35johnny has left #ltsp
04:35johnny has joined #ltsp
04:35johnny is now known as Guest72565
04:43maalsa has joined #ltsp
04:46nubae has joined #ltsp
04:47nubae1 has quit IRC
05:10Guest72565 has left #ltsp
05:10johnny has joined #ltsp
05:11johnny is now known as Guest13149
05:11
<nubae>
hi Guest13149
05:11
:p
05:13
<ogra>
nubae, ssshhh ... he pretends to be not himself :)
05:27alkisg has joined #ltsp
05:37
<nubae>
ogra: I pasted a bug on launchpad concerning building xubuntu clients... the universe repo is required by default in 000-basic-configuration
05:37
or ltsp-build-client will die
05:37
<ogra>
nubae, thats fixed since ages
05:38
<nubae>
not in hardy
05:39
<ogra>
sure
05:39
<nubae>
(12:44:42 AM) staffencasa: nubae, I added universe to the 000-basic-configuration file and it worked
05:39
<ogra>
hardy != 8.04 :) we dont offer 8.04 isos anymore
05:40
hardy = 8.04.1
05:40
the .1 CD and iso are built from hardy-updates which has the fix
05:40
nobody should use 8.04 isos or CDs as they carry the ssh bug
05:41
if they d, they should at least immediately upgrade which wil get them the fix as well
05:41
s/d/do/
05:42* ogra points at https://bugs.launchpad.net/ubuntu/+source/ltsp/+bug/47278
05:42
<ogra>
oh, sorry, wrong one
05:43
but there is a bug for it
05:43
i just dont happen to find it right away
05:43
<nubae>
MIRROR=http://archive.ubuntu.com/ubuntu
05:43
COMPONENTS="main restricted"
05:43
thats from intrepid
05:43
<ogra>
https://bugs.launchpad.net/ubuntu/+source/ltsp/+bug/221894
05:44
<nubae>
thats my current intrepid install... no universe in there
05:46
<ogra>
hrm, right
05:46
stgraber, ^^^
05:46* ogra notices that this isnt fixed upstream ...
05:46
<nubae>
how did it get label fixed on launchpad :d ?
05:46* ogra does so
05:46
<ogra>
nubae, it was fixed in the package
05:47
stephane rolled a new upstream but seems that didnt get carried over ... my fault, i should have fixed it in the upstram brach as well
05:47
<nubae>
ah...
05:47
<ogra>
now we need to fix it in two places :/
05:50wima has joined #ltsp
05:59alkisg has left #ltsp
06:09
<nubae>
ogra: on a fat client, where should portmap and nfs be in the RC? I have to restart portmap through lts.conf
06:09
but thats not pretty
06:18pmatulis has joined #ltsp
06:24nubae has left #ltsp
06:31alkisg has joined #ltsp
06:34
<alkisg>
Hi, what's the difference between ltsp-server and ltsp-server-standalone packages? Only the dhcp-server? Because I've installed ltsp-server and in the clients X doesn't start...
06:34
(I can manually logon to the clients and start it though, and see the X cursor)
06:45nubae has joined #ltsp
06:46
<nubae>
gahhh... I just restarted may computer and network manager (even in unmanaged mode) took over all applications, meaning internet was there via ping, but no apps worked... I had to get rid of it entirely to make networking work again
06:47
<alkisg>
OK, next question... :) What would be faster, nbd swapping on the very fast server hard disk over gbit ethernet, or local swapping on very old 10Gb local hard disks?
06:47
<nubae>
I event tried deleting /etc/network/interfaces, to let NM take over
06:47
but stupid thing then froze
06:47
I feel sorry for those upgrading when intrepid comes out
06:48
<alkisg>
nubae, yes, I'm using the beta and I've @#$^*^#@ the NM a lot of times...
06:48
<nubae>
alkisg: pretty sure u can answer that question yourself ;-)
06:49
<alkisg>
nubae, it's not my own school, so I only go there for 4 hours a week.. no time for benchmarks! :)
06:49
<nubae>
nah, I mean from logic... the gigabit network swap would be faster
06:49
but dont know how that would affect bandwidth
06:49
<alkisg>
Even with all the network latency and with 8 clients on it?
06:49
<nubae>
if there is a lot of swapping... it might slow it down
06:50
<alkisg>
clients = 64Mb ram :(
06:50
<nubae>
8 clients is nothing
06:50
<stgraber>
ogra: hmm, right I didn't check for non-fixed fix released bugs :)
06:50
<alkisg>
nubae, I really wonder about that. Do you people really use 1 server with, say, 20-30 clients?
06:51
<nubae>
I use 1 server with 150 clients
06:51
<alkisg>
nubae, !!! You got fiber? :)
06:51
nubae, what is the typical client usage? Just openoffice?
06:51
<nubae>
granted, all multimedia stations are fat clients
06:51
<alkisg>
ah, ok!
06:51
<nubae>
and others are just internet, openoffice, some educational apps
06:52
but gigabit on all switches
06:52
gigabit to the multimedia stations
06:52
and 100mb to others...
06:52
but I seem to be one of the few people with little problems... along with Asmo, who also has a massive setup
06:53
strangely the complaints all come from people with smaller setups
06:54
<alkisg>
What about LDM_DIRECTX? True or false?
06:55
Because with 450MHz clients, I can't put it True, and it makes my local net suffer...
06:55
<nubae>
true
06:55
why cant u put it true?
06:56
<alkisg>
No, true is fine, I meant the default, which is false, so data is compressed and less bandwidth is needed
06:56
I can't put it *false* I meant
06:57wima has left #ltsp
06:57
<ogra>
stgraber, there is a bunch in the diff.gz i'm pushing upstream now, but we'll sadly have to carry them for intrepid, want a patch system ?
06:58
<nubae>
yeah I need it for the older machines too
06:58
the multimedia machines use the default though, not that it matters since everything runs locally
06:58mhterres has joined #ltsp
06:58
<alkisg>
nubae, so how many non-multimedia clients?
06:58
<stgraber>
ogra: while you are at it, can you also fix the "missing -updates and -security in chroot" one ?
06:59
<nubae>
hundred or so
06:59
<ogra>
thats not so easy, i already looked
06:59
<nubae>
I have some in between systems too for primary school
06:59
<ogra>
we'll need to modify the add mirror plugin
06:59
<alkisg>
nubae, I would have imagined that was viable!!! :) I shouldn't complain anymore with just 8 clients! ;)
07:00
s/would/wouldn't
07:00
<nubae>
well, u do need a hell of a lot of RAM and good dual or quad core xeon processors
07:00
and a fast drive with raid if possible
07:00
but not for 8 clients
07:00
<alkisg>
I have quad + 4 Gb ram (in 32 bit ubuntu), no raid, and CPU usage < 10% in normal usage
07:01
But in wine or flash, cpu gets 100% for even 4 clients...
07:01
<nubae>
yeah well, wine and flash are simply not meant to be run on the server
07:01
either local apps or fatclient
07:01
or they will bring down the network....
07:02
alkisg: did u see my plugin: http://nubae.com/ltsp-linux-terminal-server-project-netbooted-fat-client-for-ubuntu-hardy-and-intrepid
07:02
<stgraber>
ogra: hmm, right doesn't look that easy
07:02
<alkisg>
Yeah, I had to install flashblocker and tell the students to use flash only one at a time...
07:02
nubae, looking...
07:03
<nubae>
which reminds me, should the plugin go into Intrepid, ogra, stgraber?
07:03
or test some more and put into jaunty?
07:04
<ogra>
nubae, we can put it into the extra-plugins like the mythtv one
07:04
nubae, but that requires extra dependencies we dont have, right ?
07:04
which is a bit tricky
07:05Q-FUNK has joined #ltsp
07:05
<stgraber>
and dependencies must be in main
07:06
(that's the good thing with ltsp-cluster our outside-PPA dependencies are all in main already :))
07:06Egyptian[Home] has quit IRC
07:06
<nubae>
right, though think I add universe and multiverse in the script itself
07:07
<alkisg>
nubae, so the user accounts in the server are valid for the fat clients, just that when the admin adds new users, he has to update the chroot?
07:07
<nubae>
no
07:07maalsa has quit IRC
07:07
<nubae>
updating is done via a cron job every hour, and on restart of client
07:08
<alkisg>
Oh, that's fine! Wish my clients had 256Mb RAM... :(
07:08
<nubae>
but one can push the changes across with a script /etc/copyusers + restart client
07:08
ram is cheap... go buy some!
07:09
<alkisg>
SDRAM is >50 euros for each 256 MB, but I can't even find it!
07:10
<nubae>
2nd hand shop?
07:10
I know a place here in Vienna sells old dims for next to nothing.. like less than 10 euros
07:10
<alkisg>
I didn't find any that sells SDRAM... Cheap PCs with SDRAM, yes, but not just RAM
07:11
You're in Vienna? I was in Viener Neustadt in May!
07:11Egyptian[Home] has joined #ltsp
07:11
<alkisg>
Does this shop ship internationally? :)
07:13
<nubae>
no, but I have to be close to there early next week, so I can see if they still have the ram
07:14
tft screens for ~50 euro too... great for grabbing stuff for thin client setups
07:14
<alkisg>
Well if they do, I know many teachers here in greece that would be interested... So maybe something could be done.
07:15elisboa has joined #ltsp
07:16
<ogra>
stgraber, how about using EXTRA_DIST for updates
07:16
<alkisg>
Other question: why (in intrepid beta) doesn't update-manager *ever* close itself but just hangs in the "ldconfig deferred processing" phase?
07:17
<ogra>
alkisg, likely a bug, i saw that too yesterday, but havent talked to mvo in #ubuntu-devel about it yet (feel free to do so)
07:18
<alkisg>
ogra, I'll try, but I don't have the expertise to tell him exactly what's going wrong...
07:18
<ogra>
you just did :P
07:19
<alkisg>
heh
07:19
<ogra>
i didnt even know it was in ldconfig :)
07:36elisboa has quit IRC
07:36elisboa has joined #ltsp
07:39
<ogra>
stgraber, ok, we are missing the COMPCACHE by default change, the CONFIGURE_X being unset change (i cant push the latter upstream, compcache is now)
07:39
both need to get into the package somehow plus support for -updates
07:42
<nubae>
one of my teacher reports a very strange problem... everyday at 12 and again at 3 there is a massive spike in memory and cpu usage
07:42
<ogra>
cron ?
07:42
<nubae>
nothing jumps out
07:42
<ogra>
apt-get update run from update-manager probably
07:43
though that would only explain one of them
07:43plamengr has joined #ltsp
07:45elisboa has quit IRC
07:45elisboa has joined #ltsp
07:46
<nubae>
that ldconfig deffered process thing is new... as of latest update
07:47
<alkisg>
nubae, yes, I've just this morning installed a new server (intrepid beta), on the first update it didn't do it, but after this first update it does it...
07:48
<nubae>
oooh new gimp, neat....
07:48Q-FUNK has quit IRC
07:48
<nubae>
alkisg: did u report a bug, I can add + 1 to it :-)
07:49
<alkisg>
no, but here's one: https://bugs.launchpad.net/ubuntu/+source/update-manager/+bug/280236
07:51alkisg has left #ltsp
07:52mhterres has quit IRC
07:52mhterres has joined #ltsp
08:04wima has joined #ltsp
08:14bjohnson has joined #ltsp
08:21ATA_Dark_Shadow has joined #ltsp
08:26Gadi has joined #ltsp
08:38
<nubae>
@books ltsp
08:40
@gg ltsp
08:40ltsplogbot has quit IRC
08:41ltsplogbot has joined #ltsp
08:45
<nubae>
@g define:ltsp
08:50monteslu has quit IRC
08:50monteslu has joined #ltsp
08:50
<nubae>
@g define:ltsp
08:51
<ltsplogbot>
09:06bjohnson has quit IRC
09:15vagrantc has joined #ltsp
09:24
<stgraber>
ogra: how likely are we to get a second FFe for LTSP ?
09:24
ogra: that diff won't be huge and I don't like the idea of adding dpatch or similar thing unless it's necessary (as if the changes can't go upstream)
09:25
<ogra>
well, the CONFIGURE_X one cant
09:25
until debian switched to Xorg 1.5
09:25* ogra wonders if warren uses CONFIGURE_X=True by default
09:26
<warren>
ogra: I do? I don't even have a script that will do it.
09:26
<vagrantc>
CONFIGURE_X=false has been working reasonably well in lenny, even.
09:26
<ogra>
well, its set to true b default
09:26
by
09:26
<vagrantc>
though all the X_* stuff probably won't work.
09:26
<ogra>
the kayboard stuff will
09:27
as well as mouse settings for serial mice
09:27
both are not handled by xorg.conf
09:27
<vagrantc>
i mean in lenny without all the hal magic
09:27toscalix has joined #ltsp
09:27
<ogra>
still, warren added teh setxkbmap thing to ldm
09:28
so you wouldnt need hal anywy
09:28
<vagrantc>
ah, right.
09:28
<warren>
except the new X wont have any input without hal
09:28
stupid X
09:28
<ogra>
and mice are handled by inputattach
09:28
warren, but setxkbmap to change the keymap settings will still work
09:28
and vagrantc doesnt have the new X yet
09:28
<warren>
right
09:29
<ogra>
thats why i was worried to set CONFIGURE_X=false by default
09:29
<stgraber>
ogra: can't we check the X version and change CONFIGURE_X's default based on that ?
09:29
<vagrantc>
ogra: it's easy enough to enable
09:29
<ogra>
stgraber, if we can just drop it thats an easier change
09:30
<stgraber>
sure
09:33* vagrantc wonders why xprop doesn't have a -get corresponding to set
09:33Guest13149 is now known as johnny
09:33
<ogra>
isnt .get the default ?
09:34
<vagrantc>
how do i get the value of a particular property?
09:34johnny is now known as Guest30920
09:34
<vagrantc>
ah...
09:35
xprop -root PROPERTY
09:35
<ogra>
right
09:35
<vagrantc>
although it still outputs ... xprop -root LTSP_COMMAND
09:35
LTSP_COMMAND(STRING) = "bazbat"
09:36
be nice to just get bazbat ...
09:37
<ogra>
so, do i comment the CONFIGURE_X setting code or should i drop the whole chunk ?
09:37
<vagrantc>
-notype is almost there
09:37
don't see any reason to clutter it with comments
09:38
<ogra>
ok
09:38* ogra removes then
09:38* vagrantc experiments with running Xsession as a localapp
09:42
<ogra>
nubae, change 886 is probably worth to be mentioned in the docs
09:43Guest30920 is now known as johnny
09:43Egyptian[Home] has quit IRC
09:43johnny is now known as Guest35440
09:48
<nubae>
change886?
09:48
<vagrantc>
CONFIGURE_X != True by default
09:48
<ogra>
well, unset
09:48
<nubae>
ah 886 from ltsp bzr... ok
09:48
<vagrantc>
hrm. xprop still returns 0 if the property isn't found.
09:49
<nubae>
I'll change that... warren also wanted me to take out instructions for Fedora, as its still going through changes
09:49Guest35440 is now known as johnny
09:49
<warren>
nubae: leave it for the moment, I'll go ahead and change it myself later
09:49johnny is now known as Guest59984
09:49
<nubae>
ok
09:54alkisg has joined #ltsp
09:54Guest59984 has left #ltsp
09:57johnny has joined #ltsp
10:06wima has quit IRC
10:10warren has quit IRC
10:23Loto_ has joined #ltsp
10:24staffencasa has joined #ltsp
10:29Egyptian[Home] has joined #ltsp
10:31
<monteslu>
anyone have 64bit firefox with flash working on 32bit clients?
10:32
<johnny>
it possibly can
10:32
if you have nspluginwrapper
10:32
and if you use pulse.. you may or may not need libflashsupport.. depending on which version of flash
10:33
that is.. for sound
10:33
my clients have no speaker
10:34
<monteslu>
I have nspluginwrapper, and made sure that libflasuppoprt is not installed
10:34
and pulse
10:34
using fedora 9
10:34
<johnny>
it depends on the flash version
10:34
<monteslu>
ok, I'll give the 10 beta a shot
10:34
9 doesnt work
10:34
<ogra>
falsh 9 needs libflashuppoprt
10:35
<monteslu>
oh boy
10:35
<ogra>
lubflustchappurt ... :)
10:35
<monteslu>
10 doesnt work
10:35
<ogra>
labflashyport
10:35
<monteslu>
i'll go back to 9 with libflashsupport
10:36
<ogra>
you wont really notice the crashers anyway if you use nspluginwrapper
10:36siki has quit IRC
10:36
<ogra>
at least in ubuntu it works fine
10:37
thats the only combo where it works fine in ubuntu actually :)
10:37
<monteslu>
that's one more combo than I currently have working
10:38
man, still no go
10:38
<ogra>
in ubuntu 32bit we dont have nspluginwrapper ... so the libflashsupport bugs make firefox crash
10:38
<nubae>
gnash worked ok
10:38
on 64 bit
10:38
<ogra>
without libflashsupport on pulse you dont have sound with flash9
10:39
<nubae>
but its the reason I left 64bit... was way more trouble than its worth
10:39
<ogra>
yeah
10:39
it definately isnt worth the trouble
10:39
<monteslu>
warren told me to do it
10:39
:)
10:39
<nubae>
hehe blame someone els :p
10:39
<johnny>
flash 10 works here
10:40
<monteslu>
johnny, 64 server?
10:40
64 bit server rather
10:41
RAM manufucturers need to start leaning on Adobe. They're holding the world back
10:41twinprism has quit IRC
10:42
<johnny>
they don't care until more people use 64bit windows
10:43plamengr has quit IRC
10:43
<ogra>
well, there are 64bit atoms now
10:43
so i guess MS will react at some point
10:43
<johnny>
sure.. but i bet they run 32bit xp 99.9% of the time :)
10:44
or will
10:45
<nubae>
what would help is if hollywood started boycotting their products... the entire industry is linux based... and uses adobe products through wine heavily
10:46
<johnny>
thatwon't be a problem tho ?
10:46
afaik
10:46
<nubae>
they should just say, support linux or we move to gimp
10:46
<johnny>
lol.. not until the gimp is easier to use
10:46
is gimpshop still around.. i wonder
10:46
<nubae>
have u see latest one?
10:46
it is...
10:46
but now gimp is almost gimpshop
10:46
<johnny>
i installed 2.6 just recently
10:46
<nubae>
extremely similar, 2.6 is an amazing change in the right direction
10:46
<johnny>
haven't opened it since
10:46
<nubae>
really?
10:46
u werent impressed?
10:47
<johnny>
ogra, gimp 2.6 in intrepid?
10:47
:(
10:47
<nubae>
yes
10:47* johnny bets not
10:47
<johnny>
oh
10:47
cool
10:47
<ogra>
sure :)
10:47
<nubae>
only as of a couple hours ago though
10:47
hehe
10:47
<johnny>
i just said i haven't opened it since, haven't had any images to edit
10:47
can it do raw files natively yet?
10:47
<nubae>
ah... ok, well it works great
10:48
<johnny>
and when it does.. will we need ufraw and the like anymore? :)
10:48
<nubae>
well, thats what cinepaint is for
10:48
<johnny>
huh? i thought decent cameras nowadays did raw files?
10:48
not that i am a digital photo expert of ANY kind .,.
10:49
<nubae>
cinepaint is for the professional photographer
10:49
<johnny>
sure.. what what is for the amateur ? :)
10:49
with a camera does raw files
10:49
<nubae>
as far as I can tell most cameras have the dcm
10:50
format
10:50
kodak, canon, sony all do that format
10:51
anyway, just been working with gimp and the new layout is lovely
10:53
<johnny>
my users will be happy
10:53
i saw some lady trying to use the gimp.. and it was acting funny for her
10:54
it was reversing actions for no reason.. and the undo was not applying to every step
10:54
it was weird
10:54* johnny awaits intrepid heavily
10:56
<nubae>
do u maintain the ubuntus at work?
10:57
<johnny>
it's an ltsp setup
10:57
altho users use the server as the POS server
10:57
err workers
10:58
yes.. i maintain the ubuntu at the shop
10:58
i'm one of the worker owners
10:58
i just happen to be one of the techies
11:02
<nubae>
heh
11:03TheGuy has joined #ltsp
11:03
<TheGuy>
anyone have any idea why I try to login as a certain user and I get this error connecting to D-BUS system bus: Did not receive a reply. Possible causes include: the remote application did not send a reply, the message bus security policy blocked the reply, the reply timeout expired, or the network connection was broken.
11:03
<nubae>
groups permissions?
11:04
<TheGuy>
it works for everyone else in the same group
11:04
47 other users
11:04
<nubae>
same groups u mean, right?
11:04
<TheGuy>
yeah
11:04
also when I su to this user and try to launch dbus manually I get: Failed to start message bus: Memory allocation failure in message bus
11:05
but if I su to another user and run dbus manually it runs
11:05
<nubae>
and u googled that?
11:05
<TheGuy>
yeah
11:05
nothing helped really
11:05
<nubae>
@g Failed to start message bus: Memory allocation failure
11:05
<ltsplogbot>
Apple - Support - Discussions - Terminal @ http://discussions.apple.com/thread.jspa?threadID=1739602&tstart=1 | Nabble - fink-devel - Failed to start me @ http://www.nabble.com/Failed-to-start-message-bus:-Memory-allocation-failure-in-message-bus-td19393426.html | Error running DBUS on ARM @ http://lists.freedesktop.org/archives/dbus/2007-March/007189.html | [Fink-devel] Failed to start message bus @
11:05
http://www.mail-archive.com/fink-devel@lists.sourceforge.net/msg17825.html
11:06
<nubae>
that second link seems pertinent
11:07Gadi has left #ltsp
11:08
<TheGuy>
any idea why gnome-keyring-daemon is kicking off or caring about D-BUS?
11:09Gadi has joined #ltsp
11:09
<nubae>
not really... perhaps #ubuntu might have more help, since this isnt specifically about ltsp
11:09
<TheGuy>
it's opensuse i'll try there
11:09
thanks
11:09
<nubae>
but I'm just guessing
11:09
hehe
11:10
yeah thats a whole different kettle of fish
11:10twinprism has joined #ltsp
11:11
<nubae>
sbalneav: u there?
11:11TheGuy has left #ltsp
11:14
<nubae>
ogra: the X_CONFIGURE is not in the manual right now
11:15
<ogra>
right, bit to change things like X_MODE etc you need it now
11:16
<nubae>
can u give me a little more explanation on it, I'll add it now
11:18japerry has quit IRC
11:19
<ogra>
X_CONFIGURE creates a xorg.conf
11:19japerry has joined #ltsp
11:20
<Gadi>
er, CONFIGURE_X
11:20
<ogra>
but Xorg 1.5 doesnt use xorg.conf anymore ... so by default we leave that out
11:20
right, Gadi is right
11:21
<Gadi>
ogra: will Xorg 1.5 still produce one with X -configure?
11:21
<nubae>
ok, so I put it just above the X_MODE and X_CONF and all that, explaining that it is required for users that want to set individual options for xorg.conf
11:21
<ogra>
Gadi, yep
11:23
<Gadi>
ogra: is compcache only in the newer ubuntu kernels?
11:24
<alkisg>
So we need a CONFIGURE_X=True in order for X_MODE_0=1024x768, X_VERTREFRESH=50.0-60.0 to work? Does this apply to intrepid beta?
11:24
<nubae>
If you want to be able to configure the individual settings of the X configuration file, without
11:24
having the X automatically configure the graphics card for you, you must enable this option. By
11:24
default this option is turned off. To turn it on do: CONFIGURE_X = True
11:24
<alkisg>
nubae, thanks
11:24
<ogra>
alkisg, yes, but you should first try withour it
11:24
<nubae>
ogra: does that sound right?
11:25
<ogra>
as Xorg is gotten a lot more intelligent
11:25
nubae, well, you dont need CONFIGURE_X for: mouse or keyboard settings
11:25
all others need it
11:25
<nubae>
ok, I'll add that disclaimer, though it will just confuse people :-)
11:26* vagrantc uses X_MODE_0 to get consistant size on classrooms with disparate hardware capabilities
11:26
<alkisg>
Well, I had problems with a ltsp lab I setup this morning, and I wonder if this was the problem... X wasn't starting on the clients, but if I ran it manually then I was able to see the "cross" cursor.
11:26
<vagrantc>
choosing the best isn't always what you even want
11:26
<nubae>
You don't need this option just for keyboard and mouse settings. It
11:26
corresponds to the graphic card options only.
11:26* alkisg does that also! :)
11:27
<ogra>
vagrantc, yes, but first people should always try without xorg.conf now
11:27
<nubae>
ogra: ?
11:27
<ogra>
nubae, and monirtor options ...
11:27
<nubae>
ok
11:27
<vagrantc>
ogra: what about turning on CONFIGURE_X if the necessary settings are configured?
11:28
<ogra>
vagrantc, hmm
11:28
<nubae>
sounds like work
11:28
<ogra>
thats not the point
11:28
i would really like to get rid of parsing the xorg.conf
11:28
<vagrantc>
as more of those settings become possible to configure without configuring X directly
11:29
then it just starts to work properly...
11:29* Gadi likes that idea
11:29
<ogra>
thats a lot of options though
11:29
<vagrantc>
preserves backwards compatibility, too.
11:30
it's a fair number, yes.
11:30
<Gadi>
also, need to know what version of Xorg is used
11:30
<ogra>
and i'd like to have people *first place* using the wutoconfiguration
11:30
and only then start tweaking
11:30
<vagrantc>
agreed
11:30
<Gadi>
is this not distro-specificc?
11:30
<vagrantc>
X version dependent ... not sure about distro
11:31
<Gadi>
didn't ogra have to do a lot of fiddling with hal ?
11:31
<ogra>
nope
11:31
<Gadi>
and aren't hose tweaks only in ubuntu?
11:31
oh
11:31
<ogra>
our X etam had :)
11:31
and debian has all our changes in experimental already
11:31
<Gadi>
right - but will all distros on X 1.5 use hal?
11:31
<ogra>
yes
11:31
<johnny>
we will
11:31
<Gadi>
and to the same extent?
11:31
<ogra>
yes
11:31
<johnny>
yes.. they have to :)
11:31
<ogra>
upstream requires hal
11:31
<Gadi>
oh ok
11:32
<ogra>
the hal sided implementation might be distro dependent
11:32
<Gadi>
could we set CONFIGURE_X=False if Xorg >=1.5?
11:32
<ogra>
i.e. debian and ubuntu use /etc/default/console-setup as the base for keymap settings
11:32
other distros dont have that
11:33
<johnny>
doesn't fedora and suse have console-setup ?
11:33
<ogra>
so they will grab the values from elsewhere
11:33
<johnny>
we don't
11:33
but that's ok
11:33
<vagrantc>
i'm not sure we're actually making use of console-setup ...
11:33
on debian
11:33
but it's installed :)
11:33
<ogra>
i think console-setup is a debian thing
11:33
or even ubuntu and backported to debian
11:33
<johnny>
Gadi, i already set CONFIGURE_X=F due to configure-x.sh not working very well for me
11:33
<vagrantc>
console-tools, kbd, console-setup ... aieee!
11:33
lets not forget console-data
11:33
<ogra>
kbd is dead
11:34
err, no
11:34
xkb is dead
11:34
:)
11:34
console-data and console-tools are dead in ubuntu iirc
11:34
and debian might do that switch as soon as lenny is released
11:34
<johnny>
so.. when will the silly nvidia people use xrandr :(
11:34
that's what we really need now :(
11:35
<Gadi>
brb
11:36Gadi has left #ltsp
11:37
<vagrantc>
gah.
11:37
well, i've got my crazy "run-entire-session-as-localapp" stuff nearly working ...
11:38
just need to figure out a way to communicate when the session is ended ...
11:38
<johnny>
drop kick it
11:39
<vagrantc>
ltsp-localappsd backgrounds it, and then unsets the xprop for LTSP_COMMAND ... i need a way to conditionally tell it "don't unset it until LTSP_COMMAND is finished ..."
11:40Gadi has joined #ltsp
11:40
<alkisg>
Guys, what's the difference between ltsp-server and ltsp-server-standalone packages? Only the dhcp-server? Because I've installed ltsp-server in a new lab this morning, and in the clients X doesn't start. I can manually logon to the clients and start it though, and see the X cursor...
11:40
<sbalneav>
vagrantc: You could do a "wait $PID && xprop..."
11:40
Morning all, BTW
11:41
<vagrantc>
sbalneav: but it needs to be conditional ... normally, we want it backgrounded ...
11:41
<ogra>
alkisg, which package version ?
11:41
<nubae>
so ogra XFS is not supported in any distro anymore right?
11:41
<ogra>
right
11:41
<vagrantc>
how to distinguish between a "normal" localapp, and a "whole session" localapp
11:41
<sbalneav>
vagrantc: You gonna be around all this afternoon?
11:41
<alkisg>
intrepid beta, original && also full updated (this morning)
11:41
<vagrantc>
sbalneav: possibly, probably...
11:41
<nubae>
sbalneav: Ive branched from your branch and will merge into trunks
11:41
<sbalneav>
Ah, ok
11:41
<nubae>
editing through vi of course =)
11:42
<ogra>
alkisg, set CONFIGURE_X=False in lts.conf and try again ;)
11:42
<vagrantc>
sbalneav: busy at work, i'm guessing?
11:42
<alkisg>
ogra, I'll try on Monday (no access on weekends!), thanks! :)
11:43
<vagrantc>
i've thought about using another x property ... LTSP_COMMAND_WAIT or something ...
11:43
<nubae>
now that I'm editing the lts.conf variables, are there any new /outdated ones?
11:44
<Gadi>
nubae: I don't think I ever added: LDM_USER_ALLOW
11:44
(which is new)
11:44
<nubae>
k
11:45
<alkisg>
nubae, is what you're writing online? Or you'll upload it all when it's finished?
11:45
<nubae>
upload when I'm done... to ltsp-doc-trunks
11:46
<ogra>
Gadi, do you have any expecience with L1730Sf touchscreens ?
11:46
<nubae>
but there is a very similar version here:
11:46
!docs
11:46
<ltspbot>
nubae: "docs" is For the most current documentation, see http://wiki.ltsp.org/twiki/bin/view/Ltsp/LtspDocumentationUpstream
11:46
<alkisg>
nubae, thanks
11:46
<Gadi>
ogra: is that ELO?
11:46
<ogra>
evtouch
11:46
<Gadi>
no
11:47
i have only played with elo
11:47
<nubae>
ldm_allow_user does what exactly?
11:48
<Gadi>
LDM_USER_ALLOW allows the admin to limit access to a select group of usernames
11:48
LIKE: LDM_USER_ALLOW = alex,bob,carl,dave
11:48
<nubae>
ok, so for a mixed server environment
11:48
<Gadi>
no
11:48
for any environment
11:49
where there are some workstations that only certain users should use
11:49Egyptian[Home] has quit IRC
11:49
<nubae>
OK, gotcha...
11:49
<Gadi>
so, I could have a certain workstation identified by MAC in lts.conf with LDM_USER_ALLOW = alex so only alex can log in there
11:50
<nubae>
right
11:51
<Gadi>
multiple users are comma-separated with no spaces
11:51
as above
11:51
the absence of the variable means all users
11:52
setting to a non-existent user is effectively denying access to all
11:53
<nubae>
heh
11:53
<ogra>
and setting to "Gadi" gives everyone in ltsp-upstream root on your server
11:53* ogra hides
11:53
<Gadi>
hehe
11:53* Gadi wished he thought of that
11:54
<nubae>
that would be funny, a backdoor in ltsp
11:54
first get everyone to use ltsp, then sabotage them all
11:55
<ogra>
well, we should at least thig abotu some nice easteregg
11:55Egyptian[Home] has joined #ltsp
11:55
<ogra>
an option that flips all client screens upside down or so :)
11:56
and scrolls the names of all contributors in front of it
11:56
(not upside down indeed)
11:56
<nubae>
that would be fun indeed
11:57
how about on ldm login, if u dont put anything in and hit enter
11:57
:-) then people will be sure to see it
12:02
<johnny>
as long as it doesn't break LDM_ALLOW_GUEST or whatever it is called now
12:02
<ogra>
well, it would be a special option
12:02
and i think it should actually prevent you from doing stuff untl you unset it again
12:02
to make sure the toying doesnt interfere with actual work
12:03
EGGS_4_FREE=True in the default section :)
12:03
<nubae>
u mean like EASTER_EGG = True
12:03
now who would think of putting that in?
12:03
:p
12:04
<ogra>
nobody
12:04
did you ever try out the gnome eastereggs ?
12:04
<nubae>
isnt someone supposed to discover an easteregg
12:04
no... the openoffice ones though I did
12:05
<ogra>
alt+f2 ... enter "free the fish" (no quotes here)
12:05
or
12:05
<nubae>
heh, oh yeah I have seen that one
12:06
<ogra>
alt+f2 ... enter "gegls from outer space" (no quotes here)
12:06
<nubae>
lol, I like that one
12:06
<ogra>
as the name says, eastereggs are hidden :)
12:07
<nubae>
yeah but, unfindable?
12:07
hmmm there is no LDM_AUTO_LOGIN in the docs either
12:08* vagrantc can't wait to get bug reports about broken easter eggs
12:08
<ogra>
hehe
12:08
<nubae>
make it complex enough and the egg will break :-)
12:08
especially on Suse
12:09
<vagrantc>
nubae: LDM_AUTOLOGIN (boolean), LDM_GUESTLOGIN (boolean), LDM_TIMEOUT_ENABLE (boolean), LDM_TIMEOUT_TIME integer
12:09
<nubae>
oof.. ok, non of those are in there...
12:09
<vagrantc>
i figured not
12:09
<nubae>
I remember there was a bug with one of those, they all work?
12:09
<vagrantc>
last i tried, they worked
12:10
<ogra>
xrandr -y && sleep 5 && xrandr -o normal
12:10
<vagrantc>
nubae: LDM_USERNAME (string), LDM_PASSWORD (string)
12:10
<ogra>
trivial :)
12:11
<vagrantc>
maybe be should rename LDM_TIMEOUT_ENABLE to LDM_TIMEOUT ...
12:11
<stgraber>
ogra: so, what's the plan ? upload one more new ltsp to Intrepid ?
12:11Eghie has joined #ltsp
12:11
<vagrantc>
and LDM_TIMEOUT_TIME to LDM_TIMEOUT_SECONDS
12:11
<ogra>
stgraber, up to you ... i will support your decision
12:12
<nubae>
I'll put them as they are now then, right vagrantc or?
12:12
<ogra>
(i.e. help fighting with slangasek if you want a FFe)
12:13
<vagrantc>
nubae: put them as they actually are ... i'll put in backwards compatibility if we change them.
12:13
<nubae>
ok
12:14
<vagrantc>
ok, i've got a reasonable patch to run whole sessions as localapps... avoids all the need for pesky network auth :)
12:14
<nubae>
XDM_SERVER... that still used?
12:14
<vagrantc>
if you're using xdmcp
12:14
<nubae>
ah right
12:16F-GT has quit IRC
12:16
<stgraber>
ogra: let's make sure all the bugs we want to close in Intrepid are fixed upstream, then I'm fine doing one more package
12:17
<ogra>
well, i pushed all changes that were missing from my last package i think
12:17warren has joined #ltsp
12:18* nubae has a feeling intrepid will be the most stable LTSP Ubuntu yet
12:20
<nubae>
will LDM_AUTOLOGIN work without defining a user and pass?
12:20
or does it requrie LDM_USERNAME and LDM_PASSWORD?
12:20
<stgraber>
no
12:20
it requires at least LDM_USERNAME
12:20
<ogra>
how should it ? :)
12:20
<stgraber>
you can have an empty password though (for example if you use SSH key authentication)
12:20
<nubae>
well, guestlogin kinda implies that
12:20
<stgraber>
(I do that for ltsp-cluster autologin)
12:22elisboa has quit IRC
12:22elisboa has joined #ltsp
12:22
<nubae>
for a normal user though without getting into the intricacies of ssh keys, they need password too
12:22
right=
12:22
?
12:22F-GT has joined #ltsp
12:23
<nubae>
and the user and pass are the machine name, right?
12:23
or was that for guestlogin?
12:24
<ogra>
wow, #ubuntu-meeting is fascinating (at least if yoour client has proper utf8 support)
12:24
<johnny>
i tried that once..
12:24
my user acounts got hacked
12:24
somebody ran an irc bouncer
12:24
had to switch to ssh keys
12:24
i wrote a script to handle ~/.ssh
12:24
when i cleared out all the profiles
12:25
<nubae>
well cant u lock down the machine accounts?
12:26
without doing ssh keys?
12:26
<vagrantc>
nubae: actually, LDM_USERNAME and LDM_PASSWORD default to the hostname of the thin client, so it doesn't require it
12:26
<stgraber>
ogra: yeah, saw that too :) I was first wondering if my terminal was having some issues, then noticed it was just an ubuntu arabic meeting :)
12:26
<vagrantc>
but you can also set up ssh keys
12:26
<nubae>
ok so autologin does work without user and pass
12:28
<vagrantc>
nubae: the only difference between AUTOLOGIN and GUESTLOGIN is that guest login displays the login screen, but includes a button to "kick in" autologin ... or logs in as whatever user is specified if the button isn't pressed
12:29ATA_Dark_Shadow has quit IRC
12:31
<stgraber>
ogra: I'll just go through all the Ubuntu ltsp bugs checking for bugs with workaround/patch that should be fixed in Intrepid, if I don't see any I'll just package the current bzr
12:31
<ogra>
stgraber, ok, poke me if i can help or dump anything into upstream
12:32* vagrantc pushes vagrantc's crazy localapp login sessions
12:33
<stgraber>
I also have a small patch for ltsp-cluster that I'd need other ltsp guys approval before I push it
12:36
ogra, vagrantc, warren: Opinions ? http://paste.ubuntu.com/56051/
12:37
<ogra>
stgraber, where is the source for getltscfg-cluster ?
12:37
<stgraber>
that's a bit ltsp-cluster specific but it's the only change we do on upstream LTSP that we haven't merged yet
12:37
ogra: ltsp-cluster-client in ltsp-cluster (LP)
12:37
<laga_>
what is ltsp-clusterß
12:37
<stgraber>
it's an ugly shell script that performs queries to our control center component
12:37
<ogra>
thats fine then, i'm not happy by calling the file /etc/ltsp/directory.conf though
12:38
thats very generic
12:38
<warren>
stgraber: well, I haven't seen the entire architecture of ltsp-cluster yet
12:38
<ogra>
warren, well, it will fire only if /etc/ltsp/directory.conf xists anyway
12:38
<johnny>
directory.conf ?
12:38
<stgraber>
hmm, you're right the name sucks
12:38
<ogra>
so i dont really care about the binary for upstream
12:38
<stgraber>
especially as the component was removed :)
12:38
<ogra>
but the naming seems dangerous
12:39
someone might want to use that file for a distro specific slapd config with such a name
12:39
<warren>
ogra: yeah, so on the surface I have nothing against it
12:39
<johnny>
so.. getltscfg-cluster needs it's own binary ?
12:39
<ogra>
i would make it more custer specific
12:39
<stgraber>
what about /etc/ltsp/cluster-control.conf ?
12:39
<ogra>
that sounds sane
12:40
<vagrantc>
stgraber: why not "if [ -x /path/to/getltscfg-cluster ]; then ..." ?
12:40
<stgraber>
ok, the server component is called ltsp-cluster-control so that name sounds better (it used to be called ltsp-directory)
12:40
<ogra>
vagrantc, ++
12:40
thats even better
12:40
<stgraber>
vagrantc: right, will do that
12:42sepski has joined #ltsp
12:42siki has joined #ltsp
12:42* vagrantc makes a commit to the command ogra doesn't even want to exist
12:43
<stgraber>
vagrantc: oh, you commiting ? I was about to commit the change, I'll just wait to avoid a LP conflict then :)
12:43
<vagrantc>
stgraber: done.
12:44
<ogra>
vagrantc, i want it to exist ... but on the client :)
12:45
i'm just complainint about the implementation details, not about the implementation itself :)
12:46nubae1 has joined #ltsp
12:46toscalix has quit IRC
12:46
<stgraber>
EVERYONE: Any issue with me tagging a new LTSP now ?
12:47
<ogra>
not me
12:47Lns has joined #ltsp
12:47nubae has quit IRC
12:47nubae1 is now known as nubae
12:52
<stgraber>
ogra: do we have a bug for CONFIGURE_X!=False ?
12:52
<ogra>
no, that was a change i made before FF
12:52
<stgraber>
ok, so regression from before FF
12:52
<ogra>
a regressing from FF
12:53
i uploaded that package at FF day
12:53
<vagrantc>
stgraber: no problem with tagging. though when you make your tarball, please do it from a clean export or something ... saw some editing cruft in the last tarball
12:53
<stgraber>
ogra: for the Xubuntu thing, do we have a bug report ?
12:53
<vagrantc>
ldm, anyways
12:53
<stgraber>
vagrantc: I used the mkdst thing
12:53
<vagrantc>
stgraber: --from-tag ?
12:55
<ogra>
stgraber, for hardy, yes, it was fixed with the first intrepid upload in the package
12:55
staffencasa, one min
12:56
stgraber, ^^^
12:56
i'm just checking if we unlock the root account so it doesnt come up expired, i want that fixed as well
12:56
<stgraber>
ok, I just had a look at LP and couldn't find the bug but I remember seeing it in Hardy
12:56
ogra: yeah, got that bugmail too :)
12:57
<ogra>
i dont care aboru ask-rootpass, bt i know i had fixed it in hardy
12:57
<alkisg>
ogra, I tried this morning (intrepid beta), it was unlocked (and --ask-rootpass worked)
12:57
<ogra>
alkisg, well, beta had my package, stgraber missed some of my changes i didnt have committed upstream
12:57
<stgraber>
ok, I also use it here and it works correctly (last I tried)
12:58
<ogra>
ok
12:59
i thought i saw it as expired last time i checked
12:59
<nubae>
sbalneav: I just commited to the main branch, and added in my revision that this was a merge from your branch where many changes had been made
13:00
hopefully thats ok, beause it wont show the revisions u did in the recent changes text
13:00
anyway it should be current now, maybe have a check that you can branch it
13:03
<laga_>
ogra: so, configure-x.sh is not called anymore in intrepid?
13:03siki has quit IRC
13:04
<stgraber>
laga_: except if you force CONFIGURE_X=True
13:04
<ogra>
laga_, not by default
13:04
<laga_>
okay, i will have to adjust mythbuntu-diskless for that
13:05
<ogra>
just export it :)
13:05
its just an odd hackish script
13:05
and i want it to die asap
13:05
<laga_>
export what?
13:05* ogra hopes corg in jaunty will support driver settings through hal
13:06
<ogra>
*xorg
13:06
laga_, CONFIGURE_X=True
13:06
<laga_>
ogra: ah, yes, but i think the path has changed, didnt it?
13:06
<ogra>
somewhere in your initscript
13:06
the path ? for the variable ?
13:07
al i dropped was the default setting of CONFIGURE_X=True, its unset now
13:07
nothing else
13:08
<laga_>
okay
13:10
<stgraber>
ogra: compcache was enabled pre-FF too ?
13:12
<ogra>
yes
13:12
there were three ubuntu specific changes i didnt have upstream
13:12
<stgraber>
ok
13:12* ogra gets pizza
13:12
<alkisg>
Where does nbd_swap actually swap? Server RAM or HD?
13:13
<ogra>
hd
13:13
<alkisg>
eeek...
13:13
<ogra>
you can bind mount a tmpfs to where the swapfiles are created
13:14
<alkisg>
ogra, what do I look for? (like man nbd)
13:14
<ogra>
but be aware that this can indeed eat tons of server ram
13:14
<alkisg>
I can't find documentation for nbd anywhere...
13:14
<ogra>
look at nbdswapd
13:14
<alkisg>
ogra, no problem, I got 3 gb on the server and only 64mb on the clients
13:14
ogra, thanks, bon appetite!
13:14
<stgraber>
ogra: Does that sound scary enough to get the FFe: http://paste.ubuntu.com/56067/ ?
13:14
<ogra>
thabks :)
13:15
stgraber, add (committed upstream) to the regression lines
13:18
<stgraber>
ogra: http://paste.ubuntu.com/56068/
13:18
ogra: I had to change a bit the wording so the lines aren't split
13:19
<ogra>
yef, pfounds bepper
13:20
(sorry, mouth full)
13:21
<sbalneav>
Last night was a bad night. My dad had a low blood sugar episode, and I didn't get home 'till 12:30, and of course I was too bloody wired to get to sleep right away.
13:21
<ogra>
oh man
13:22
<Lns>
sbalneav: i'm sorry :(
13:22
<sbalneav>
Ah, everything's AOK now.
13:22
So, I've fixed the segfault with cdpinger
13:23
<Lns>
alkisg: how many clients do you have? Keep in mind if you have 256mb swap per client...that can add up quickly.
13:23
one of my client's HDD filled up right after I enabled nbd_swap :p
13:23
they had 35 clients
13:23
<alkisg>
Lns, 8 clients, but right now performance suffers a lot
13:24
My server is new, quad core, 2x640Gb disks, 4 Gb ram (=3 in 32bit os)
13:24
<Lns>
alkisg: that's gonna take 2/3GB out just for client swap....
13:24
alkisg: why not install -server kernel to get all 4GB?
13:24
<ogra>
alkisg, if you buit it just now, you will miss the compcache love
13:24
try agan with the next upload
13:24
<alkisg>
Can't I somehow use caching for that? I mean, if ram is enough, use it, else swap on normal swap file...
13:25
<Lns>
alkisg: well you're swapping over the NETWORK anyway..i would think that's your bottleneck
13:25
<laga_>
ogra: is the current ltsp package in ubuntu kep tin VCS? or do i just send a patch?
13:25
<Lns>
i might be wrong, but that's what i would think
13:26
<alkisg>
Lns, the clients also have some ancient hard disks, (10Gb), do you think I should use local swap?
13:26
<ogra>
laga_, stgraber does it now
13:26
<Lns>
alkisg: if you wanna partition the disks, sure.. but i think ogra is working on fixing the nbd_swap bug we found the other day, might not be worth the trouble
13:26
<ogra>
for the mythbuntu plugin a patch is ok i guess
13:26
<vagrantc>
alkisg: local hard disk will swap much more cleanly.
13:26
<laga_>
stgraber: ah, i'll redirect my question: is the current ltsp package in ubuntu kept in VCS? or do i just send a patch?
13:26
<alkisg>
vagrantc, thanks, I'll do that
13:26
<vagrantc>
alkisg: and fast.
13:27
<ogra>
alkisg, really, wait until compcache is enabled agai, it will raise your clent ram by default without external needs
13:27* ogra goes afk now
13:27
<alkisg>
ogra, OK, I'll try both! :P
13:28monteslu has quit IRC
13:28
<stgraber>
laga_: not in VCS yet, but I guess I'll upload it to LP soon
13:29
laga_: what change would you like ? I'm preparing an upload for Intrepid
13:31
<Lns>
What's compcache?
13:31
<laga_>
stgraber: a small fix for the mythbuntu plugin. when are you going to upload?
13:31
<Lns>
ah, nm http://code.google.com/p/compcache/
13:32
<stgraber>
laga_: now
13:32
<laga_>
stgraber: ah, hold on a minute please :)
13:33* vagrantc wonders when compcache will hit debian
13:33
<vagrantc>
is compcache just integrated directly into the kernel on ubuntu?
13:34
<stgraber>
stgraber@castiana:~$ find /lib/modules/2.6.27-6-generic/ | grep compcache
13:34
/lib/modules/2.6.27-6-generic/kernel/ubuntu/compcache
13:34
/lib/modules/2.6.27-6-generic/kernel/ubuntu/compcache/compcache.ko
13:34
/lib/modules/2.6.27-6-generic/kernel/ubuntu/compcache/tlsf.ko
13:34
<vagrantc>
most out-of-kernel modules on debian are in separate packages.
13:35
<stgraber>
it's in linux-image-2.6.27-6-generic but we got rid of of ubuntu-modules in Intrepid
13:35
that's very likely an out-of-kernel module that you could easily package using dkms
13:36* Lns wonders if compcache is stable enough to use at v0.4
13:39
<laga_>
stgraber: http://laga.ath.cx/lts_conf_fixes.diff
13:39
stgraber: but wait a few minutes please, i need to do a test build
13:39
<vagrantc>
looks like there's an ITP for compcache on debian...
13:39bobby_C has joined #ltsp
13:42
<alkisg>
Lns, so I can just install linux-image-2.6.27-6-server (on intrepid) and my kernel will see all 4 Gb, even though it will still be 32bit?
13:43
<vagrantc>
stgraber: have you uploaded 5.1.27 anywhere yet?
13:43
<Lns>
alkisg: i can't vouch for intrepid, but yes :) I always install -server kernel on my i386 hardy installs to get all 8GB on my servers
13:44
<alkisg>
Lns, wow, I guess with all the windows-not-seeing-all-4-gb talks, I forgot I'm using Linux now... :D
13:44
<Lns>
(uses PAE)
13:45
<alkisg>
I guess this means a little overhead, but for one extra Gb, no problem! :)
13:45
<stgraber>
vagrantc: no
13:45
<Lns>
alkisg: that's what i say too
13:45
<stgraber>
laga_: ok, tell me when it's tested
13:45
vagrantc: just pushed the revision to LP but haven't uploaded the new package anywhere as I'm waiting for laga_'s patch
13:46
<laga_>
stgraber: changelog entry: * mythbuntu-diskless: explicitly install xfce4-terminal
13:47
* mythbuntu-diskless: fix lts.conf generation
13:47
(still building)
13:48
<Lns>
And now for some comic relief - http://www.ubergeek.tv/article.php?pid=54
13:49
<stgraber>
laga_: ok
13:49
<laga_>
Lns: "i'm all about open sores. changed my love life" - what?
13:50
<Lns>
laga_: sores... :p
13:50
<stgraber>
vagrantc: why ?
13:52sepski has quit IRC
13:58warren has quit IRC
14:02
<Lns>
Yikes...nbd_swap files don't like to get cleaned up.... root@binkleyubuntu:/tmp# ls -l tmp.*|wc -l 97
14:04maalsa has joined #ltsp
14:05
<laga_>
stgraber: okay, please upload the patch
14:05
it should be safe. i can't test properly now due to unrelated breakage
14:06
<Lns>
13G total - this is completely filling up my hdd :( swap is only at 128mb, too
14:06
<stgraber>
laga_: ok
14:06
<vagrantc>
stgraber: oh, just lookin to do an upload, and if there's already an tarball uploaded to ubuntu, i usually just grab that.
14:09
<stgraber>
laga_: test building the package here, if it's ok I'll ask for a freeze exception
14:09
<laga_>
stgraber: feature freeze exceptions?
14:09mccann has joined #ltsp
14:10
<stgraber>
yeah, I'm pushing a new LTSP
14:10
<laga_>
ah. ok
14:16monteslu has joined #ltsp
14:18artista_frustrad has quit IRC
14:23
<stgraber>
ogra: https://bugs.edge.launchpad.net/ubuntu/+source/ltsp/+bug/281415
14:26artista_frustrad has joined #ltsp
14:31Pascal_1 has joined #ltsp
14:33Pascal_1 has left #ltsp
14:34makghosh has joined #ltsp
14:38mccann has quit IRC
14:43alkisg has left #ltsp
14:46pmatulis has quit IRC
14:53
<Lns>
ogra: any word on the nbd_swap/ltsp-client-setup script bashism fixes? Is this in any way related to swap files sticking around if a client crashes/reboots? I've got a site with 97x 128mb swap files and it's filling up the HDD on a daily basis
14:53
(there are only 35 physical clients there)
14:59
<stgraber>
doh, just got the password expired issue ...
14:59
this one was fixed last I checked
15:00
if [ "$ROOTPASS" = "prompt" ]; then
15:00
$CHROOTEXEC $ROOT passwd -l root
15:00
fi
15:00
vagrantc: ^
15:00
Am I wrong in thinking that the above locks the root password when we choose to set it ?
15:00
shouldn't that be != "prompt" ?
15:01
<vagrantc>
stgraber: indeed, indeed.
15:01
stgraber: sorry about that.
15:02
<stgraber>
ok, I change the logic and tag a new release.
15:02
<vagrantc>
i just changed it quick and didn't really test.
15:02
it was previously a -z test
15:02
so if ROOTPASS was set to "true" or something like that ...
15:03
<johnny>
Lns, buy a bigger HD
15:03
<Lns>
johnny: yeah, that'll buy me about a week's worth of time...
15:03
<vagrantc>
stgraber: and i just downloaded the orig.tar.gz from the bug report ... *sigh*
15:04
<johnny>
Lns, delete em at night?
15:04
<Lns>
johnny: ugh
15:04
not entirely elegant solution, now is it :)
15:05
<johnny>
no
15:05
i don't know what's causing it
15:05
so i can't propose a real solution
15:06
<ltsppbot>
"Lns" pasted "ltsp-client-setup script - swap portion" (46 lines) at http://pastebot.ltsp.org/29
15:07
<johnny>
huh?
15:07
<Lns>
From here (not sure if it's handled anywhere else too) it doesn't look like there's any cleanup mechanism at all
15:07
<johnny>
why are you pasting it
15:07
it shouldn't be cleaning up there should it?
15:07
<Lns>
why not?
15:07
<johnny>
shouldn't it be cleaning up on the server? and reusing existing files?
15:08
<Lns>
I would hope so - but what mech. is responsible for that?
15:08
<johnny>
look at the nbdswap stuff on the server
15:10
<stgraber>
vagrantc: doing a test rebuild just to make sure, I'll then push .28
15:13makghosh has quit IRC
15:15
<stgraber>
vagrantc: new tarball attached
15:16makghosh has joined #ltsp
15:22
<Gadi>
Lns: if you kill the nbd-server procs, then nbdswapd will exit prematurely and not clean up
15:22
Lns: are you killing procs by cron job?
15:22
<Lns>
Gadi: no...i just enabled nbd_swap this week and am experiencing these problems
15:23
<Gadi>
do you kill nbdrootd procs?
15:23
or nbd* procs?
15:23
by cron job?
15:23
<Lns>
I've been manually deleting the files and killall-ing nbdswapd and nbdrootd
15:23
<Gadi>
well, the killing part will leave swap files
15:23
if you kill, you should delete
15:23
but best not to kill
15:24
<johnny>
Gadi, why can't it reuse existing files?
15:24alkisg has joined #ltsp
15:24
<Lns>
Gadi: yeah, i get that part but the thing is even if i don't touch the server, the swap files pile up
15:24
<johnny>
instead of making new ones?
15:24
<Gadi>
johnny: the files are random filenames
15:24
for security
15:24
<johnny>
should they be?
15:24
you think?
15:24
<Gadi>
only the running process knows which file goes with which proc
15:24
<Lns>
Gadi: there's swap encryption code in ltsp-client-setup that would take care of the security aspect..
15:25
<Gadi>
Lns: not completely
15:25
<johnny>
not the issues he's talking about
15:25
<Gadi>
that secures the data in transit
15:25
<Lns>
ah
15:25
<Gadi>
not /tmp symlink redirects
15:26
if a hacker should do: ln -s / /tmp/myfile
15:26
and we do: mkswap /tmp/myfile
15:26
bad things happen
15:26
or worse: dd if=/dev/zero of=$SWAP bs=1024k count=$SIZE 2> /dev/null
15:27
<johnny>
put them somewhere other than tmp? :)
15:27
somewhere that only nbd can write to?
15:27
<Gadi>
that is writable by user nobody?
15:27
<johnny>
change the user
15:27
to nbd
15:27
<Lns>
Gadi: ok, security aside - what about cleanup? i don't see any timeout functionality by default even though there's functionality in nbd-server...although the manpage doesn't say whether it removes the files, just the procs get killed
15:27
<Gadi>
# clean up the swap file
15:27
rm -f $SWAP
15:28
<Lns>
Gadi: where is that?! =p
15:28
<Gadi>
last line of /usr/sbin/nbdswapd
15:28
<Lns>
ah jeez
15:28
:p
15:28
<Gadi>
thats why when you kill the proc it doesnt get executed
15:28
<johnny>
Gadi, he wasn't killing the proc tho
15:28
<Lns>
yea
15:28
<johnny>
not until it was already leaving a bunch
15:28
<Gadi>
well, the proc only dies after 2 hours
15:29
so, you will have lingering files for 2 hours
15:30
in theory
15:30
unless u tell me that tcpd is killing nbdswapd and not just nbd-server
15:30
in which case, that would be a bug worth fixing
15:32
perhaps we should set nbd-server timeout to a time less than 2 hours
15:32
<Lns>
Gadi: i'm not sure really what's going on, all I know is that since 2 days ago (last time I touched the server) there were 97x 128mb swapfiles in /tmp. There are 35 total clients in the computer lab.
15:32
<Gadi>
instead of indefinite
15:32
<Lns>
So *something* isn't cleaning up correctly, even after 2 hours.
15:33
<Gadi>
Lns: try adding: -a 3600 to the nbd-server line in nbdswapd
15:33
and see if that fixes things
15:33
oh, wait
15:33
we have NBD_SERVER_OPS
15:34
cool
15:34
<Lns>
whaaa?
15:34
=p
15:34
<Gadi>
try:
15:34
echo "NBD_SERVER_OPTS=\"-a 3600\"" >> /etc/ltsp/nbdswapd.conf
15:35
and see if that fixes things moving forward
15:35
<Lns>
mmk
15:35
<Gadi>
that should cleanup idle connections after an hour
15:35
<Lns>
Gadi: well lemme set it to something smaller so i can test here at my office
15:36
<Gadi>
ok
15:36
its in seconds
15:36
so, try 30
15:36
<Lns>
k
15:36elisboa has quit IRC
15:38
<Lns>
ok first i'm going to cleanly shut the client down
15:38
woa
15:38
umm
15:38
my swapfile just got created, then disappeared (sitting at ldm)
15:39
maybe i should kick up the timeout ;p
15:39
<Gadi>
well, there's the rub
15:39
because we only call nbdswapd on boot
15:39
<Lns>
?
15:40
<Gadi>
so if the machine stays on for longer than your timeout, and swap is idle, you will lose your swap until reboot
15:40
<Lns>
ah
15:40
<Gadi>
we need to distinguish between idle because of power off
15:40
versus idle because of inactivity
15:40
<Lns>
well that sure doesn't make much functional sense since potentially there will be any number of people sitting on a session, leaving their desk for X number of minutes/hours...
15:41
yea
15:41
<Gadi>
or, we need to repeatedly call for nbdswap
15:41
while the terminal is on
15:41
(this is a chronic problem in client/server systems)
15:41
no pun intended
15:41
<Lns>
IANAP but I'd think that monitoring the session itself would be cleanest
15:42
<Gadi>
client monitoring or server monitoring?
15:42
<Lns>
maybe a gnome-watchdog type thing
15:42
server monitoring client
15:42
<Gadi>
hmm... maybe
15:42
<Lns>
see i'm sure we'll keep running into these things until we have something solid to rely on for client state
15:43
<Gadi>
but, the server is ill equipped to distinguish between a client that is on and one that is unreachable
15:43
<Lns>
gnome-watchdog won't do it (unless we modify it to watch the specific swapfile) since its owner/group is nobody/nogroup
15:43makghosh has quit IRC
15:43
<johnny>
well.. why can't nbd use it's own user acount ?
15:43
and then name the files to be reusable
15:44
<Gadi>
but what would trigger your watchdog?
15:44
<johnny>
no need to trigger a watchdog if it could just pick up and reuse the file?
15:44
<Gadi>
do you not want swap before logging into a session?
15:44* Lns agrees with johnny that reusing files would be best
15:44
<Gadi>
how do you ensure that you delete the correct file?
15:44
<johnny>
no need to delete anything..
15:45
<Lns>
Gadi: well with gnome-watchdog the gnome-panel process is the "main" process that triggers it
15:45
<johnny>
is there some simple security concern i'm missing?
15:45
<Gadi>
so, create a swap file for every workstation regardless of if it is used?
15:45
<johnny>
only if NBDSWAP is on
15:45
<Gadi>
and never delete
15:45
maybe
15:46
<johnny>
i think the tradeoff is much better with that
15:46
<Lns>
that might work...or maybe if client mem >= XX%, launch nbd_swap
15:46
i dunno
15:46
<Gadi>
of course, there is no nbd user
15:46
<johnny>
uhmm.. we can make one
15:46
it's not very hard.. :)
15:47
<Gadi>
this is true
15:47
<vagrantc>
what is the proposal?
15:47
<johnny>
vagrantc, more like brainstorming
15:47
<vagrantc>
the problem is the nbd swapfile deletion issue?
15:47
<Gadi>
to keep swapfiles in /var for nbdswap owned by a special user
15:47
<johnny>
vagrantc, create a user and a directory owned by the users and allow sweap file reuse
15:47
<vagrantc>
or the lack of deletion...
15:48
<Gadi>
vagrantc: ot appears to be broken
15:48
<johnny>
instead of deleting on a timeout
15:48
<vagrantc>
swap file re-use is not a good idea security-wise...
15:48
<Lns>
vagrantc: my issue is the non-deletion of /tmp/tmp.* swapfiles
15:48
<johnny>
well.. we can still delete them
15:48
when the client does properly lgout
15:48
<vagrantc>
although, network swap isn't a good idea security-wise, either.
15:48
<Lns>
lol
15:48
<Gadi>
johnny: you mean shutdown not logout
15:48* johnny thinks nothing is good security wise
15:49
<johnny>
Gadi, sure
15:50
we can still make a best effort to handle such things
15:50
<Gadi>
perhaps we have a better scripting tool at our disposal
15:50
<johnny>
you would say that wouldn't you..
15:50
:)
15:50mhterres has left #ltsp
15:52
<Gadi>
is there a way to issue a write command to swap space?
15:52
does anyone know?
15:52* Lns doesn't
15:53
<Lns>
i'm sure ther eis something..i mean, normal RAM cleanup will swap out unused portions to swap
15:53
<laga_>
swapon/swapoff?
15:53
<Lns>
right?
15:53
<laga_>
well, that's not a smart idea, ignore me
15:53
<Lns>
laga_: that'd be a bit unstable i think :)
15:54
<Gadi>
I am thinking combining an idle timeout on nbd-server with a periodic write to swap on the client
15:54
that way, when the client is off the timeout times out
15:54
but not when it is on
15:54
<Lns>
Gadi: that would probably be the cleanest
15:54
<laga_>
why don't you *read* from the block device?
15:54
<Gadi>
ah, or read
15:55
but how?
15:55
<laga_>
gotta make sure it's not cached, but that's doable i guess
15:55
Gadi: dd if=/dev/nbdswap
15:55
<Gadi>
perfect
15:55
so periodically: dd if=/dev/nbdswap of=/dev/null bs=1 count=1
15:55
<laga_>
there are ways to enforce it to read from the device w/o caching
15:55
<Gadi>
or some such
15:56
<laga_>
O_DIRECT might work
15:57
<Gadi>
I leave the rest as an excercise for the reader ;)
15:57
*exercise
15:57
<Lns>
Gadi: if we do that, say, every 5min... we can also make the 2 hour thing much less, thus making cleanup much more effective
15:57
<Gadi>
Lns: u bet
15:57
in fact, might want something similar for nbdrootd
15:58
<Lns>
for sure
15:59
i don't want to get ahead of this specific thing now..but couldn't we use this method to create psuedo-state client monitoring?
15:59
not necessarily via nbd
15:59
but a custom method
15:59
that's always on by default
15:59
like a real watchdog
16:00
<Gadi>
well, this would monitor if a client is on
16:00
<Lns>
right
16:00
<Gadi>
not if a session is running
16:00
but, it is a good scheme for that
16:01
<Lns>
and if/when someone logs in, a secondary run of that..so you get both client-state and session-state
16:02* Lns thinks there MUST be something already for this that can be easily integrated into ltsp
16:09
<Lns>
I think gnome-watchdog is a good start to building something like this, and it seems to work great for cleaning up processes, but it's too limited. If we integrate what we're talking about with swap into gnome-watchdog and make it more generalized (ltsp-client-watchdog?) we can use that functionality for any number of future issues. We just need to build on the concept and make it see user session state vs. client state.
16:10
<vagrantc>
Gadi: i've been down this road, and apparently it didn't work....
16:10
the whole dd directly thing...
16:10
<Lns>
like what Gadi/laga were discussing w/periodic 1byte writes
16:10
<vagrantc>
the -a timeout option is borked.
16:10
<Gadi>
I dont see the need for watchdog at all
16:10
simply preventing a timeout
16:11
<vagrantc>
http://bugs.debian.org/409531
16:12
does setting in /etc/hosts.allow:
16:12
nbdswapd: ALL: keepalive
16:12
not work?
16:12
or is it just too slow?
16:12
<Lns>
Gadi: why don't you see the need for a watchdog?
16:14
<vagrantc>
stgraber: actually, the whole passwd -l thing could be fixed with a call to chage to unset the expiry time ...
16:14
<Gadi>
Lns: do you have an nbdswapd line in /etc/hosts.allow
16:14
?
16:15
<vagrantc>
i've definitely tested the keepalive option in the past, and it took a couple hours, but did eventually time out
16:16
<Lns>
Gadi: no, i just saw a thread w/that in it actually too
16:16
<Gadi>
ah, there you go
16:16
<Lns>
http://www.mail-archive.com/ltsp-discuss@lists.sourceforge.net/msg34885.html
16:16
is this in hosts.allow by default in newest versions?
16:16
<johnny>
it should be..
16:16
since hardy
16:17
<Lns>
mine only has nbdrootd: ALL: keepalive
16:17
hardy w/hardy-updates
16:17
<johnny>
oh
16:17
for swap
16:17
i guess you do need both
16:17
<Lns>
yup
16:17
<johnny>
that should be in the default
16:17
<Lns>
lemme test that
16:18* Gadi needs to run
16:18
<Gadi>
happy hacking, all
16:18* Lns waves to gadi
16:18
<Lns>
thanks =)
16:18Gadi has left #ltsp
16:20
<Lns>
"And now, we play the waiting game." -Peter Griffin
16:22
vagrantc: nope! my swap just got nuked w/no activity, even with keepalive in hosts.allow
16:22
<vagrantc>
Lns: you still have -a in the nbdswapd ?
16:22
<Lns>
nbdswapd.conf yes
16:22
<vagrantc>
DON'T USE -a with nbd-server.
16:23
<Lns>
-a 120
16:23
<laga_>
WHY
16:23
<vagrantc>
IT'S TOTALLY BROKEN.
16:23
<Lns>
oooook :)
16:23
had it in there from earlier..
16:23
re-testing now
16:24
<vagrantc>
http://bugs.debian.org/409531
16:24
^^ that mentions almost everything people have been talking about for the last good long while
16:26Eghie has quit IRC
16:26
<vagrantc>
i should request an update to the manpage and/or the removal of the option ...
16:27
<Lns>
aah.. ok, sorry i didn't realize the implications of that thing
16:28
ok it's definitely sticking around now, cool.
16:29
now lemme just turn it off
16:31
<alkisg>
To enable local swapping, do I have to (1) create a swap partition, (2) enable the local disks (http://www.ltsp.org/~sbalneav/LTSPManual.html#AEN1296) and (3) set LOCALDEV=True in lts.conf?
16:36
<vagrantc>
alkisg: (1) and (4) set USE_LOCAL_SWAP=True
16:36
<alkisg>
vagrantc, that's all? ok :)
16:37
I thought the "clients don't have access to the hard disks" also included the swap partitions!...
16:38
<vagrantc>
that's LOCALDEV stuff
16:46alkisg has left #ltsp
16:47Q-FUNK has joined #ltsp
16:55Egyptian[Home] has quit IRC
16:56Egyptian[Home] has joined #ltsp
17:01
<Lns>
Just created: https://help.ubuntu.com/community/UbuntuLTSP/EnableNBDSWAP - Can someone verify everything is in there and correct?
17:29alekibango has quit IRC
17:32alekibango has joined #ltsp
17:35bobby_C has quit IRC
17:41Pascal_1 has joined #ltsp
17:41Pascal_1 has left #ltsp
17:41japerry has quit IRC
17:43
<Lns>
ogra: https://bugs.launchpad.net/ubuntu/+source/ltsp/+bug/281498 FYI
17:44
<stgraber>
vagrantc: well, I guess the question is why do we lock the root account at all ? What'd happen if we just don't set it at all ?
17:50
<Lns>
vagrantc: also just filed https://bugs.launchpad.net/ubuntu/+source/ltsp/+bug/281501 FYI..hope all is correct there
17:54Q-FUNK has quit IRC
18:06
<vagrantc>
stgraber: then users can log into the console as root without a password.
18:07
i guess i haven't tried it recently ... but that's how it used to be.
18:08* vagrantc is experiencing swap trhashing ...
18:11
<vagrantc>
that's the last time i try to import 470 more keys into my gpg keyring ...
18:16
<stgraber>
vagrantc: well, I don't get the same error with the account locked as I get with Ubuntu without a root password. I'd need to see how it's done in Ubuntu
18:16
root:!:14146:0:99999:7:::
18:17
that's what we have in Ubuntu for root
18:17
<vagrantc>
stgraber: looks like it's locked by default for ubuntu, then.
18:17
<stgraber>
root:!:14146:0:99999:7::1:
18:17
that's when it's locked
18:18
<vagrantc>
debian reverted a change where passwd -l also sets account expiry dates as well
18:18
sounds like ubuntu still has that
18:20johnny has left #ltsp
18:22warren has joined #ltsp
18:28
<stgraber>
sbalneav: ok, looking at your patch now. As it's bugfix only I won't even need a FFe for Intrepid. For hardy you'll still have to go through the SRU process but that should be made easier with .2 being released in January.
18:35
sbalneav: and I have a good test candidate for it :) (a co-worker who's complaining about it crashing on his no-so-thin client at home)
18:39jc2it has left #ltsp
18:42
<stgraber>
sbalneav: failed to build, looking at the code now
18:43maalsa has quit IRC
18:56petre has joined #ltsp
18:58Lns has quit IRC
19:06johnny has joined #ltsp
19:11staffencasa has quit IRC
19:20maalsa has joined #ltsp
19:26
<warren>
monteslu: ping
19:26
monteslu: are you using the standard fedora firefox and nspluginwrapper?
19:26
monteslu: with all fedora updates?
19:26
monteslu: does flash work on the X login of the terminal server itself?
19:30
monteslu: oh, is this firefox as a local app?
19:58johnny has left #ltsp
20:05vagrantc has quit IRC
20:15moldy has joined #ltsp
20:15
<moldy>
hi
20:20johnny has joined #ltsp
20:21johnny has joined #ltsp
20:21johnny is now known as Guest55303
20:23Guest55303 has left #ltsp
20:49johnny has joined #ltsp
20:50johnny is now known as Guest72217
20:51nubae1 has joined #ltsp
20:52nubae has quit IRC
20:58Guest72217 is now known as johnny
21:35maalsa has quit IRC
21:43maalsa has joined #ltsp
21:54mikkel has quit IRC
21:57maalsa has quit IRC
21:59Loto_ has quit IRC
22:02
<monteslu>
warren, pong
22:02
yeah all standard
22:02
not as a local app
22:02ogra has quit IRC
22:02Loto has joined #ltsp
22:03moldy has quit IRC
22:03
<monteslu>
didnt try flash on the server itself
22:03
which combination of versions with 64bit did you get to work?
22:06
I'll try it on the server this weekend.
22:07
Oh, also. Seems like all users get the yum update notifications
22:08
guess I could just remove whatever applet that is
22:10petre has quit IRC
22:17ogra has joined #ltsp
22:27japerry has joined #ltsp
22:45
<warren>
monteslu: oh, you're probably missing nspluginwrapper.i386
22:45
monteslu: it is not installed by default
22:45
monteslu: you need both archs of nspluginwrapper for it to work
22:49maalsa has joined #ltsp
22:54powellc has joined #ltsp
22:55
<powellc>
I am trying to boot a ppc client off of an i386 server and have got it down to where it looks for the root. It complains about not being able to mount a whole bunch of filesystems before choking up on /sbin/init
22:55
<monteslu>
warren, ok. I'll try that out. I'll follow up on the list as soon as I get it working.
22:56
512MB is sketchy for local apps?
22:57
that's all my ntavo's can do, and they're less than two years old
22:57
i think i can only do 256 on my t150s
22:58powellc has left #ltsp
23:31Loto has quit IRC
23:36Loto has joined #ltsp
23:38
<warren>
monteslu: have you looked at how much RAM firefox alone uses?
23:39
monteslu: normal desktops use firefox by using virtual memory on their local hard drives