[12:40] <Banana> hey folks. First of all kudos for the LTSP5 packages.
[12:41] <RichEd> Banana: kudos go to ogra & sbalneave mainly :)
[12:42] <Banana> I have managed to get everything working on top of Xen amd64 which was kind of a nightmare
[12:42] <Banana> i have a small issue
[12:43] <Banana> i can't seem to find it in the bugzilla so i figured i asked here.
[12:43] <Banana> when i shutdown a client, the / gets umounted too soon
[12:43] <Banana> and the halt script cannot be executed.
[12:44] <Banana> there are I/O errors on nbd-client which complains about a closed socket
[12:44] <Banana> hence the machine hangs just before calling halt
[12:45] <Banana> (non xen here, on a real diskless client).
[12:48] <Banana> uhhh i thikn i found the cause
[12:49] <Banana> the  S60umountroot in runlevel 0 is trying to remount / readonly.
[12:49] <Banana> i'll try to deactivate it.
[01:29] <ogra> Banana, thats feisty ?
[01:30] <Banana> no gutsy
[01:30] <Banana> i found the culprit
[01:30] <Banana> the script sendsigs which sends killall -15/-9 to everyone
[01:30] <Banana> it kills nbd-client early in the shutdown process
[01:31] <Banana> there is a file /var/run/sendsigs.omit
[01:31] <Banana> i'm trying to add pidof nbd-client in it to see
[01:31] <ogra> well, bug 126444 was supposed to be fixed
[01:31] <ubotu> Launchpad bug 126444 in ltsp "[gutsy-tribe2]  <ctrl><alt><del> on tty0 fails to shut down ltsp thin clients" [Undecided,Fix released]  https://launchpad.net/bugs/126444
[01:32] <Banana> i'll go check the bugreport; thanks for the pointer.
[01:33] <ogra> /opt/ltsp/i386/etc/rc6.d/ and /opt/ltsp/i386/etc/rc0.d/ should only have the shutdown/reboot scripts
[01:33] <ogra> at least for all ltsp-build-client runs after version 5.0.25
[01:35] <Banana> hooo
[01:35] <ogra> (note that wont be the case if you built your client chroot with an older version)
[01:35] <Banana> that might be why.
[01:35] <Banana> i built the image early in gutsy.
[01:35] <ogra> yeah
[01:35] <Banana> and then chroot in it and dist upgraded
[01:36] <ogra> you should rebuild with the recent version :)
[01:36] <ogra> there are somw steps ltsp-biild-client performs that we cant do on upgrades
[01:36] <Banana> yes its understandable.
[01:37] <Banana> thanks for the hint.
[01:39] <ogra> also note, if you dist upgrade the chroot, you need to run ltsp-update-image
[01:39] <ogra> (the image is whats served now, the chroot is only for maintenance)
[01:41] <Banana> yes i saw
[01:41] <Banana> the speed up is amazing
[01:41] <Banana> bravo :)
[01:41] <Banana> (for the boot process)
[01:41] <ogra> :)
[01:42] <ogra> it has its drawbacks
[01:42] <ogra> i.e. you have to rebuild the image after changes
[01:42] <ogra> :)
[01:42] <Banana> yes but it seems to make better use of the bandwidth than nfsroot
[01:43] <ogra> yeah
[01:43] <ogra> its a lot faster and smaller
[01:43] <Banana> which matters to me as I have ~50 - 75 diskless pc with raging student working on it at all time.
[01:43] <ogra> it puts some extra load on the client cpu for unsquashing though... but thats only affecting the boot
[01:43] <Banana> and only 100mb connection
[01:44] <ogra> oh, yeah there it surely matters
[01:44] <Banana> i also tested with a xen domU thin client (living on the same machine as the ltsp server)
[01:45] <Banana> works allmost like a charm.
[01:45] <Banana> (mainly xen problems)
[01:45] <ogra> cool
[03:03] <sbalneav> Morning all
[03:03] <sbalneav> ogra: Hey, how's it going?
[03:04] <sbalneav> I did some testing on the weekend.  Localdevs seems to be working great!  Thanks for the upload.
[03:06] <ogra> pulling my hair out over ltsp-update-image
[03:07] <ogra> that inetd/port code is such a bad mess
[03:08] <sbalneav> yeah, I improved it a little, but it's kind of a sticky problem.
[03:08] <sbalneav> What can I help with?
[03:08] <ogra> we have two places that need to be identical all the time (PXE and inetd)
[03:09] <ogra> beyond that there are x special cases that need to be taken into account
[03:09] <sbalneav> Oh, by the way, couple of minor bugs squashed on the weekend, most important one related to X_RAMPERC.  ulimit -v isn't there under dash's shell, so changed it to ulimit -m
[03:09] <ogra> we also need to make sure that we always exit 0 ... unless there is no image
[03:10] <ogra> ah, thats good
[03:10] <sbalneav> Hey, something else I was wondering.
[03:10] <ogra> yeah, i found that mkfifo isnt in the d-i environment :/
[03:10] <ogra> have to use mknod ... grrr
[03:12] <sbalneav> I did some testing, and on all the boxes I tested, an X_RAMPERC=80 makes sure nothing bad happens, kills firefox on insane web pages, etc.  I was wondering, should we think about setting a default X_RAMPERC in ltsp_config?  Might help to elimintate some of those "my workstation locks up for no apparent reason" type bug reports?
[03:12] <ogra> could go into the initramfs code that checks for low ram
[03:13] <sbalneav> heh ram < 512m then X_RAMPERC=80?
[03:13] <ogra> lol
[03:13] <sbalneav> I tested that insane website that Jim K posted
[03:14] <sbalneav> it allocates, like 512 megs of pixmaps!
[03:14] <ogra> well, i was planning to release with an empty lts.conf so it should be set elsewhere
[03:14] <ogra> (read we'll need an override var for that )
[03:14] <ogra> NO_RAM_LIMIT
[03:14] <ogra> or so
[03:15] <ogra> or X_RAMPERC=0 disables it
[03:15] <sbalneav> Well, or X_RAMPERC = 100?
[03:15] <sbalneav> RAMPERC's such a bad name.
[03:15] <sbalneav> What else can we call it?
[03:15] <ogra> so people not wanting it can switch it off
[03:15] <sbalneav> sure = 0
[03:15] <ogra> MAX_USED_RAM ?
[03:15] <ogra> MAX_RAM ?
[03:16] <ogra> RAMLIMIT
[03:16] <sbalneav> X_MAX_RAM?
[03:16] <sbalneav> X_RAMLIMIT
[03:16] <ogra> well, you set all ram
[03:16] <sbalneav> I like that
[03:16] <ogra> not only X's
[03:16] <sbalneav> No, just X's
[03:16] <sbalneav> it's done in the ldm
[03:16] <sbalneav> so it won't affect other things.
[03:17] <ogra> doesnt ulimit set kernel params ?
[03:17] <sbalneav> Not as far as I know.  Just that process and all subprocesses
[03:18] <sbalneav> so we set it in the ldm screen script, it should set for that process, and children.
[03:18] <ogra> ah, ok
[03:18] <ogra> well, just because we're root
[03:22] <ogra> sbalneav, cant we just read /proc/meminfo and automtically limit to the ramsize ?
[03:22] <ogra> or better ramsize + swapsize
[03:22] <ogra> seems more elegant somehow
[03:23] <ogra> and have a switch tghat only en or disables that fucntion
[03:29] <sbalneav> That's what we do
[03:29] <sbalneav> The question is just WHERE you set the limit
[03:30] <sbalneav> if you set it too close to the max, some things don't like it.  If you set it too low, you're not making full use of the ws's resrources.
[03:31] <sbalneav> we add up free ram and free swap.  And though some (long) testing this weekend, I found 80 makes the best sense.
[03:31] <ogra> leave a 10% buffer ?
[03:31] <ogra> 80 doesnt make sense on a client with 256M and 512M nbd swap in a tmpfs :)
[03:32] <ogra> i'd go with a certain percentage below max
[03:33] <ogra> not friday anymore :)
[03:33] <sbalneav> Well, what I was coming across was that, if I set it much above 85, mozilla would crash, and the desktop would return, but the mouse and keyboard would be frozen.  If, on the server, I pkilled the user, the desktop would recyle, but the mouse and keyboard would still be latched. At 80, I found everything worked as expected.
[03:34] <sbalneav> ok, maybe we can do some math, and set some sensible values.  I'll put some thought into it today.  Should we put the code in ltsp_config?  That'd be the simplest.
[03:34] <ogra> at with what physical ram size ?
[03:34] <ogra> s/at/at 80/
[03:34] <sbalneav> yeah, physical ram size
[03:34] <sbalneav> i.e. if you've got 512 ram, probably 95 is fine
[03:35] <ogra> only ?
[03:35] <ogra> why not more ?
[03:35] <sbalneav> 98
[03:35] <sbalneav> I'd have to play around with some values.
[03:35] <ogra> how much phys ram did you have on the machine you tested 80 on ?
[03:35] <sbalneav> 256
[03:35] <ogra> urgh
[03:35] <sbalneav> 32 meg swap
[03:35] <ogra> thats a huge waste
[03:36] <ogra> we probably need 20M for everything !X
[03:36] <ogra> so you would have 150M idling there
[03:37] <sbalneav> 256 + 32 * .8 = 230 meg
[03:37] <ogra> 80 is percentage ?
[03:37] <sbalneav> yeah
[03:37] <sbalneav> yeah, percentage
[03:37] <sbalneav> ramperc :)
[03:37] <sbalneav> ram percentage :)
[03:38] <ogra> sorry, better ignore me today
[03:38] <sbalneav> lol, NP
[03:38] <ogra> :)
[03:38] <ogra> ok, then 80 is perfect indeed :=
[03:38] <ogra> :)
[03:39] <sbalneav> Anyway, merge from me now, so we get the default behavior of it not being set at all, and then I'll fool around and try to set a sane value based on ram in the client, swap size, etc.
[03:39] <ogra> did you make a change the last hour ?
[03:39] <sbalneav> no
[03:39] <ogra> i just merged this mornoing
[03:39] <ogra> *morning
[03:39] <sbalneav> Effecient as always :)
[03:39] <sbalneav> They're not paying you enough :)
[03:40] <ogra> i just want -update-image half way sane before i package it
[03:40] <ogra> else the package would be up already
[03:40] <sbalneav> ok, I'll quit bending your ear, and let you get to it :)
[04:27] <bddebian> Heya
[08:19] <LaserJock> hi ogra
[08:19] <ogra> hey
[08:20] <LaserJock> ogra: how was your weekend?
[08:20] <ogra> busy
[08:21] <LaserJock> manage to get that upload done?
[08:53] <bdoin> ogra: latest GCompris release is 40MB smaller, I am sure you will appreciate ;)
[08:53] <LaserJock> wow
[08:53] <LaserJock> how'd you do that?
[08:53] <ogra> \o/
[08:54] <bdoin> we re-encoded all voices in a lower bitrate
[08:54] <LaserJock> ahhh
[08:54] <LaserJock> well done
[08:54] <ogra> yay
[08:56] <bdoin> take care, in the latest release there are several changes that impacts packaging
[08:57] <LaserJock> oh man, so 40MB comes at a price? :-)
[08:58] <bdoin> you can see it this way but it's unrellated
[08:58] <ogra> well, we went fine with the debian packages the recent time ...
[08:59] <bdoin> we integrated some libraries in order to ease packaging and switched to gstreamer instead of GTK
[08:59] <ogra> gstreamer ++
[08:59] <ogra> very good
[09:00] <bdoin> oops sorry, gstreamer instead of SDL_Mixer
[09:00] <ogra> indeed :)
[09:01] <bdoin> do you have cooperations with the olpc team on some subject ?
[09:02] <ogra> only very loosely
[09:03] <sbalneav> gstreamer = ++++good.
[09:03] <sbalneav> \o/ for bdoin
[09:03] <bdoin> thanks
[09:04] <bdoin> at some point we could imagine that the sugar user interface is also a good choice for regular children oriented desktop
[09:04] <sbalneav> ogra: too late for a UVFe for something that big, eh?
[09:05] <ogra> yeah
[09:05] <ogra> its on the addon CD anyway
[09:05] <ogra> all we'd gain would be some MB in the live squashfs
[09:05] <sbalneav> bdoin: gstreamer's good, as it plugs in very nicely with the remote sound stuff on ltsp.
[09:05] <bdoin> what's a UVFe ?
[09:05] <ogra> !uvf
[09:05] <ubotu> uvf is Upstream Version Freeze.  For an exception, see https://wiki.ubuntu.com/FreezeExceptionProcess#head-9523bc4076ff011324d67cddc97969ec609618d6
[09:05] <sbalneav> Upstream Version Freeze exception.
[09:06] <ogra> sbalneav, lazier is faster :P
[09:06] <sbalneav> heh, there a way to see what's ubotu's list of factoids is?
[09:06] <sbalneav> !factoids
[09:06] <ubotu> I am ubotu, all-knowing infobot. You can browse my brain at http://ubotu.ubuntu-nl.org/factoids.cgi - Usage info: http://wiki.ubuntu.com/UbuntuBots
[09:06] <ogra> "all-knowing infobot"
[09:06] <ogra> tsk
[09:07] <ogra> ubotu, whats the color of my socks ?
[09:07] <ogra> bah
[09:07] <LaserJock> supposedly I'm a factoid editor
[09:07] <sbalneav> Ubotu needs a "seen " function.
[09:07] <LaserJock> I could whip up a nice response
[09:07] <ogra> and a socks function
[09:08] <LaserJock> "Edubuntu colors, of course" would be a good one
[09:08] <ogra> a force feedback socks function that also pulls them :)
[09:08] <LaserJock> haha
[09:08] <sbalneav> ogra: if they're white, they're clean, if they're green, time to change 'em :)
[09:08] <ogra> lol
[09:08] <LaserJock> lol, such wisdom
[09:19] <stgraber> :)
[09:20] <stgraber> ogra: saw my comment about italc ? (posted on Saturday)
[09:20] <ogra> that you got it running ?
[09:21] <stgraber> yep
[09:21] <stgraber> except things like shutdown/reboot of course
[09:21] <ogra> yeah
[09:21] <stgraber> and I still have some problems with the demo mode
[09:21] <ogra> but that sounds good
[09:21] <ogra> we'll need to get it building without libssl dependency though
[09:22] <stgraber> yes, you basically run one ica (the user daemon) for each user, then make it listen on a different port (default+client number in my case)
[09:22] <stgraber> it should also be easy to do a launcher for the teacher tool so it grabs the list of connected clients and generate the XML
[10:43] <clarkkent93> looking for help....installed edubuntu but missing education programs
[10:47] <clarkkent93> can anybody help?
[10:56] <monteslu> clarkkent93, "gksudo synaptic"
[10:57] <monteslu> just install whatever you want from there
[10:57] <clarkkent93> hmmm...i've tried to go to add/remove programs and it shows the stuff but says that it can't install due to either computer type or something
[10:57] <clarkkent93> should what you wrote help?
[11:01] <monteslu> not sure, but give it a shot, and make sure to have it update availabe packages
[11:06] <clarkkent93> ok..i'll wait until i get a lan cable this afternoon...didn't realize that i lost it when i moved.
[11:07] <clarkkent93> what should i be looking for in that gksudo synaptic?
[11:10] <clarkkent93> i'm completely new to linux and edubuntu...i imagine you got that from my questions
[11:13] <monteslu> clarkkent93, I'm actually on fedora right now, but I believe you can browse the different sections for software you want to install, and there are some educational categories
[11:13] <clarkkent93> shoot...i didn't see them
[11:27] <cliebow> clarkkent93, got it figgered out?
[11:27] <clarkkent93> no
[11:27] <clarkkent93> i'm an idiot
[11:28] <clarkkent93> i used the gksudo synaptic command but don't see the packages
[11:28] <cliebow> i think there was a second add-on cd with lots of ed apps..but you can wing it with synaptic
[11:28] <clarkkent93> do i need to be hooked up to the internet to use it?
[11:28] <cliebow> system-administration->synaptic
[11:29] <cliebow> yes pretty much\
[11:29] <cliebow> to download
[11:29] <clarkkent93> maybe i'll wait until later to ask the question then....
[11:29] <clarkkent93> i did notice in the add/remove programs that it had the education programs listed but gave some reason why it couldn't be installed.
[11:30] <cliebow> ill try to keep an eye out for you..
[11:30] <cliebow> for what good it may do you 8~)
[11:30] <cliebow> you need a network connection for sure
[11:32] <cliebow> i see there are meta packages for edubuntu-addon-science and edubuntu-addon-young
[11:32] <cliebow> etc etc
[11:32] <cliebow> i am in gutsy so mine wont match yourn
[11:32] <clarkkent93> says doesn't support my computer type...x86 because it doesn't meet special hardware requirements or vendor doesn't support computer type
[11:34] <cliebow> what is your machine??anything special?
[11:35] <clarkkent93> nope...amd athlon...homebuilt system
[11:36] <clarkkent93> small hd..only 13gb...imagine that!  hehe...there's still about 9 gigs left though
[11:36] <clarkkent93> 256 mg ram
[11:36] <cliebow> shouyld be just fine..
[11:37] <clarkkent93> thought so...weird...maybe it will work out fine once i get the network connection going
[11:37] <cliebow> get hooked up.. give it a whirl..usually helpful folks on #edubuntu
[11:37] <clarkkent93> once i hook up the lan...anything special i need to do to set it up?
[11:37] <clarkkent93> or should it be automatic?
[11:37] <cliebow> maybe run update-manager first
[11:38] <cliebow> get latest of everything..then choose packages
[11:44] <clarkkent93> thaks cliebow!  i'll be back on later after i do the updates and let you know how it goes