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


Channel log from 18 January 2010   (all times are UTC)

00:08
<sbalneav>
Heading to bed. Night all.
00:09
<stgraber>
night sbalneav
00:15
<elmo40>
interesting... http://sf.net/projects/symbiont/ webmin-based tool for managing ltsp. Jan 2009.
00:24
<GGD_>
elmo40: it is but its not been kept up
00:27FGXR6 has quit IRC
00:30FGXR6 has joined #ltsp
00:32pasmen has joined #ltsp
00:40
<elmo40>
one year old. for 4.1, i believe
00:41
hrmm, trying to boot this laptop with the NIC and it is hanging on the routers IP. How do i tell the router to show the server to it? is there a port to open?
00:54GGD_ has quit IRC
01:07
<Appiah>
"hanging on the routers IP" ?
01:14
<alkisg>
elmo40: do you mean that you have 2 dhcp servers and 1 nic on your ltsp server?
01:22Selveste1 has joined #ltsp
01:35shogunx has joined #ltsp
01:35
<shogunx>
hi guys.
01:36
i need a bit of a advice
01:36
i have just installed one of my solarnetone ltsp networks on abemama, kiribati in the south pacific.
01:37
i am running out of time on the island, and i need a method to allow timed logins
01:38
the operator does prepay billing, and they need to be able to sell an hour of service and have that hour essentially logged and the session to time out when complete.
01:38
any ideas? ogra?
01:39
not sure i can write that in an day and a half...
01:47
<alkisg>
shogunx: http://projects.gnome.org/nanny/ enforces time limits on sessions, but it's oriented to kids. Maybe you could get some code out of it, though...
01:47
<shogunx>
cool
01:47
thanx... i'll check it
01:47
<alkisg>
np
01:53jcastro has quit IRC
01:56jcastro has joined #ltsp
02:01mikkel has joined #ltsp
02:13NeonLich1 has quit IRC
02:14NeonLicht has joined #LTSP
02:30KaRoe_ has quit IRC
02:59vagrantc has quit IRC
03:27shogunx has quit IRC
03:38ftherese has quit IRC
04:17sebas891 has quit IRC
05:00hersonls has joined #ltsp
05:28lucascoala has joined #ltsp
05:37Faithful has quit IRC
05:45hersonls has quit IRC
05:45hersonls has joined #ltsp
05:47ogra has quit IRC
05:48ogra has joined #ltsp
06:12pmatulis has joined #ltsp
06:21Selveste1_ has joined #ltsp
06:21Selveste1 has quit IRC
06:40bobby_C has joined #ltsp
06:41
<alkisg>
SOUND should be forced=FALSE for fat clients. So I want to put an "if boolean_is_true "$LTSP_FATCLIENT"; then export SOUND=False; fi"
06:41
Should I put that in ltsp_config or in ltsp-init-common?
06:50GGD_ has joined #ltsp
06:55Selveste1__ has joined #ltsp
07:04mikkel has quit IRC
07:08_UsUrPeR__ has quit IRC
07:09Selveste1_ has quit IRC
07:11GGD_ has quit IRC
07:16sene has joined #ltsp
07:23seppl has joined #ltsp
07:35try2free has joined #ltsp
08:07Selveste1___ has joined #ltsp
08:14seppl has left #ltsp
08:16evilx has joined #LTSP
08:18Faithful has joined #ltsp
08:19gnunux has joined #ltsp
08:19GGD_ has joined #ltsp
08:21Selveste1__ has quit IRC
08:25Gadi has joined #ltsp
08:29_UsUrPeR_ has joined #ltsp
08:34GGD_ has quit IRC
08:44prpplague_afk is now known as prpplague
08:49try2free has left #ltsp
09:26
<sbalneav>
MoMorning all
09:27
<alkisg>
!s
09:27
<ltspbot>
alkisg: "s" :: Scotty!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
09:28
<sbalneav>
Morning alkisg
09:29litlebuda has quit IRC
09:32Selveste1___ has quit IRC
09:39rhodan_ has joined #ltsp
09:39gentgeen__ has joined #ltsp
09:44
<alincoln>
how does ltsp-build-client decide what packages to put in the chroot? i've been poking at the scripts for a while and can't tell.
09:46
oh, i think it's 040-copy-package-lists.
09:47
<alkisg>
It puts a hardcoded set of packages, unless the user specifies one of the '--early-packages' or '--late-packages' options
09:47damianos has joined #ltsp
09:47
<alincoln>
alkisg: ah, ok. i was wrong.
09:47
alkisg: where is the hardcoded list?
09:47
<alkisg>
In the source code :P :D
09:47
No seriously it's in more than 1 plugins
09:47etyack is now known as etyackAKAmrMom
09:48
<alincoln>
alkisg: heh, fair enough
09:49
<damianos>
Hello
09:49
Gia sou aklisg
09:50
<alkisg>
Geia sou damianos :)
09:50
<alincoln>
alkisg: if i go into the chroot and do 'dpkg --get-selections', i get a list of packages installed by ltsp-build-client into the chroot. but these package names aren't in any of the plugin scripts. they must be defined another way, right?
09:51
<damianos>
Can somebody point me to some documentation on how to boot a thin client with a floppy?
09:51
<Appiah>
umm
09:51Barbosa_ has joined #ltsp
09:51
<Appiah>
gPXE maybe has something like that
09:51
<damianos>
Everything I've found online is years old
09:52
<Appiah>
yes thats beacuse the floppy is old
09:52
people use usb sticks today
09:52
<alkisg>
alincoln: many of them are dependencies....
09:52
<gnunux>
alincoln: package are installed with debootstap
09:52
<Appiah>
http://www.etherboot.org/wiki/removable#creating_a_bootable_floppy_disk
09:52
<alkisg>
alincoln: E.g. if you run `sudo apt-get install ubuntu-desktop`, you'll get hundrends of programs
09:53
<alincoln>
alkisg: ah, ok.
09:53
gnunux: thanks
09:53
<damianos>
Thanks Appiah...I'll go check it out
09:53
<alkisg>
damianos: http://rom-o-matic.net/gpxe/gpxe-0.9.9/contrib/rom-o-matic/
09:54
Just click the button there.
09:54CAN-o-SPAM has joined #ltsp
09:55rhodan has quit IRC
09:55
<alkisg>
(the get image button - you don't need to modify any setting)
09:55
<damianos>
Thanks alkisg....btw a thin client i've been using for testing was working fine then I started getting a media test failure. Is the media being referred to system RAM?
09:56
<alkisg>
I don't think so. Maybe the network cable?
09:56
<damianos>
Nope....I setup a virtual machine on my laptop to act as a client and it boots fine
09:57Barbosa has quit IRC
09:57litlebuda has joined #ltsp
10:11otavio has quit IRC
10:18
<alincoln>
i'm testing LTSP 5.1.98 in Ubuntu 10.04, and getting some errors on startup because lspci isn't available in the image. it seems that it's been removed from EARLY_PACKAGES for ltsp-build-client for 5.1.98, even though it was in this variable for LTSP 5.1.90.
10:18
was it intended to remove hal from the image? if so, should the scripts that rely on it be removed?
10:19
<alkisg>
alincoln: that was fixed recently, lspci should be on the image on newer builds
10:19
<alincoln>
alkisg: ah, ok
10:19
<alkisg>
For now just put it there manually
10:19
<alincoln>
gotcha
10:31damianos has quit IRC
10:33
<_UsUrPeR_>
ok, so I just upgraded a ltsp server from 9.04 to 9.10, and almost all of the ltsp users (aside from my own personal user) are not able to start X. Here's the auth.log and syslog files:
10:33
http://pastebin.com/m2ef79a2f
10:33
it's saying that it can's start an nm-applet.desktop, and a /proc directory can't be read :/
10:36
I am able to log on as these users on the server, just no clients
10:37
<Gadi>
the proc thing sounds like italc or some such
10:37
and nm-applet is network-manager
10:37
your users don't need nm-applet
10:38
<sbalneav>
Gadi: Morning!!!
10:38gnunux has quit IRC
10:38
<Gadi>
!s
10:38
<ltspbot>
Gadi: "s" :: Scotty!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
10:38
<_UsUrPeR_>
!s
10:38
<ltspbot>
_UsUrPeR_: "s" :: Scotty!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
10:38
<_UsUrPeR_>
hey the bot works! :D
10:38
<Gadi>
_UsUrPeR_: neither of those errors should be blocking
10:38* sbalneav reminds gadi about a breif one pager on rdesktop for the manual
10:38
<Gadi>
_UsUrPeR_: check auth.log for ssh auth issues
10:38
<alkisg>
Also check for compiz problems :)
10:39
<Gadi>
sbalneav: ah, thx! Todays a good day to do it, too
10:39
bless u, MLK,Jr.
10:39
<_UsUrPeR_>
gadi: auth.log on the server is in the pastebin I provided
10:39
<sbalneav>
Ah, yes, happy MLK day to our american cousins. Here's to dreaming!
10:39
<_UsUrPeR_>
hmm. another weird thing: SCREEN-02 = shell does not seem to be working properly. I hit ctrl+alt+F2 to go to tty2, and instead I am seeing a garbled version of ldm background
10:40
<Gadi>
SCREEN_02 you mean
10:40
<_UsUrPeR_>
gadi: yeah
10:40
SCREEN_02
10:40
<sbalneav>
xplash brokenness?
10:40
<Gadi>
yeah, sounds like a wonky compositing manager
10:40
uninstall compiz
10:40
and see if everything comes up Milhouse
10:40
<sbalneav>
ahh, compiz
10:41
Some people have the bogeyman, we have compiz
10:41
<_UsUrPeR_>
Gadi: milhouse = bring, blue hair, glasses, divorced parents?
10:41
s/bring/boring
10:41
<sbalneav>
"The program you love to hate" :)
10:41
<Gadi>
indeed
10:41
on all counts
10:41
<_UsUrPeR_>
I hope my ltsp session doesn't have a broken childhood
10:42
<alkisg>
_UsUrPeR_, try this: sudo gconftool-2 --direct --config-source xml:readwrite:/etc/gconf/gconf.xml.mandatory --type string --set /desktop/gnome/session/required_components/windowmanager metacity
10:43
You may need to restart gconfd... (or reboot or something)...
10:43
<_UsUrPeR_>
alkisg: is that in chroot, server or client?
10:43
<alkisg>
Server, it disables compiz for everyone
10:43
<_UsUrPeR_>
k
10:43* _UsUrPeR_ restarts client
10:45
<_UsUrPeR_>
alkis: that worked. huh.
10:45
also fixed the screen_02 issue
10:47
alkisg: thanks.
10:47
<alkisg>
np - hope your driver is fixed soon :)
10:47
<_UsUrPeR_>
oh that's a VIA issue?
10:47
I'll keep chasing that rainbow.
10:48
I.E. "fat chance" ;)
10:56pasmen has quit IRC
10:57Pod2 has joined #ltsp
11:04mikkel has joined #ltsp
11:07vvinet has joined #ltsp
11:11
<Gadi>
sbalneav: where in the manual do you intend to put the rdp stuff?
11:11* Gadi wonders what context he should put stuff in
11:12
<sbalneav>
Gadi: I'd probably give it it's own chapter
11:12
since lots of people have an interest in it, it's important enough to separate out as a chapter.
11:14
<Gadi>
ok
11:15
<sbalneav>
Just do plain text, or an OO document, whatever's easier.
11:15
I can docbook it after that.
11:59otavio has joined #ltsp
12:00
<etyackAKAmrMom>
sbalneav: is there a document describing you want documentation submitted?
12:04
<sbalneav>
etyackAKAmrMom: You mean, how I want documentation submitted?
12:04
I'll take it any way I can get it
12:05
Reformatting for docbook's just an excercise in text processing
12:10Pod2 has left #ltsp
12:12
<Gadi>
sbalneav: keep an eye out for my email
12:13
<etyackAKAmrMom>
sbalneav: k, and your are the docmaster too
12:14alkisg has quit IRC
12:14
<sbalneav>
etyackAKAmrMom: docmaster. Cool title.
12:15
I was gunning for something more along the lines of "Keeper of ye olde mysikal LTSP Lore and Spells"
12:15
but docmaster's shorter :)
12:19
<etyackAKAmrMom>
yah, that sounds more like you. tough to print on a badge though, maybe a belt buckle, a BIG belt buckle.
12:19
<CAN-o-SPAM>
heh
12:25
<sbalneav>
Gadi: Got it! I'll add it now.
12:42alkisg has joined #ltsp
12:55alincoln has quit IRC
12:55alincoln has joined #ltsp
12:57alincoln has quit IRC
12:58alincoln has joined #ltsp
12:58otavio_ has joined #ltsp
12:58dro has joined #ltsp
13:06otavio has quit IRC
13:09pasmen has joined #ltsp
13:19alkisg1 has joined #ltsp
13:21alkisg has quit IRC
13:24alkisg1 is now known as alkisg
13:28pasmen has quit IRC
13:30pasmen has joined #ltsp
13:34
<sbalneav>
Gadi: OK, check out the docs link
13:34
!docs
13:34
<ltspbot>
sbalneav: "docs" :: For the most current documentation, see https://sourceforge.net/apps/mediawiki/ltsp/index.php?title=Ltsp_LtspDocumentationUpstream
13:34
<sbalneav>
I've pushed a new version of the pdf
13:37
<knipwim>
sbalneav: why is there no mediawiki representation of the pdf?
13:37
<alkisg>
sbalneav: "Unable to find any mirror information for the "/LTSPManual.pdf" file. Please select another file."
13:40
<sbalneav>
knipwim: Do you know of a docbook to mediawiki conversion routine?
13:41
<knipwim>
xslt?
13:41
i've never done it
13:41
could give it a try
13:41
<sbalneav>
knipwim: Soon as you tell me how to do it, I'd be more than happy to post it.
13:42
Only problem is, if you pasted the content on a wiki, then people may change it.
13:42
how do I go from the wiki back to docbook?
13:42
Inverstigaet that at the same time.
13:42
<knipwim>
:)
13:42
okay
13:43
i'm going on holiday next week
13:43
that sounds like something to do then
13:43
<sbalneav>
yeah yeah
13:43
<knipwim>
do you have a docbook source somewhere?
13:43
<sbalneav>
Which, the source for the manual?
13:44
<knipwim>
yes
13:44
<sbalneav>
bzr branch lp:~ltsp-docwriters/ltsp/ltsp-docs-trunk
13:44
<knipwim>
check
13:45
<sbalneav>
crumb, why did it upload to the wrong place?
13:47
eugh, SF's as slow as molassas in January
13:47
it's there, it's just not under docs-admin-guide
13:47
I'll have to see about moving it tonight.
13:49pasmen has quit IRC
13:58otavio_ has quit IRC
13:58otavio has joined #ltsp
14:01Selveste1___ has joined #ltsp
14:07etyackAKAmrMom has quit IRC
14:19otavio has quit IRC
14:38otavio has joined #ltsp
14:44otavio has quit IRC
14:45pasmen has joined #ltsp
14:46juub has left #ltsp
14:48otavio has joined #ltsp
15:01hersonls has quit IRC
15:02pasmen has quit IRC
15:03pmatulis has quit IRC
15:14pasmen has joined #ltsp
15:32pasmen has quit IRC
15:35spectra has quit IRC
15:38vvinet has quit IRC
15:46tstafford has quit IRC
16:07mikkel has quit IRC
16:09artista-frustrad has quit IRC
16:09artista_frustrad has quit IRC
16:11pasmen has joined #ltsp
16:11artista_frustrad has joined #ltsp
16:12allinone has joined #ltsp
16:13dro has quit IRC
16:16
<alkisg>
Gadi: since you're optimizing ltsp_config, wouldn't it be a good idea to do `if [ "$LTSP_CONFIG" = "True" ]; then return 0; fi` as its first line? Kind of like #ifdef in C header files?
16:21
<Gadi>
what is LTSP_CONFIG?
16:21
<alkisg>
It's set at the end of that file
16:21
# set to determine if ltsp_config has already been sourced
16:21
export LTSP_CONFIG=${LTSP_CONFIG:-"True"}
16:22
So it's usually used by the scripts that source ltsp_config
16:22
if [ -z "$LTSP_CONFIG" ] then source it...
16:22
But it's not done consistently
16:22
It'd be much faster and more consistent if it was done inside ltsp_config, I think...
16:23
The idea is the same as in C header files: #IFNDEF _HEADER_, #DEFINE _HEADER_, code, #ENDIF...
16:24
<Gadi>
well, you can only return from a function, so we would need to put all the upfront stuff in a function
16:25
<alkisg>
Gadi: You can return from a sourced script as well...
16:25
<Gadi>
something like: set_defaults()
16:25
<alkisg>
To "cancel" the sourcing
16:25
<Gadi>
are you sure?
16:25
<alkisg>
Yeah, we already use that in the fat client script
16:26
I don't know how POSIX-standard that is, but sh works fine with it, and I think a lot of ltsp-build-client plugins use it
16:26
<Gadi>
which fat client script?
16:26
the plugin?
16:26CAN-o-SPAM has quit IRC
16:26
<alkisg>
Yeah
16:26
<Gadi>
plugins are already functions
16:27
ie the different sections are functions
16:27
<alkisg>
Ooooh
16:27
Hmmmm :)
16:27
<Gadi>
I think
16:27pasmen has quit IRC
16:27
<Gadi>
iirc
16:27
its been a while since I looked at plugins
16:27* alkisg makes an example script to try it...
16:27
<Gadi>
I was gonna say... :)
16:29
<alkisg>
Yeah, works fine...
16:29
<allinone>
hey Gadi
16:29* alkisg also tries with bash...
16:29
<alkisg>
Works with bash as well.
16:29
<allinone>
just saying howdy, the library is closing now
16:30
Gadi: catch you later
16:30allinone has quit IRC
16:31
<Gadi>
ok
16:31
Ill add it in then
16:31
and put you in the commit message :)
16:32pasmen has joined #ltsp
16:32
<alkisg>
Heh, no need, I'm trying to avoid being blamed :P :D
16:34prpplague is now known as prpplague_afk
16:34
<Gadi>
alkisg: why is it:
16:34
export LTSP_CONFIG=${LTSP_CONFIG:-"True"}
16:34
and not simply:
16:35
export LTSP_CONFIG=True
16:35
<alkisg>
Gadi: I've *no* idea!!! I couldn't understand it either
16:35
<Gadi>
unless you wanted to be able to ALWAYS source it
16:35
<alkisg>
Maybe in the case someone doesn't want it to be bypassed?
16:35
<Gadi>
by setting LTSP_COCFIG=False in lts.conf
16:35
<alkisg>
Right, that
16:35
I think debian does that
16:36
<Gadi>
how come?
16:36
are we not understanding its use?
16:36
<alkisg>
We have to ask vagrantc...
16:37
But I don't think that an "if [ "$LTSP_CONFIG" = "True" ] on top causes any harm
16:37
<stgraber>
I just mentioned it on #edubuntu but will repeat it here as it's a more appropriate channel ;)
16:37
<alkisg>
Just leave the bottom line intact, and we'll ping vagrantc on the first chance
16:37
<stgraber>
the idea is to have LTSP 5.1.99 be the last release before 5.2, that's why I haven't tagged it yet
16:38
I'm waiting for a few other fixes and tests as I want it to be rock solid, then we can do some more bugfix + translation updated and release 5.2
16:42
<Gadi>
stgraber: can you test the rdesktop script with sound to make sure we didnt break it?
16:42
RDP_OPTIONS = "-r sound:local"
16:46
<stgraber>
Gadi: not right now but I can test it tomorrow at the office
16:48
<Gadi>
alkisg: we should do the same with ltsp-common-functions
16:48
alkisg: and then call ltsp-common-functions in ltsp-config
16:48
<alkisg>
Gadi, sounds good, I hope we'll see some speed improvements from that.
16:49
<johnny>
now.. if only every distro used plymouth and upstart...
16:50
<Gadi>
hmm...
16:50
<johnny>
one day ltsp will be able to have native upstart scripts in any case..
16:50
<stgraber>
I already have a few at the office ;)
16:50
but that's for testing, nothing worth pushing upstream yet
16:51
most useful ones (that I'll probably push soon) are shutdown/reboot ones
16:51
they basically catch any shutdown/reboot and force it (as if you did a -f) so you won't get any squashfs error anymore.
16:51
(it also does a sync before doing so, just in case an external device is plugged in)
16:53
<johnny>
one day.. it'd be great if the ltsp scripts did less :(
16:53
<Gadi>
anyone know how to check for a built-in function in shell?
16:53
<johnny>
and we could rely on the distro provided scripts more..
16:54
try to run it? :)
16:54
/me doesn't know if such things are stored anywhere..
16:56mushroomtwo has quit IRC
16:56
<stgraber>
Gadi: an hackish way to do it would be to try to export it
16:56mushroomtwo has joined #ltsp
16:56
<stgraber>
but I suspect there's an easier way
16:56
stgraber@castiana:~/data/code/ltsp-trunk$ test() { echo test; }
16:56
stgraber@castiana:~/data/code/ltsp-trunk$ export -f test
16:56
stgraber@castiana:~/data/code/ltsp-trunk$ export -f test1
16:56
bash: export: test1: not a function
16:58pasmen has quit IRC
16:59* Gadi thinks thats a bashism
17:00mushroomtwo has quit IRC
17:03
<alkisg>
Gadi, type function || echo "not defined" ?
17:03
<Gadi>
takes too long
17:04
<alkisg>
It's just one word, type.. :D
17:04
<Gadi>
shell will search built in functions and then PATH
17:04
:P
17:04
<alkisg>
Ah
17:04
<Gadi>
I suppose one could do: PATH="" function
17:04
:)
17:06
wow - thats a hack
17:06
<stgraber>
looks like one
17:08
<Gadi>
PATH="" boolean_is_true True 2>/dev/null|| echo Need to source ltsp-common-functions
17:08
lol
17:09
<stgraber>
Gadi: declare -F | grep boolean_is_true
17:09
<Gadi>
mine's faster
17:09
:)
17:09
<alkisg>
sh: declare: not found
17:09
<stgraber>
alkisg: yeah, it's a bashism
17:10
<alkisg>
bash -c "declare -F..." :P
17:11
<stgraber>
that's really really hackish and won't work unless boolean_is_true is exported ;)
17:12
<johnny>
i'm sure somebody has searched this problem before
17:13
<stgraber>
johnny: not according to google, only answers are slow ones ;) (usually involves parsing type)
17:13
<Gadi>
pushed
17:14
now I will add alkisg's check for LTSP_CONFIG
17:14tstafford has joined #ltsp
17:14
<stgraber>
.99 is gonna rock ;)
17:18mushroomblue has joined #ltsp
17:23mushroomblue has joined #ltsp
17:31bobby_C has quit IRC
17:38alkisg has quit IRC
17:41
<Gadi>
so, if we run cdpinger out of udev, do we still need all the cdpinger code in ltsp-setup?
17:42
ie. configure_localdev() (now in ltsp-init-common)?
17:42
lots of slow pgreps in there
17:42
<stgraber>
we shouldn't have to
17:43
<Gadi>
how about making the ltspfs_token and starting ltspfsd?
17:44
seems those are done in udev, too
17:44
we should be safe to omit configure_localdev || true
17:45
<stgraber>
we can probably drop it and test
17:46
<Gadi>
cool - something for tomorrow
17:46
gnight, all
17:47Gadi has left #ltsp
17:48pasmen has joined #ltsp
18:11vvinet has joined #ltsp
18:14artista_frustrad has quit IRC
18:38litlebuda has quit IRC
19:03rhodan has joined #ltsp
19:05pasmen has quit IRC
19:07otavio has quit IRC
19:10shogunx has joined #ltsp
19:10
<shogunx>
hi guys
19:11
any way to define on a client by client basis the amount of nbs swap in /etc/ltsp/nbdswapd.conf?
19:12
to conserve disk space, i want one client to have a larger swap file than the others, as it houses a madwifi based access point as well.
19:13
<johnny>
i don't think so
19:13
not yet anyways
19:13
stgraber, do you know anything about that?
19:16otavio has joined #ltsp
19:17pasmen has joined #ltsp
19:19rhodan_ has quit IRC
19:21try2free has joined #ltsp
19:33pasmen has quit IRC
19:38try2free has left #ltsp
19:43japerry has joined #ltsp
19:51pasmen has joined #ltsp
19:51shogunx has quit IRC
19:51Selveste1___ has quit IRC
19:52Selveste1___ has joined #ltsp
20:05shogunx has joined #ltsp
20:07pasmen has quit IRC
20:17Ahmuck-Sr has joined #ltsp
20:18Ahmuck-Sr has left #ltsp
20:18try2free has joined #ltsp
20:19pasmen has joined #ltsp
20:35pasmen has quit IRC
20:35Selveste1___ has quit IRC
20:35Selveste1___ has joined #ltsp
20:36rhodan_ has joined #ltsp
20:41try2free has left #ltsp
20:50rhodan has quit IRC
21:21vagrantc has joined #ltsp
21:23ball has joined #ltsp
21:26
<ball>
How hard is it to find a reliable company to roll out (and perhaps support) ltsp?
21:26
(I'm in Illinois, USA)
21:31
<sbalneav>
ball: I know that disklessworkstations.com provides thin clients and support.
21:32
Other than that, any firm in your area that can do GNU/Linux support should be able to do LTSP support
21:32* ball ponders
21:33
<ball>
Finding a local firm that does Linux may be a challenge.
21:33
<stgraber>
ball: how large a deployment are we talking about ? (My company does large-scale LTSP if you need several application servers and a good number of thin clients or things like load-balancing/high availability)
21:33
yeah, I know, self publicity ... ;)
21:34
<ball>
Probably ten desktops to start with, a few more later. Two servers for availability.
21:34
Pondering ltsp Vs. Sun Ray
21:35
(Vs. ten new fat clients)
21:36
<stgraber>
hmm, ok, then you're probably better with more regular LTSP, at least at the beginning. Diskless would probably be a good company as they offer both the hardware and the support for that.
21:36
well, I guess for LTSP vs Sun Ray, the main question is what you'll be doing with it
21:36
<ball>
stgraber: can they do two servers lashed together, with fail-over?
21:36
stgraber: general office stuff mostly.
21:36
Current desktop image has Firefox, OpenOffice.org, the GIMP and a few other little things.
21:37
s/Firefox/SeaMonkey/
21:37
<stgraber>
last I checked Sun Ray, was using a bitmap (VNC-like) protocol, it has its advantages (suspending/resuming sessions being the biggest one) but also a lot of issues (as in, anything multimedia will completely fail)
21:37
<ball>
I'm told multimedia works fine on Sun Ray 2.
21:38
<stgraber>
hmm, maybe they improved that then, would be great, because last I tried, going to youtube was almost impossible and would kill the server with only 2-3 users (not even speaking of the bandwidth usage and poor quality)
21:38
<ball>
...but we're a small site and Sun just refer us to resellers who don't seem up to much.
21:39
<stgraber>
main advantage of LTSP is of course that it's Linux based, using standard hardware and that you can easily find someone to customize it to your needs
21:39
<ball>
"easily" if we were in a large city perhaps.
21:40
Standard hardware and open source are both big attractions though.
21:40
<stgraber>
well, it's still easier to find someone to customize a Linux distro than a closed source platform.
21:40
<ball>
Sun Ray Server Software is closed, so that's a fail waiting to happen.
21:41
I've come close to rolling out more Windows machines just because I know we'll be able to get support for those.
21:41
I know it's the wrong solution, but it's almost the path of least resistance.
21:43
<stgraber>
getting back to your fail-over question, that's typically something that's more ltsp-cluster specific (load-balanced application servers, sending everyone to the single working one in case of failure, ...) and the actual high availability is the kind of stuff Revolution Linux does (the company I work with and who developped ltsp-cluster)
21:43
<ball>
I hear that's php-based. I'm not keen on php.
21:44
<stgraber>
the control interface is PHP yes with a PostgreSQL backend. Load-balancer and other daemons we use are all python though.
21:44
<ball>
Hmm...
21:44
<stgraber>
the PHP choice for the Web UI was mostly made because it's the easiest to install and all packages are supported as part of Ubuntu's LTS (kind of important as we only work with Ubuntu ;))
21:45
<ball>
Why use a web UI though?
21:46
<stgraber>
when you have several hundreds or thousands of thin clients as our customers sometimes have, having them displayed in a tree-structure with configuration inheriting is really useful.
21:46
<ball>
(Is the Web UI mandatory?)
21:47
<stgraber>
the control center (web service + DB) is required as it's what's returning the configuration to the thin clients at boot time (based on their physical location, hardware or other defined groups)
21:47
you can directly change the configuration in the database if you prefer though
21:48
but the php code still needs to be there as it's called by the thin client to get its generated lts.conf and send/receive information when being used (such as logging who's logged on which thin client, checking for already opened session, sending hardware inventory, ...)
21:48rhodan_ has quit IRC
21:49
<stgraber>
oh, btw, all that is opensource ;)
21:49
<ball>
Hmm... I can learn to ignore its webification provided it's reliable enough.
21:49
(and I can get support for it)
21:50
<stgraber>
well, considering we have it deployed at a few customers with several thousands (biggest is 5000) thin clients and over 50 applications servers in a HA setup, it's reliable ;)
21:52
<ball>
That's encouraging.
21:53
I'll need to rewire our network, but that shouldn't be much of a chore.
21:54
stgraber: I'll have to look around for a company that supports smaller sites.
21:56
<stgraber>
ball: sure, if you need something bigger or full high-availability setup at some point, please do no hesitate to poke me.
21:57
<ball>
stgraber: I need failover at a minimum.
21:57
Two servers. I'd love to go with dual-socket, but may have to resort to two quad core machines.
21:57
(quad core, 1-way)
21:58
<stgraber>
if you have powerful enough thin clients, running firefox as localapp will help a load decreasing the load on your network and on the server's CPU.
21:58
<ball>
...but then the terminal LAN would have to be routed to the Internet
21:59
<stgraber>
we were at FETC last week, running between 40 and 60 Asus thin clients on a single Q6600 with 8GB of RAM. But that was with Firefox and Blender running as localapp.
21:59* ball doesn't know what "Blender" is.
21:59
<stgraber>
right, you need your thin clients to have internet access then, guess it depends on your security standards then
21:59
ball: 3D design application
21:59
<ball>
ok
22:00
stgraber: wouldn't it also confuse the users if Firefox were run on the terminal? File -> Save... where's my home folder?
22:00
<stgraber>
ok, looks like after breaking the Ubuntu builders, fixing them, I finally have a new LTSP snapshot in my PPA ! alkisg will be happy. Also fixed a good old bug where the initramfs wouldn't get rebuilt after updating the package !
22:00
ball: home folder is mounted over sshfs
22:01
ball: so the user has access to his documents and firefox shares its configuration with the application server
22:02
<ball>
I only have a few users, so I think I'll keep it on the server if possible.
22:02
Need to find servers, switches and UPS.
22:02
...and software.
22:04
<vagrantc>
hrm.
22:05artista_frustrad has joined #ltsp
22:06
<ball>
Perhaps HP DL160s or something
22:07
<stgraber>
vagrantc: good catch (screen-x-common)
22:07map7 has quit IRC
22:08
<vagrantc>
stgraber: i'm working on some others :)
22:09
sometimes, i feel that is my role...
22:13
<ball>
I have to go. Back later.
22:14ball has quit IRC
22:14
<vagrantc>
we can probably ditch all the 'if [ "True" != "$LTSP_CONFIG" ]; then' stuff, as it's implemented directly in ltsp_config now ...
22:15
though it'd be at a slight performance cost, but overall simpler maintenance.
22:16
<stgraber>
if it makes it cleaner and doesn't add a lot of external calls (grep/awk/sed/... as we saw in my bootcharts), I +1
22:16
<vagrantc>
it would add some additional sourcing calls, but not grep/awk/sed
22:16
so a few extra filesystem calls
22:17
<stgraber>
right, sourcing a file is ok I guess
22:18
<vagrantc>
overall stuff's looking nice.
22:20
stgraber: i poked the debian nbd maintainer (maybe upstream now?), Wouter Verhelst, about the nbd-proxy stuff, and he was going to ask around a bit maybe with the nbd kernel developer about how to handle that stuff better.
22:22
<stgraber>
vagrantc: ok, my view on how things should be isn't exactly clear at the moment, the nbd kernel part is doing really weird things, especially that -persist code ;)
22:22
vagrantc: so I'm not yet sure if the solution is to port what we fixed with nbd-proxy to the kernel module or to move some stuff from the kernel module to userspace
22:23
anyway, it wasn't likely we'd have anything fixed in the kernel before the LTS release, so fixing in the userspace was the easiest :)
22:23
<vagrantc>
and my looking at the code was useless, but i'd figure i'd rope in the few folks who i knew might make sense of the code :)
22:24
<stgraber>
right, we still have one bug to catch at the moment (the proxy returning an extra byte of information when we have two pending requests at the moment of a crash) but I hope to have that fixed tomorrow and so will be able to finally tag .99 ;)
22:25artista_frustrad has quit IRC
22:25
<stgraber>
(and yeah, even though the bug seems pretty hard to reproduce, it's happening way too often to just ignore it ;))
22:26
<shogunx>
anybody in need of servers on the cheap check this out: http://www.weirdstuff.com/cgi-bin/item/62025
22:26
i have bought and used several of these no issues
22:26
<vagrantc>
and wrap up a few bugfixes from 5.1.99 and then call it 5.2.0 :)
22:26
<stgraber>
that's the plan
22:27
I hope to have 5.2 out either next week or the week after and have it rock solid ;)
22:29
<vagrantc>
seems like someone introduced some new translation strings or changed something...
22:30
<stgraber>
probably the fat client stuff
22:31
<vagrantc>
i was hoping to have updated translations by 5.2.0 timeframe...
22:31
also, i didn't want to futz with the ALTLinux translations at all, since we haven't seen any activity in over a year from that distro.
22:31
<stgraber>
Is it breaking a lot of strings ?
22:32
<vagrantc>
close to two years
22:32
they introduced a TON of strings.
22:33
i *think* some of which are fuzzied variations on some of the strings we just changed... hmmmm.
22:35
ah yes, at least two of them are copies.
22:35
<stgraber>
vagrantc: I just merged fr.po with an updated template, I don't have any fuzzy string there
22:35
<vagrantc>
stgraber: the weird thing is, i didn't have any updated strings when i made the commits ...
22:36
stgraber: but just look at the .po file ... probably roughtly 25-30% are ALTLinux specific.
22:37
<stgraber>
that's really weird, doing a update-po gets me some fuzzy strings for pretty much all languages except french ;)
22:37alexqwesa_ has joined #ltsp
22:37
<stgraber>
german and spanish are ok too
22:38
stgraber@castiana:~/data/code/ltsp-trunk/po$ grep fuzzy * | awk '{print $1}' | cut -d : -f1 | sort | uniq -c | grep -v pot 21 ca.po 21 cs.po 21 fi.po 21 pl.po 25 pt_BR.po 21 vi.po
22:38
... so much for irssi copy/pasting
22:38
<vagrantc>
i'm going through and updating some of the altlinux and gentoo plugins that just have pointlessly fuzzied strings
22:38
<stgraber>
http://pastebin.com/f7421f50b
22:39
number of fuzzy string after a template update + po update for languages affected
22:39
doesn't that match the languages where you haven't received new translations yet ?
22:39
(haven't looked at the commit log, just wondering)
22:41morfic has joined #ltsp
22:41
<vagrantc>
stgraber: looks about right, yeah.
22:41
pt_BR is so far behind.
22:46alexqwesa has quit IRC
22:55
<vagrantc>
stgraber: oh no... i just remembered that the debian french translation team contacted me the very next day saying they were working on it...
22:56pasmen has joined #ltsp
23:09ball has joined #ltsp
23:12pasmen has quit IRC
23:15
<vagrantc>
ok, those last commits reduced the overall translated strings by 4-5.
23:15
since some were merely out of sync with the various distro plugins.
23:20
heh. of course, no commits on ltsp-docs for weeks, i release a new version, and the commits roll in :)
23:21pasmen has joined #ltsp
23:38artista_frustrad has joined #ltsp
23:40pasmen has quit IRC
23:47shogunx has quit IRC
23:48artista_frustrad has quit IRC
23:57artista_frustrad has joined #ltsp