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


Channel log from 3 March 2007   (all times are UTC)

00:14
<astroboy>
Gentlemen, I let ltsp-build-client run while doing other things, and on return .....
00:14
"info: LTSP client installation completed successfully"
00:17
<vagrantc>
that's more what i'm used to.
00:21
<astroboy>
WS01 is searching for server (DHCP) .. so I have to get dhcpd.conf right ...
00:33nofxx has quit IRC
00:39
<vagrantc>
astroboy: example should be in /usr/share/doc/ltsp-server/examples
00:41
astroboy: that example should handle both etherboot and PXE
00:41* vagrantc sleeps
00:41vagrantc has quit IRC
00:42
<astroboy>
I've used it as a template for eth1 but I'm not sure about eth0 ..
00:50uwe has quit IRC
00:51
<ltsppbot>
"astroboy" pasted "/etc/dhcp3/dhcpd.conf .. for ltsp5" (25 lines) at http://pastebot.ltsp.org/7
00:52
<astroboy>
Anyone care to comment ?
00:57g333k_laptop has joined #ltsp
01:03g333k_laptop has quit IRC
01:22uwe has joined #ltsp
02:16vanya has joined #ltsp
02:17astroboy has quit IRC
02:19Egyptian[Home] has quit IRC
02:21Egyptian[Home] has joined #ltsp
03:56nsar has joined #ltsp
03:56
<nsar>
hello
04:04
how can i download the latest version of ltsp? i look in the site but the download section is confusing me
04:13
<uwe>
nsar, do you want to install muekow ltsp (ltsp5)? if yes, i havent tried installing it yet, but apparently its still distro dependent
04:13
what distro are you using?
04:13
i know it works on debian etch
04:14
<nsar>
fedora core
04:15
6
04:15
you mean is not yet ready to use it?
04:16
<uwe>
um, i dont see any straight forward for installing it on fedora, i suppose you can use the tool supplied by ltsp (ltsp utilities) and itll get pretty up-to-date packs
04:17
<nsar>
ok thanks
04:17
<uwe>
nsar, its not that, its that the "new" way of installing ltsp is slightly diffrent, and needs efforts from the distro side
04:17
<nsar>
i see
04:18
<uwe>
and the maturity of it will depend on the maturity of the host OS/distro ... i think
04:18
<bip>
it is very well supporete d by debian and ubuntu
04:18
<nsar>
i am comming back just a moment
04:18nsar has quit IRC
04:18
<bip>
i think suse and fedora r slightly behind right now
04:19
<uwe>
shows how good/bad rpm package management is
04:19
shows how good/bad rpm/deb package management is
04:20
<bip>
or how lil interest eexist in ltsp from big players maybe
04:20
or how that just play wait and see ...
04:20
and leverage solutions developed by somebody else ;-)
04:20
<uwe>
em, heh... i think nsar needed this -> http://wiki.ltsp.org/twiki/bin/view/Ltsp/MueKow#Fedora_Core_3_Testing
04:21
<bip>
he might
04:21
r just google for fedora ltsp
04:21
<uwe>
he wanted the "latest"
04:21
<bip>
maybe somebody tried it with a more redcent distro
04:21
yup well i sorta have the same problem
04:22
i wanna doo a test install with latest suse ...
04:22
but i have found only one howto and it isn t very clear ...
04:23
http://www.novell.com/coolsolutions/feature/17942.html
04:25
<uwe>
i have a strong feeling that the packages provided in the classical way are just as new as those in mue
04:25
i mean the init scripts and such stuff
04:26
<bip>
that howto isn t about muekow
04:26
it stil using 4.x
04:26
s
04:27
<uwe>
well, i was trying to say that 4.x and mue are using the same stuff when it comes to the contribution of ltsp
04:27
the diffrence is what the distro ships for stuff that is common between ltsp and a normal distro
04:28
im not sure not authentic
04:28
nor*
04:30
<bip>
welll ltsp integartion into distros is still lacking imho
04:30
fo example i really like what debian and ubuntu r doing but they both lack something
04:31
for exampple i really like the possibility of picking LTSP as option if i install from a xubuntu edgy alternate cd that s almot pug and play
04:32
but ubutu support fro fat diskless option is still lacking
04:32
and ldap/AD integration too
04:39sepski has joined #ltsp
04:45
<uwe>
yay ... that would be great
04:53
<bip>
novell has some great AD integration tools ...
04:53
and now that they r becomin partner with ms they will get even better at that
05:05FernandoMM has quit IRC
05:43spectra has joined #ltsp
06:00FernandoMM has joined #ltsp
06:12Alkhouli has quit IRC
06:18freemind has joined #ltsp
06:32sepski has quit IRC
06:33cliebow_ has joined #ltsp
06:34
<cliebow_>
moquist: have you migrated ldap to feisty? mine seems balky..as in doesnt work..
06:35cliebow_ has quit IRC
06:37cliebow_ has joined #ltsp
06:38cliebow_ has quit IRC
06:42chupacabra has quit IRC
06:53freemind has quit IRC
06:59FernandoMM has quit IRC
07:01bip has quit IRC
07:01cliebow_ has joined #ltsp
07:09
<moquist>
cliebow_: nope. we're running it on Dapper and Edgy.
07:16
<cliebow>
moquist, they add a ppolicy.schema amongst other things..i dont see any errors to speak of..i dont see my database either..
07:16freemindx has joined #ltsp
07:17
<cliebow>
i frigged around for quite a while yesterday..
07:36Stevecar has joined #ltsp
07:47oldwolf has joined #ltsp
08:07vanya has quit IRC
08:09
<jammcq>
ogra: ping
08:10cliebow_ has quit IRC
08:10sbalneav has joined #ltsp
08:10
<jammcq>
Scotty !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
08:10
<sbalneav>
Morning!
08:10
Snausages
08:10
!seen ogra
08:10
<ltspbot>
sbalneav: ogra was last seen in #ltsp 22 hours, 51 minutes, and 51 seconds ago: <ogra> right
08:10Egyptian[Home1 has joined #ltsp
08:12Egyptian[Home] has quit IRC
08:14
<bronze>
42
08:19oldwolf has quit IRC
08:26cliebow_ has joined #ltsp
08:50JMBarbosa has joined #ltsp
08:53jammcq has quit IRC
08:58sbalneav has quit IRC
09:00JMBarbosa has joined #ltsp
09:02JMBarbosa has quit IRC
09:15Cipher_ has joined #ltsp
09:19sbalneav has joined #ltsp
09:19jammcq has joined #ltsp
09:19
<jammcq>
ogra: ping
09:43uwe has quit IRC
09:45latif has joined #ltsp
09:46
<latif>
hello can u help me regarding dhcp?
09:47
i am used mandrake 8.2
09:47
install but the dhcp its problem
09:47
only dhcp?
09:48cliebow has quit IRC
09:55
<ltsppbot>
"latif" pasted "dhcp probelm" (96 lines) at http://pastebot.ltsp.org/8
09:55
<sbalneav>
latif: WHat problem are you having?
09:56
<latif>
dhcp?
09:56
i am using mandrake 8.2
09:56
<sbalneav>
yeah, but WHAT
09:56
<latif>
the prob only dhcp
09:56
<sbalneav>
Does it start? Does it give an error? etc.
09:56
You have to tell us what's WRONG
09:57
<latif>
everting its ok...only part of DHCP yes....yes...no?
09:57
i need to know
09:57
<sbalneav>
No
09:57
You need to know what>
09:57
?
09:57
<latif>
mandrake have the kernel....and ltsp also have the kernel?...
09:57
<sbalneav>
Do you get an error when you start dhcp?
09:57
<latif>
which kernel i need 2 used
09:58
<sbalneav>
Which version of LTSP are you using?
09:58
<latif>
yes get erorrr only dhcp part only
09:58
3.0
09:58
<sbalneav>
What error do you get?
09:59
<latif>
went i cek dhcp start no problem
09:59
all ok only
09:59KeepWalking_ has joined #ltsp
09:59
<latif>
went run ltspadmin dhcp not running
09:59
<KeepWalking_>
buen dia
10:00
i cant get gdm working, its like remote login its not enable...
10:00
<jammcq>
KeepWalking_: bom dia
10:00
<sbalneav>
latif: check and see if it's running with ps
10:01
<KeepWalking_>
jammcq, can you help me about gdm please?
10:01
<latif>
if u thing i must do the grub before run back dhcp?
10:01
<jammcq>
umm, maybe
10:01
<KeepWalking_>
jammcq i get X working
10:01
bu gdm never load
10:01
<sbalneav>
latif: First, check and see if gdm is running.
10:02
err dhcpd
10:02
<latif>
ok
10:02
tq
10:02
<jammcq>
KeepWalking_: have you followed this document: http://wiki.ltsp.org/twiki/bin/view/Ltsp/Troubleshooting-GrayScreen
10:02
<KeepWalking_>
letme see
10:02J45p3r has joined #ltsp
10:11latif has quit IRC
10:16bronze has quit IRC
10:17bronze has joined #ltsp
10:19BlueDragon has joined #ltsp
10:21Stevecar has quit IRC
10:26KeepWalking has joined #ltsp
10:26
<KeepWalking>
jammcq
10:27
i have already checked all, but everything looks ok
10:27
only
10:27
that i get in ltspcfg
10:27
gdm installed, configured, but no running using gdm
10:27
and i have gdm running
10:27
im seeing on server
10:28KeepWalking_ has quit IRC
10:28
<cliebow_>
KeepWalking, can you..Balls!
10:28
<KeepWalking>
cliebow_ what?
10:29
<cliebow_>
Oh there you are
10:29
netstat -anp|grep ":177"
10:29
is it listening?
10:29
<KeepWalking>
no, it isnt
10:30
<cliebow_>
cant trust ltspcfg for xdmcp
10:30
what distro?
10:30
<KeepWalking>
cliebow_ oh ok
10:30
cliebow_ edubuntu
10:30
<cliebow_>
there is a gconf-editor wherer you can turn it on
10:31
<KeepWalking>
cliebow_ ok, let me see please
10:31
<cliebow_>
balls i cant find it either
10:33
be sure in gdm.conf enable=true and port=177commented are un
10:34
<KeepWalking>
yes
10:34
its ok
10:34
but
10:34
nothing on port 177
10:34
<cliebow_>
was it ok already?
10:34
cause if you just did it you need to restart gdm
10:34
<KeepWalking>
in gdm.conf enable is true and port is 177
10:34
yes, i have
10:35
<cliebow_>
just now uncommented?
10:35
<KeepWalking>
uncomented, and restarted gdm
10:35
<cliebow_>
but not listening?
10:35
<KeepWalking>
i do netstat | grep 177
10:35
and nothing
10:36
<cliebow_>
hang a sec
10:36
<KeepWalking>
sure
10:39
<cliebow_>
in redhat there is a line in Xsession that needs uncommented..like * #any client can get a login
10:41
i yes..sudo gdmsetup should have a remote tab..
10:42
choose how you want remote to look
10:42
......drum rolll..............
10:51BlueDragon has quit IRC
10:53psyco-obiwan has quit IRC
10:57
<sbalneav>
mistik1: ping
10:58psyco-obiwan has joined #ltsp
11:00Alkhouli has joined #ltsp
11:05
<cliebow_>
KeepWalking, what happened??
11:12BlueDragon has joined #ltsp
11:28cliebow_ has quit IRC
11:36cliebow has joined #ltsp
11:50cliebow_ has joined #ltsp
11:55KeepWalking has quit IRC
12:01Topslack has quit IRC
12:14Topslack has joined #ltsp
12:14anahuac has joined #ltsp
12:15anahuac has left #ltsp
12:15cliebow has quit IRC
12:34chupacabra has joined #ltsp
12:43spackle has joined #ltsp
12:46mathesis has joined #ltsp
12:56cliebow has joined #ltsp
13:24MaCaDe has joined #ltsp
13:34Cipher_ has quit IRC
13:44nsar has joined #ltsp
13:45
<nsar>
hello
13:47
inside the Xorg command is included a parameter -nolisten tcp that for this reason last time i had tried to install ltsp that was my big headache in which file is that parameter to remove it?, i am using fedora core 6
14:01sepski has joined #ltsp
14:05nsar has quit IRC
14:19mistik1 has quit IRC
14:24mistik1 has joined #ltsp
14:58Alkhouli has quit IRC
15:08Stevecar has joined #ltsp
15:12nick125_lappy has joined #ltsp
15:12
<nick125_lappy>
afternoon ltspers
15:33sepski has quit IRC
15:37cliebow has quit IRC
15:59astroboy has joined #ltsp
16:18astroboy has quit IRC
16:22BlueDragon has quit IRC
17:15nick125_lappy has quit IRC
17:27
<jammcq>
hey, anybody have access to a Fedora core 6 box ?
17:27
I just need to know if the 'tempfile' command exists on fc6
17:55
<spackle>
I thought xdpyinfo named the X server in use
17:59
<sbalneav>
Which, you mean what driver's being used?
17:59
<spackle>
yes
17:59
like s3 or svga
17:59
<sbalneav>
No, it doesn't as far as I know. It shouldn't care.
18:00
<jammcq>
well, technically speaking, xdpyinfo doesn't care about anything. it's the user who's running xdpyinfo who cares, and generally, they'd prolly like to know the name of the driver
18:01
<spackle>
is checking the x configuration file in a terminal the only way to know then - unless you catch it on startup
18:01
<sbalneav>
No, the user doesn't care either. The ADMINISTRATOR's the only one who cares :)
18:01
<jammcq>
the "PERSON" running xdpyinfo is the USER of xdpyinfo
18:01
call him an "ADMINISTRATOR" if you want to
18:01
<sbalneav>
:)
18:04* spackle rolls eyes
18:04
<sbalneav>
spackle: I guess the 50 dollar question of the day is, why do you need to know? Are you trying to debug a problem. Contextless questions will get you silly answers :)
18:05
<spackle>
oops, sorry, I know the chip in the term is an S3, but the detection is set to auto, and I want to know if there is a btter choice
18:07
<sbalneav>
Well, the only two s3 choices I know of are either s3 or s3virge
18:07
<spackle>
well, framebuffer is another choice
18:07ogra has quit IRC
18:08
<sbalneav>
I'd expect you'll probably want to avoid that one. You'll get little or no accel that way. Not that there's a ton of accel in s3 anyway.
18:08
<spackle>
;)
18:08
<sbalneav>
could always try vesa as well.
18:08
<spackle>
is there any way to get sound from JAVA apps on a client?
18:09
where app != running *on* client
18:10
<sbalneav>
yeah. Grab Sun, slap 'em around a little bit, and tell 'em to quit writing to /dev/dsp and use a real audio stack. I suppose now that java's been free'd, the community should look at it.
18:11
Java's currently a problem.
18:11
<spackle>
those Ba******
18:11
K, just wanted to be sure there wasn't an ingenious workaround
18:12
<sbalneav>
Now, with the new stuff that ogra's done with pulseaudio, and faking out an alsa device, it MAY work better on Ubuntu feisty
18:13
but ultimately, the way to make network audio work SEAMLESSLY everywhere, is to fix anything that still writes to /dev/dsp, and convert it to use gstreamer.
18:13
kde's moving to gstreamer, and anything that currently uses gstreamer works flawlessly.
18:13
So, totem plays movies fine, etc.
18:14
<spackle>
excellent. Is that in the latest edubuntu - edgy I think
18:14
?
18:14
<sbalneav>
no, edgy's the current
18:14
feisty's the next one.
18:14
that's in beta now.
18:14
<spackle>
6.10
18:14
or later?
18:14
<sbalneav>
7.04
18:14
<spackle>
OK.
18:15
thanks sbalneav
18:15
<sbalneav>
NP
18:15
<spackle>
I'm off to make the donuts.... later
18:15
<jammcq>
mmmm, donuts
18:15
<mistik1>
hey guys
18:16
<jammcq>
hey mistik1
18:16ogra has joined #ltsp
18:17
<mistik1>
hola ogra
18:17
<sbalneav>
ogra!!
18:17
hey mistik1
18:18
<mistik1>
hey sbalneav
18:18
<jammcq>
mistik1: remember in Maine at BTS, we whipped up a command to do a commandline paste to the pastebot ?
18:18
<mistik1>
yea
18:18
<jammcq>
any chance you have it?
18:18
<mistik1>
nope, lost in the shuffle
18:19
<jammcq>
any chance you remember what we called it?
18:19
I may still have it
18:19
<mistik1>
nopaste maybe
18:19
hmm
18:20
Its easy to reproduce with www-perl though
18:22Stevecar has left #ltsp
18:22
<mistik1>
jammcq: found a ruby one,sec
18:23
<jammcq>
ahh,,, paste2ltsp
18:23
<mistik1>
hehe
18:23
that may be what you named it ;)
18:23
<jammcq>
there's a file in an old backup called that
18:23
<mistik1>
prolly it
18:24
or a first draft
18:25
<ltsppbot>
"jam" pasted "STDIN: #!/usr/bin/perl -w ..." (252 lines) at http://pastebot.ltsp.org/9
18:26
<jammcq>
:)
18:26
I used paste2ltsp to paste "paste2ltsp" to the LTSP pastebot
18:28
<mistik1>
hehe
18:28
cool
18:28
<ltsppbot>
"jam" pasted "STDIN: #!/usr/bin/perl -w ..." (253 lines) at http://pastebot.ltsp.org/10
18:29
<mistik1>
jammcq: that's the latest one alright
18:29
Thanks
18:29
<jammcq>
it's got a slight problem
18:29
<mistik1>
what's that?
18:29
<jammcq>
it's not parsing the result correctly. it says 'your paste was Terribly unsuccessful'
18:29
when, indeed, it succeeded
18:30
<mistik1>
ok
18:30
You have the source Luke
18:30
<jammcq>
yep, and the will
18:32
<ltsppbot>
"jam" pasted "STDIN: #!/usr/bin/perl -w ..." (257 lines) at http://pastebot.ltsp.org/11
18:34
"jam" pasted "STDIN: this is a test ..." (1 line) at http://pastebot.ltsp.org/12
18:35
"jam" pasted "STDIN: this is a test ..." (1 line) at http://pastebot.ltsp.org/13
18:36
"jam" pasted "STDIN: this is a test ..." (1 line) at http://pastebot.ltsp.org/14
18:42
"root" pasted "STDIN: this is a test ..." (1 line) at http://pastebot.ltsp.org/15
18:42
<jammcq>
mistik1: all fixed. pardon my flooding :)
18:42robbie has joined #ltsp
18:43
<robbie>
mornings
18:43
<mistik1>
okie dokie
18:43
<jammcq>
robbie: hey robbie
18:43
how ya doing?
18:43
<mistik1>
send me a copy
18:44
<jammcq>
sure
18:44
where?
18:44
<mistik1>
just paste it ;)
18:44
I bet I know what it was though
18:44
<jammcq>
k, coming right up
18:45
before I post it, tell me what you think it was
18:45
<mistik1>
the url string return is not longer .../paste
18:45
<ltsppbot>
"root" pasted "STDIN: #!/usr/bin/perl -w ..." (251 lines) at http://pastebot.ltsp.org/17
18:45
<jammcq>
well, it doesn't have the port number in the url anymore
18:45
AND, it doesn't have "/paste'
18:45
<mistik1>
noted that also
18:46
<jammcq>
either of those broke it
18:46
<mistik1>
thanks mod_proxy ;)
18:46
<jammcq>
anyway, the latest is on the pastebot, #17
18:46
ah
18:46
mod_proxy took out our /paste ?
18:46
<mistik1>
I think so
18:46
also the port
18:46
<jammcq>
well, the port is now :80, so it's not needed
18:47
<mistik1>
yea
18:47
<jammcq>
the code can still handle :port, but it's not looking for /paste
18:47
<mistik1>
right
18:49
is the source not wonderfull
18:51
<jammcq>
like socks in a box
19:23
<robbie>
hi guys
19:24
you guys getting those storms ?
19:24
<jammcq>
not so much
19:24
had horrible weather on thursdya
19:24cliebow has joined #ltsp
19:25
<sbalneav>
Hey robbie!
19:26
How's the kids?
19:26
<robbie>
hi scotty
19:26
we now have 1.93
19:26
<sbalneav>
:)
19:26
<jammcq>
what's the due date?
19:26
<robbie>
19 days till "Number 2" arrives :)
19:26
23rd
19:26
<sbalneav>
Got any photos? I enjoyed your last photo album from a year ago.
19:26
<robbie>
heh, one sec
19:27
my dsl is rate limited at the moment so it may be slow
19:27
http://rotapile.ods.org/gallery/0702/IMGP3428.JPG
19:27
an di took him sand toboganing
19:27
http://rotapile.ods.org/gallery/0701/IMGP3404.JPG
19:28
<sbalneav>
Cute!
19:28
<robbie>
http://rotapile.ods.org/gallery/0701/IMGP3413.JPG
19:28
day out with sister and friends
19:28
and nephew and niece
19:29
sbalneav: actualy, I have a photo that a man from a snowy country would appreciate :)
19:29
<jammcq>
great pics
19:29
<sbalneav>
tsk tsk. EVERYONE's gotta be wearing a life jacket.
19:29
<robbie>
just kids
19:29
lots of power boats near there
19:29
and water is silty, babies sink
19:30
http://rotapile.ods.org/gallery/0610-tasmania/0610%20042.jpg
19:30
<sbalneav>
LOL
19:30
Nice aim :)
19:31
<robbie>
my friend said it was great penismanship
19:32
my old man has built model a steam railway
19:32
http://rotapile.ods.org/gallery/0701/038.jpg
19:33
i love working on windows 2000 servers
19:33
<sbalneav>
Wow! Electric?
19:33
what voltage?
19:33
<robbie>
2 ide drives on same bus set up with software mirrored dyanmic disks
19:34
sbalneav: all steam except for that particular one which is 4hp petrol/hydraulic
19:34
<sbalneav>
Steam! Make his own engine from scratch?
19:34
<robbie>
bought engines
19:34
<sbalneav>
cool
19:35
<robbie>
built 650m track
19:35
heh
19:35
very popular with kids in street
19:35
<sbalneav>
I can imagine
19:37
<robbie>
if you can handlye the slow load there are quite a few photos of the construction etc here
19:37
http://rotapile.ods.org/gallery/0608-mumdad/
19:38
http://rotapile.ods.org/gallery/0608-mumdad/100_0276.JPG
19:38
thats one of the steam engines
19:39
<sbalneav>
Who's telescope?
19:39
<robbie>
mine
19:41
<sbalneav>
nice
19:41
Schmidt-Cass?
19:41
<robbie>
i believe so
19:41
that is where we got it from, the dome is still to arrive
19:42
and the telescope is stil packed up in the shed
19:42
its a 25 year old meade
19:47daniQ has joined #ltsp
19:52cliebow has quit IRC
20:03
<cliebow_>
anyone using 2.6.20 for a kernel?
20:06daniQ has quit IRC
20:32cliebow has joined #ltsp
20:34bronze has quit IRC
20:34bronze has joined #ltsp
20:40MaCaDe has quit IRC
20:53cliebow has quit IRC
21:07ramsys has joined #ltsp
21:08spectra has quit IRC
21:11
<ramsys>
how do i copy all comments from channel to my local disk
21:12J45p3r has quit IRC
21:19robbie has quit IRC
21:26vagrantc has joined #ltsp
21:42Bhaskar has joined #ltsp
21:48
<jammcq>
vagrantc: hey
21:48
<vagrantc>
jammcq: hi
21:50
<jammcq>
vagrantc: I built an etch machine, and loaded up ltsp-5 in it
21:50
found a few little "issues"
21:50
<vagrantc>
please share your issues and i will address them to the best of my ability :)
21:51
<jammcq>
ISC changed how next-server works. the Ubuntu guys patched around it
21:51
apparently, Debian hasn't patched around it.
21:51otavio has joined #ltsp
21:51
<otavio>
Hi :-)
21:52
<jammcq>
otavio: HEY
21:52
<vagrantc>
otavio: welcome back, long time no see :)
21:52
<otavio>
jammcq: :-)
21:52
<jammcq>
wow, long time, no see
21:52
<otavio>
vagrantc: yes. :-)
21:52
busy life ;-)
21:52
I've been away from LTSP hacking lately
21:52
<jammcq>
yes, we've missed you
21:52
<otavio>
I hope I can come back in few weeks ;-)
21:52
jammcq: thank you :-)
21:52
<vagrantc>
jammcq: i haven't run into that problem myself yet ...
21:52
<jammcq>
cool. we've been pushing forward, having fun
21:53
<otavio>
jammcq: how is ltsp5 going?
21:53
<jammcq>
vagrantc: the easy fix is to add 'next-server 192.168.0.1;' to dhcpd.conf
21:53
but it's not there by default
21:53
<vagrantc>
jammcq: ah! i'm mostly using "dnsmasq" instead of ISC's dhcpd
21:53
<jammcq>
vagrantc: also, ubuntu and debian both drop their dhcpd.conf files into /etc/ltsp/dhcpd.conf
21:53
<vagrantc>
jammcq: and that's bad?
21:53
<jammcq>
BUT, debian's dhcpd startup script doesn't look there for the config file
21:54
it only looks in /etc/dhcp3/dhcpd.conf
21:54
<vagrantc>
jammcq: it's meant to be added as an include line
21:54
<jammcq>
ah
21:54
is that supposed to happen automatically? or is it documented that a manual thing needs to be done?
21:54
<vagrantc>
because, i am not confident that i will be able to perfectly script an automagic configuration of the DHCP server ...
21:54
should be documented in a manual, yes.
21:55
<jammcq>
agreed, dhcpd.conf is a big ugly beast
21:55
"should be" == "still needs to be done" ?
21:55
<vagrantc>
jammcq: if you don't see it in /usr/share/ltsp-server/* ... yes, it still needs to be done.
21:55
<jammcq>
or is there a manual that i've just not found yet
21:55
ok, gotcha
21:56
believe me, i'm not criticizing anything here. the work is fantastic
21:56
<Bhaskar>
hello everybody, i want to integrate LTSP12 in my linux debian based, how?
21:56
<jammcq>
hmm, ltsp-12 ?
21:56* vagrantc suspects it still needs to be done
21:56
<jammcq>
we're just getting ltsp-5 off the ground
21:56
<vagrantc>
jammcq: i'm not taking it as criticism so much as reminders of things i've been putting off :)
21:56
<jammcq>
yeah, cool
21:56
cuz it's all pretty sweet stuff
21:56
<vagrantc>
!integration
21:56
<ltspbot>
vagrantc: "integration" is http://wiki.ltsp.org/twiki/bin/view/Ltsp/IntegratingLtsp
21:57
<vagrantc>
Bhaskar: that was for you
21:57* vagrantc heart ltspbot
21:57
<Bhaskar>
vagrantc, we develope Nepalinux, we want to integrate
21:57* vagrantc fires up X
21:58
<vagrantc>
Bhaskar: yes, well, read that wiki page and start asking questions :)
21:58
<jammcq>
vagrantc: there's an issue with both Ubuntu and Debian's NFS. when you restart nfs, or run 'exportfs -ra', after adding the entries, it complains about not having a 'subtree_check' or 'no_subtree_check' option
21:59
<vagrantc>
jammcq: yeah... i guess we just have to learn which we want for LTSP and add it into our /etc/exports line(s)
22:00
<jammcq>
yep
22:01
i've got tarballs of both the etch-ltsp and feisty-ltsp ready to put on our website.
22:01
<otavio>
jammcq: great news. What's still missing for it?
22:01
<mistik1>
otavio: hey man, long time
22:01
<vagrantc>
jammcq: should probably call it an etch-prerelease or something, as etch will be different tomorrow, and the next day, and the next day, etc.
22:02
<otavio>
mistik1: yes, indeed ;-)
22:02
<jammcq>
just a few scripts, to ease the installation on any system you want
22:02
vagrantc: true enough
22:02
<otavio>
jammcq: ah great
22:02
<vagrantc>
i thought ogra already wrote some scripts to do that downloading stuff ...
22:02
<otavio>
mistik1: how are you doing?
22:03
<mistik1>
otavio: I'm doing well now, was not for a few months there
22:03
Had some major turbulance in my life
22:03
<otavio>
mistik1: neither I. I've been busy working at work project
22:03
mistik1: what happened?
22:03
<vagrantc>
jammcq: are you configuring it such that you can install it to /opt/ltsp/debian-etch-i386 /opt/ltsp/ubuntu-feisty-i386
22:04
jammcq: so someone could easily choose between the different distros ?
22:04
<mistik1>
otavio: in a nut-shell separation, divorce, custody, support
22:04
<otavio>
mistik1: uh!
22:04
<mistik1>
yea
22:04
<otavio>
mistik1: life sometimes is hard
22:04
mistik1: i hope you're fine
22:05
<mistik1>
I'm good now, thanks ;)
22:05
<jammcq>
vagrantc: I was just thinking of /opt/ltsp/i386, and let them move them around, but you've got a great idea there
22:05
<otavio>
mistik1: that's indeed the way to deal with that kind of problems. Pushing forward :-)
22:05
<vagrantc>
jammcq: might be good to craft the tarballs so that it makes that easier ...
22:05
<mistik1>
otavio: furthermore i'm ready to hack once more
22:05
<otavio>
jammcq: yes. I think that have them available for user's choice would be great.
22:06
<vagrantc>
mistik1 already wrote some scripts to test ltsp in qemu, if i recall...
22:06
<otavio>
mistik1: aha!
22:06
mistik1: would be good if you could come back to work on the Gentoo stuff. Have you done something about it since last time we talked?
22:06
<mistik1>
vagrantc: It was not written specificly with ltsp in mind but modification would be trivial
22:07
<vagrantc>
mistik1: i wear ltsp-colored glasses, i guess.
22:07
<mistik1>
hehehe
22:07
<otavio>
hehe
22:07* vagrantc wonders what the official colour of ltsp is
22:07
<otavio>
hahhaa
22:08
<sbalneav>
Puce.
22:08
<mistik1>
otavio: I am doing EVERYTHING in Gentoo these days
22:08
<otavio>
mistik1: please, explain
22:08* mistik1 has ~10 chroots all over his disk
22:08
<jammcq>
Puce or Puke ?
22:08
<sbalneav>
Hex triplet #CC8899
22:08
<otavio>
mistik1: but have you included the ltsp support for it?
22:08
mistik1: the scripts and like?
22:09
<mistik1>
I started working on the initscripts so far
22:09
<vagrantc>
sbalneav: you're even here this evening... this is a fun friday.
22:09
<otavio>
mistik1: but hadn't yet looked at the plugin system?
22:09
<mistik1>
dberkholz did some layout of how it would work
22:09
<sbalneav>
:)
22:09* otavio my gf is calling me. 2min
22:10
<mistik1>
otavio: Yes in fact the Gentoo plugin will be rather easy with emerge
22:10
<vagrantc>
mistik1: don't forget i've been planning on making the init scripts into plugins too :)
22:10
mistik1: in fact, i already have a branch for that
22:10
<mistik1>
explain
22:11
<vagrantc>
mistik1: splitting the init scripts into smaller pieces that get sourced on boot ...
22:11
mistik1: rather than a single monolithic script
22:11
<jammcq>
vagrantc: scotty and I are both in Toronto right now
22:11* otavio is back
22:11
<mistik1>
vagrantc: That would be welcome I think
22:11
<vagrantc>
mistik1: which makes it easier to ignore distro-specific code
22:12
mistik1: i know, that's why i wrote it :)
22:12
<otavio>
vagrantc: compating with the code we've written at debconf, how different ltsp5 is from it?
22:12
<vagrantc>
jammcq: ah, toronto. i gazed for a fair number of hours at toronto, if i remember correctly
22:12
otavio: same thing
22:13
otavio: i mean, a few small changes, but basically the same thing.
22:13
<otavio>
vagrantc: oh great :-)
22:13
vagrantc: yes yes, I gotcha it
22:13
vagrantc: will you atendee to debconf7?
22:13
<vagrantc>
otavio: i basically split the init scripts up in a similar way ...
22:13
<otavio>
vagrantc: good
22:13
<vagrantc>
otavio: if i can get sponsorship
22:14
otavio: well, i started splitting them up. it was too late for etch.
22:14
<otavio>
vagrantc: would be great. I'll bring some nice stuff to show to you ;-)
22:14
<mistik1>
You guys are moving very fast with all this, sometimes I swear my head is spinning
22:14
<otavio>
mistik1: hahaha
22:14
<sbalneav>
vagrantc: when you're done with everyone else, I've got a debian policy question to ask you.
22:14
<otavio>
vagrantc: we're finishing the company's product
22:14
<vagrantc>
sbalneav: have you read the topic? :P
22:15
sbalneav: please, ask :)
22:15
<jammcq>
otavio: where is debconf7 this year?
22:15
<sbalneav>
I wasn't asking to ask the question, I just didn't want you to have too many threads of execution at the same time. :) But ok....
22:15
<vagrantc>
sbalneav: oh, you totally got to merge my ltspfs branch into upstream. it's like, waaaay cool. :)
22:15
<otavio>
jammcq: Edinburgh
22:15
<jammcq>
ah
22:15
<sbalneav>
Oh?
22:15
<otavio>
jammcq: probably miss spelled
22:15
<sbalneav>
What's new with it?
22:16
<otavio>
jammcq: will you go?
22:16
jammcq: if yes, we can talk more :-)
22:16
<vagrantc>
sbalneav: oh, it's nothing special really, just some patches i want that make it easier to work with debian :)
22:16
<jammcq>
otavio: close enough for me to know what you are talking about
22:16
when is debconf ?
22:16
<otavio>
jammcq: June
22:16
jammcq: hold
22:16
<sbalneav>
sure, I got no problem with that, wait 'till you hear what I've got planned...
22:16
<jammcq>
hmm, i'm already doing too much travelling this year
22:16
<sbalneav>
here's the situation:
22:16
<otavio>
Sunday 17 to Saturday 23 June 2007
22:16
<sbalneav>
Right now, server side, we need to support 3 things...
22:17
<jammcq>
and it looks like June 21-27th I'll be travelling
22:17
<vagrantc>
there's also debcamp the week before
22:17
<otavio>
vagrantc: will you want to stay all over night awake with me?
22:17
<vagrantc>
where you actually do real work
22:17
<sbalneav>
1) ltspfs itself, 2) ltspfsmounter, the wrapper script, and lbmount, that makes the links in /media.
22:17
<vagrantc>
otavio: heh.
22:17
sbalneav: yeah, it's messy.
22:17
<otavio>
vagrantc: do you remember how much we've done of work... right? You called me crazy ... three times?
22:17
vagrantc: four? hehehe
22:17* vagrantc suspects more
22:18
<sbalneav>
Problem is, trying to make an upstream ltsp, to be loaded on multiple distros/arch's etc. is a pain for binaries...
22:18
<otavio>
hhaha
22:18
<sbalneav>
so, I'm thinking this:
22:18
<vagrantc>
otavio: when the colour was missing from your face, i really started to wonder :P
22:18
otavio: and our talk was sooooo awful :P
22:18
<otavio>
vagrantc: some time ago I was talking with Daniel (debian-live guy) and he said: Oh! Now I see why Vagrant said you don't sleep heheeh
22:18
<sbalneav>
1) re-implement ltspfs in python. This is a breeze, I've written fuse filesystems in python with the python-fuse bindings, shouldn't be a problem.
22:19
<vagrantc>
sbalneav: i see the ram requirements skyrocketing
22:19
but the maintainability probably improving
22:19
<sbalneav>
2) re-implement lbmount as python/shell something, since it's really just a sanity wrapper
22:19
for mount --bind.
22:20
Probably a bit, but not too much, i'd expect.
22:20
<vagrantc>
we don't even use lbmount in debian/ubuntu, do we?
22:20
<sbalneav>
Well, how do you make icons pop up on the desktop?
22:20
<vagrantc>
actually, we do. i was thinking of something else.
22:21
<sbalneav>
ok
22:21
<otavio>
hhe
22:21
<sbalneav>
so, here's the issue.
22:21* otavio waiting git to sync
22:21
<otavio>
vagrantc: I just finished the package checking. All ok to me.
22:21
<sbalneav>
reimplementing lbmount as a script's easy, it's already half done. The problem is, the setuid bit.
22:21* vagrantc wants to move the ltspfs code in the ltsp-client and ldm packages to be moved into ltspfsd
22:22
<vagrantc>
otavio: surely theres something that could be done better :P
22:22
<sbalneav>
now, I can get around THAT real easy, by doing some sudo magic.
22:22
<otavio>
vagrantc: yes, but nothing grave
22:22
<vagrantc>
otavio: like thwacking upstream with a big stick so we actually have an upstream :)
22:22
<otavio>
vagrantc: hahahha
22:22* vagrantc looks at jammcq and sbalneav
22:22
<otavio>
vagrantc: yes. This is important
22:23* otavio looks at them, too
22:23
<sbalneav>
What do you think we're WORKING on here??!? :)
22:23
<jammcq>
vagrantc: hey buddy, that's one of our goals here this weekend
22:23* vagrantc is giving a hard time or two
22:23
<jammcq>
although time is running short
22:23
<vagrantc>
jammcq: you mean you weren't just hanging out in canada to be cool?
22:24
<otavio>
haha
22:24
<mistik1>
cool being the operative word
22:24
<sbalneav>
here's the question: could a package modify the /etc/sudoers file, or is that a no-no in the debian packaging world.
22:24
<vagrantc>
sbalneav: big no no
22:24
<otavio>
sbalneav: no-no
22:24
<sbalneav>
ok
22:24
<mistik1>
huge no no here too
22:24* vagrantc notes a +1 for gentoo
22:24
<otavio>
sbalneav: you cannot touch others configuration files
22:24
sbalneav: hold on
22:25
<mistik1>
CONFIG_PROTECT=/etc
22:25
heh
22:25
<sbalneav>
so, having yet another setuid binary on your system's NOT a problem, then? :)
22:25
<otavio>
sbalneav: In this specific case you can. /etc/sudoers isn't part of any package.
22:25
<vagrantc>
sbalneav: so... we need to run lbmount (or its future cousin) as root, because it's mounting filesystems?
22:25
otavio: not true!
22:25
<otavio>
sbalneav: but usually you shouldn't do that.
22:25
<sbalneav>
lbmount is setuid todat.
22:25
today.
22:26
<otavio>
vagrantc: go check ;-)
22:26
otavio@neumann:~$ dpkg -S sudoers
22:26
sudo: /usr/share/man/man5/sudoers.5.gz
22:26
sudo: /usr/share/doc/sudo/examples/sudoers
22:26
<sbalneav>
it has to be, because it has to do a mount --bind from /tmp/sbalneav-ltspfs.... to /media/sbalneav/....
22:26
<vagrantc>
otavio: it's still not ok, at least, according to the release managers
22:26
<mistik1>
hmm
22:26
otavio is right
22:26
<sbalneav>
What if you have a COMMAND that modifies the file.
22:27
?
22:27
<otavio>
vagrantc: if you _ask_ the user it's ok
22:27
<mistik1>
mistik1@firelinks ~ $ equery f /etc/sudoers
22:27
!!! The query '/etc/sudoers' does not appear to be a valid package specification
22:27
<ltspbot>
mistik1: Error: "!!" is not a valid command.
22:27
<vagrantc>
otavio: i don't think so.
22:27
<otavio>
vagrantc: hehe
22:27
vagrantc: why not?
22:27
<vagrantc>
otavio: not in maintainer scripts, anyways.
22:27
otavio: in programs the admin runs, sure.
22:27
<otavio>
vagrantc: yes, you can. Iff you ask the user.
22:28
vagrantc: gforge does it
22:28
vagrantc: other packages does also
22:28
<jammcq>
wasn't one of the goals of dbus to rid us of all this setuid crap?
22:28
applications ask dbus to do something for them
22:28
<sbalneav>
What about ltsp-update-sudoers as a script?
22:28
<vagrantc>
sbalneav: so ... the mounting is happening on the server, or on the thin client?
22:28
<mistik1>
all that aside I dont believe a package modifying /etc/sudoers is a good idea
22:28
<sbalneav>
server
22:28
<otavio>
sbalneav: what you need to add on sudoers?
22:29
<vagrantc>
otavio: essentially a wrapper around "mount"
22:29
<sbalneav>
something like:
22:29
Cmnd_Alias LBMOUNT=/usr/share/ltsp/scripts/lbmount.py
22:29
%fuse ALL=NOPASSWD: LBMOUNT
22:29
<vagrantc>
sbalneav: why can't we mount directly to /media/USER ?
22:30
<sbalneav>
ls -al /media
22:30
drwxr-xr-x 3 root root 4096 2007-03-03 12:44 .
22:30
how's a user gonna make a dir in there? :)
22:30
<otavio>
sbalneav: but if it's suppose to be done for all users, sudo doesn't make sense. Just use setuid ;-)
22:30
<mistik1>
will mount not still complain unless its defined in fstab with -ouser
22:30
<sbalneav>
pmount normally does that.
22:30
<vagrantc>
mistik1: if it's setuid, no. :)
22:30
<sbalneav>
otavio: Then I'm maintaining a setuid binary program.
22:31
which needs to work on:
22:31
<mistik1>
vagrantc: if its setuid there is no need for sudo
22:31
<otavio>
sbalneav: well yes but it's igual of a global allowed command on sudo :P
22:31
<vagrantc>
mistik1: yes, that's the current debate
22:31
<sbalneav>
debian, ubuntu, fedora, slackware, mandrake, mepis, etc etc etc etc etc
22:31
<otavio>
sbalneav: basically same side-effect
22:31
<jammcq>
first rule: keep it simple
22:31
<vagrantc>
otavio: actually, it's not a global, it's only allowed to users in the fuse group
22:32
is it bad to make it setuid = root, and only group = fuse executable ?
22:32
<otavio>
vagrantc: virtually, every user on a LTSP server ;-)
22:32
<sbalneav>
yeah, but handing out a script is distro-independent. Maintaining a binary's a pain.
22:33
<jammcq>
somebody is whining :)
22:33
lemme call the Whambulance
22:33
<sbalneav>
Fine, then I'll hand it off for you to do. :)
22:33
<mistik1>
Why not just make the binary non-setuid and document the sudoers mods
22:33
<jammcq>
it'll become a support problem
22:34
<mistik1>
adding two lines to a file?
22:34
<otavio>
mistik1: people usually do not read docs ;-)
22:34
<jammcq>
yep
22:34
<otavio>
mistik1: eheh
22:34
<mistik1>
ugh!
22:34
<jammcq>
they'll muck it up, and lock themselves out of their own system
22:34
<otavio>
mistik1: boring human beans ;-)
22:34
<jammcq>
and it'll be OUR fault
22:34
<vagrantc>
messing with sudoers is tricky business.
22:35
<jammcq>
setuid gets my vote
22:35
<vagrantc>
not to say setuid is to be taken lightly, but its fairly simple to do it well programmatically
22:35
<otavio>
sbalneav: as jammcq suggest, is too difficult to use d-bus to do this work?
22:35
<mistik1>
we should have a disclaimer that LTSP should not be run by trained monkeys
22:35
<otavio>
jammcq: mine too. hehe
22:35
<sbalneav>
So, I still haven't had my question answered: if there's a program like ltsp-update-sudoers, and THAT updates the file, is that in keeping with the debian policy?
22:35
I have no idea about dbus.
22:35
<jammcq>
ultimately, I think d-bus is supposed to be the answer, but i don't have a clue where to start with that
22:35
<vagrantc>
sbalneav: as long as it's not called from the maintainer scripts, and the admin runs it manually.
22:35
<sbalneav>
If someone wants to point me at a tutorial, I'm happy to look at it.
22:36
vagrantc: right, so like ltsp-update-sshkeys
22:36
<vagrantc>
jammcq: does dbus work with things other than GNOME, or specific window managers?
22:36
<otavio>
sbalneav: Debian Policy forbids any program to touch a conffile. A conffile, for Debian is something that is included in a package and in /etc.
22:36
<mistik1>
sbalneav: http://dbus.freedesktop.org/doc/dbus-tutorial.html
22:36
<otavio>
sbalneav: so you can change sudoers
22:36
<vagrantc>
sbalneav: or ltsp-build-client
22:36* jammcq thinks it works with kde too
22:36
<jammcq>
and prolly xfce
22:36
vagrantc: also, d-bus doesn't depend on the window manager
22:36
<sbalneav>
mistik1: seen that one. Next to useless.
22:37
<jammcq>
background processes can use it to communicate
22:37
<vagrantc>
otavio: but youre also not allowed to edit "configuration files" of other packages, according to the etch_release_policy.txt
22:38
<otavio>
vagrantc: yes
22:38
<vagrantc>
"configuration files" being a broader category than "conffiles"
22:38
<jammcq>
people are using d-bus, we're not treading on new ground here. I know for sure the Ubuntu guys would be willing to help. I know the Fedora guys would also help. as for Debian, I don't know who would be the person. it may very well be a Ubuntu guy
22:38
<otavio>
vagrantc: but sudoers isn't part of any package
22:38
vagrantc: it's used by sudo but it's another part of history.
22:39
vagrantc: policy doesn't cover it
22:39
vagrantc: for policy what means if it's or not part of a package
22:39
<vagrantc>
otavio: years ago, when this issue came up discussing lessdisks, steve langasek commented that it doesn't matter weather it's owned by another package, you can't mess with files that are not yours.
22:39
<otavio>
vagrantc: but sudoers do not have an owner
22:39
vagrantc: dpkg -S proved it
22:40
<vagrantc>
otavio: doesn't matter if it doesn't have an owner, the owner is certainly *not* ltsp
22:40
<otavio>
vagrantc: well, that's right ... but if you does ask, you can mess with it
22:40
vagrantc: gforge mess with all over the system in that way
22:41
<vagrantc>
otavio: in the postinst scripts?
22:41
<otavio>
vagrantc: yes
22:41
<vagrantc>
(or other maintainer scripts)
22:41
<otavio>
vagrantc: both
22:41
vagrantc: but them are called on postinst
22:41
<sbalneav>
so if we have a 'ltsp-update-sudoers' command, that notifies the user that it's about to touch the sudoers file, and allows them to back out, that's ok?
22:42
<otavio>
sbalneav: yes
22:42
<sbalneav>
vagrantc: ?
22:42* sbalneav is looking for concensus :)
22:42
<jammcq>
-1
22:43
<otavio>
sbalneav: but I still think that a setuid is a better solution for it right now
22:43
<vagrantc>
i think so
22:43
<otavio>
sbalneav: later, move to d-bus
22:43* vagrantc agrees with otavio
22:43
<sbalneav>
jammcq: OK, so who's going to maintain setuid binaries for all the different archs?
22:43
and distros?
22:43
<jammcq>
we don't maintain binaries, we maintain source code
22:43
remember?
22:44
binaries aren't our problem
22:44
<sbalneav>
Ah, so what are we doing putting tarballs on our site for?
22:44
<vagrantc>
maintaining a distro
22:44
<jammcq>
we certainly aren't maintaining them
22:44
<otavio>
sbalneav: I thought it was one of the most important change since ltsp4 -> ltsp5 ;-)
22:44
<jammcq>
we're facilitating easy installation of bits for someone who's distro hasn't caught up yet
22:45
<sbalneav>
OK, so johnny slackware comes along, downloads the tarball for the chroot, and wants localdev to work. So, what do we say? What goes on the server?
22:45
Do we simply not provide the server side?
22:46
<jammcq>
well, possibly hand him the source package, tell him to ./configure && make && make test && make install
22:46
dunno for sure, that's what we're trying to figure out
22:46
this is a "transitional period"
22:47
<sbalneav>
Wait, ok, lemme get this straight: adding two lines to /etc/sudoers is a support nightmare, but stepping some guy who doesn't speak english through a binary compile's ok?
22:47
<jammcq>
umm, yeah
22:50
somebody's getting a bit grumpy about the whole thing
22:50
<ramsys>
jammcq: if i have live cd and i connected to net is there any way to connect to my college server that support thin clients (ltsp) and access my college server or my own desktop
22:51
<vagrantc>
sbalneav: ummmm... hwo do you propose to get binaries for slackware before slackware is officially supported?
22:51
<jammcq>
ramsys: you could prolly use NX or VNC
22:51
<sbalneav>
Well, that was what implementing that stuff as scripts + sudo would solve: they'd be distro independent.
22:51
<ramsys>
ok
22:51
<vagrantc>
sbalneav: except there still would be distro-specific issues.
22:52
sbalneav: for example, much of the python is dependent on python 2.4 modules
22:52
<jammcq>
my whole goal of having the tarballs is to give the fedora/novell/slackware guys a swift kick in the butt, to get them moving
22:52
<sbalneav>
Sure, but it's one HECK of a lot easier to modify a script, than it is to download source, and patch it, and recompile.
22:52
<jammcq>
maintaing binaries is a DISTRO problem
22:53
<sbalneav>
However, the concensus seems to be, we just maintain the source. Users are on their own.
22:54
<jammcq>
in an open source project, developers are free to work on whatever aspect they want to work on. if they have an itch to scratch (re-implementing something in python) then they are free (and encouraged) to do it. afterall, he who has the bits wins. So, if coming up with an implementation in Python is something you wanna do, then DO it, see how it works. prove us wrong
22:55
we're all giving our opinions about what WE think is the right way to do it
22:55
sometimes we're right
22:55
sometimes we're passionate about it
22:55
but in the end, whoever has the bits, wins
22:55
<vagrantc>
sbalneav: from a distro standpoint, there is little to no difference between a script and a compiled binary- they are treated the same.
22:55
<jammcq>
we learned that lesson with local devices and Ubuntu
22:55
<otavio>
sbalneav: if you want to improve it, looks at d-bus. That looks to be right way to go
22:56* otavio fetching kfreebsd d-i branch to work on
22:57
<jammcq>
http://en.wikipedia.org/wiki/Dbus
22:57
<sbalneav>
dbus is a communications protocol, right?
22:57FernandoMM has joined #ltsp
22:57
<sbalneav>
So, something's going to send a message over dbus, correct?
22:57
and something's going to listen and act on that, yes?
22:57
<jammcq>
D-Bus is an inter-process communication (IPC) system with three architectural layers
22:59
<otavio>
sbalneav: :-)
22:59
<sbalneav>
Sure. So, my question here is: I'm a userspace, unprivileged process.
22:59
<otavio>
sbalneav: you can "use" d-bus and ask for it do something to you
23:00
<jammcq>
look at the introduction on wikipedia, it shows what the purpose is
23:00
<otavio>
sbalneav: d-bus has the rights to make it
23:00
<jammcq>
looks to me like it fits right into what we are doing
23:00
<otavio>
jammcq: yes, I think it does
23:00
<sbalneav>
So, dbus already knows how to make ltspfs bindmounts from /tmp to /media?
23:01
<jammcq>
umm, yes, i'm sure it knows everything
23:01* vagrantc doesn't like mounting to /tmp
23:01
<jammcq>
same as python, it already knows too
23:02
dbus is not a solution, it's just one of the pieces of the solution
23:03
<sbalneav>
Well, that's what I'm trying to understand. As far as I've ever read, it's just an ipc mechanism. So I'd need some kind of server out there, so here's the question: do I write a server process, that runs as root, and the processes just send a message to it?
23:03
<jammcq>
dunno
23:03
i'm not a d-bus expert, but i'm sure we could get someone who is, to help out
23:03FernandoMM is now known as fernando[jantar]
23:03
<jammcq>
as soon as we find Ogra, we'll ask him who knows this stuff inside-and-out, and drag them in here
23:04
seems like it might be pitti
23:04
but I don't remember for sure
23:04
<otavio>
jammcq: pitti == Martin?
23:04
<jammcq>
yeah
23:04
in the end, it could be that python + dbus is the way to go
23:04
<otavio>
jammcq: yes, he probably knows it too
23:05
<sbalneav>
I mean, if it's just a case of 1) write a server that sits and listens to dbus, and processes message requests to do bind mounts and unmounts, and then a client piece that sends the requests, that's ok, but, we're back to the same problem:
23:05
<jammcq>
that way, we get around the "binary" issue AND we get the priveleged things run properly
23:05
"same problem" ?
23:05
<sbalneav>
it's stilll a bunch of binaries, with all the dependent problems. :)
23:05
<jammcq>
read my previous post
23:05
<vagrantc>
otavio: in case you're curious about my interpretation of policy why it's not ok to modify "configuration files" even if nobody "owns" them ... http://lists.alioth.debian.org/pipermail/pkg-ltsp-devel/2006-December/000540.html
23:06freemindx has quit IRC
23:07
<jammcq>
there's an article on IBM's site about d-bus, complete with examples (written in python no less). only problem is the article is kindof old
23:07
<otavio>
vagrantc: reasonable for both to be installed at the same time, one of these
23:07
packages must be defined as _owner_ of the configuration file, i.e.,
23:07
vagrantc: there's no "owner" on sudoers case
23:07
<vagrantc>
otavio: in what way is "sudo" not the owner of sudoers?
23:08
<sbalneav>
Perhaps I'm asking the wrong question. Let me ask it this way, then: if we have someone who doesn't currently have a native implementation of ltsp5, and who doesn't have the capability to build binary files, for whatever reason, is our policy simply, "Then you won't be able to get local devices to work" ?
23:08
<jammcq>
python code for both the message sender and listener is less than 40 lines of code
23:08
<otavio>
vagrantc: doesn't looks like
23:08
<vagrantc>
otavio: the *only* configuration files that dpkg -S will report are conffiles
23:08
<otavio>
People, I've to leave
23:08
<jammcq>
no, our policy is if we can help them, we will. if we can't, we'll apologize and hope they can find someone who can help them.
23:08
<otavio>
vagrantc: yes, that's why I think it has no owner
23:09
vagrantc: I'll check it tomorrow
23:09otavio has quit IRC
23:09
<jammcq>
and we'll remind them they should send an email or file a bug report with their favorite distro, to get LTSP integrated
23:10
<vagrantc>
sbalneav: so, essentially, we either need to provide a "universal cross-distro package" or we need ltsp (and related packages) integrated into all distros ...
23:11
<sbalneav>
The problem I was trying to solve was provide pieces that would be easier for us to support. If we're ultimately not worried about supporting people on distros that don't have a native ltsp5 implementation, that's fine.
23:11
<vagrantc>
i suspect a truely universal package is going to be very challenging to achive
23:11
<ltsppbot>
"jam" pasted "STDIN: ..." (54 lines) at http://pastebot.ltsp.org/18
23:12
<sbalneav>
Thanks for the dbus hello_world :)
23:12
<jammcq>
heh
23:12
you asked for pointers to get started
23:12
<sbalneav>
Now all I need to do is take that and make a rocketship out of it.
23:12
<jammcq>
if you don't want them, that's fine
23:12
<whiprush>
sbalneav: I love you dude.
23:14
<vagrantc>
sbalneav: basically, what i think you need to do is provide a good solid foundation that will likely work in any distro, and then distros will adapt it as needed.
23:15
sbalneav: like, the ltspfs Makefile*
23:15
<sbalneav>
Well, jammcq and I have been talking.
23:16
It looks as though the concensus is:
23:16
scripts == good, sudoers == bad, dbus == good.
23:16
So, now I know what I'm doing tomorrow.
23:16
<vagrantc>
is this a grudging consensus, or a happy one?
23:17
<jammcq>
scotty thought I was arguing against python
23:17
<sbalneav>
Not at all, jammcq and I disagree sometimes, but we've never got to be angry :)
23:17
<jammcq>
i'm not, although if I had to write it, it would be Perl :)
23:17
<sbalneav>
Wasn't necessarily python. I'm happy with ANY scripting language.
23:17* vagrantc is happier with python
23:17* vagrantc suspects ogra will be too
23:17
<jammcq>
most people are
23:18
and that's fine by me
23:18
<sbalneav>
It was more a "which is easier, maintain some kind of interpreted script, versus C binaries"
23:18* vagrantc is clueless with C
23:19
<jammcq>
sbalneav: this is a little bit like our decision 2 years ago to figure out lbus/ltspfs, and then sit down and do it
23:19
<vagrantc>
lbus, that was the one we don't use
23:20
<jammcq>
well, we don't use it anymore
23:20
but it worked great in LTSP-4.2
23:20
<vagrantc>
as if LTSP 4.2 isn't used anymore!
23:21
<jammcq>
well, it's just not being developed anymore
23:21
<sbalneav>
I'm just trying to have us come up with something that will be easier to maintain on distros with NO ltsp5 support, because, as much as WE see the sanity in it, I suspect the practical side of it will be: it will be a few years before all "mainline" distros support it.
23:21
lbus worked great.
23:21
<vagrantc>
i swear, for every debian user who tries ltsp5, nearly 2/3rds of them switch back to ltsp 4.2 :(
23:22
<sbalneav>
What's their biggest complaint?
23:22
<vagrantc>
which makes me feel unsettled
23:22
i think boot times.
23:22
y'all set a crazy fast expectation :P
23:22
<sbalneav>
Really? BOOT times?
23:23
<vagrantc>
boot times, or for whatever reason, sometimes people use some janky mirror and the download fails.
23:23g333k_laptop has joined #ltsp
23:23
<jammcq>
well, 22 seconds to boot a thin client is pretty darned nice
23:23
<vagrantc>
or maybe people try to use ltspadmin and it breaks everything badly.
23:24
<sbalneav>
I know perception's 9/10ths of the issue, and jammcq and I have talked about it several times. I still find it amazing that people manage to get totally turned off on it.
23:24mistik1 has quit IRC
23:24
<vagrantc>
i find it amazing, but do not deny that it really seems to rule some people.
23:24
a lot of people.
23:25
we just need dancing clowns or something during the boot process.
23:25
<sbalneav>
See for ME, it's the ssh tunnel. I need to integrate both XDMCP and SSH methods to get localdev going nice both ways. Me personally, I'm less concerned with boot, as how it runs after.
23:25
whiprush: you still there?
23:25* vagrantc knows clowns will scare some people, though
23:26
<whiprush>
sbalneav: just reading along
23:26
:)
23:26
<vagrantc>
sbalneav: yeah. i also feel like 2/3rds of the boot time is actually just LDM loading up ...
23:26
<jammcq>
sbalneav: once you become the d-bus expert, perhaps you'll do the network implementation of dbus, and use that for local dev control
23:26fernando[jantar] has quit IRC
23:26
<sbalneav>
whiprush: now, you see, if you had a BLOG, you could blog about the momentous decisions being made here!
23:26
<whiprush>
sbalneav: Canada rules.
23:27
that's all I've got.
23:27* sbalneav sticks tongue out at jammcq :)
23:27
<vagrantc>
sbalneav: wouldn't it be safer and simpler to make a setuid binary that makes the directory in the right place and mounts it directly there instead of /tmp ?
23:27
<whiprush>
sbalneav: so what's your real beef here.
23:27
tell me that
23:27
<jammcq>
/tmp ?
23:27
<whiprush>
I am trying to follow along with the discussion, but as an ltsp user I don't understand the issues here.
23:28
<sbalneav>
Yours was the only blog I read on a regular basis. Now I feel lonely that I don't get to read it. Thats my issue.
23:28
<vagrantc>
jammcq: the current approach mounts the ltspfs filesystems to /tmp, and them --bind mounts them to /media
23:28
<sbalneav>
Oh, the ltsp stuff.
23:28
<jammcq>
oh
23:28
<whiprush>
other than I expect whatever changes you make to be transparent to me. :)
23:28
sbalneav: later on ask jammcq how ltsp saved my job this week.
23:29
<vagrantc>
whiprush: i think much of the issue is we're going through a large shift in how the guts of LTSP work, and having a little trouble conceptualizing how that will all pan out
23:29mistik1 has joined #ltsp
23:29
<sbalneav>
It's just a "when we distribute server side pieces, do we hand out binaries, and the problems they entail, or scripts, and the problems THEY entail, and which is easier for US to support"
23:29
<whiprush>
vagrantc: I'm going to step out on a limb here.
23:29
and say that ...
23:30
<sbalneav>
whiprush: yeah, he had to tell me about it. AS OPPOSED TO MY READING IT ON YOUR BLOG!
23:30
<whiprush>
like, when scotty and jam talked to me about muekow .. the idea seemed to be "ltsp as a service"
23:30
<sbalneav>
:p
23:30
<whiprush>
I would bet 20 bucks (and not much more), that projects like apache and samba fixed this kind of thing years ago.
23:30
I would see how they handled this kind of thing, and look if that helped them or not.
23:31
so like ... 2 years ago ... ltsp was an add on thing.
23:31
you had to want it to get it
23:31
and as far as I know, now the goal is "out of the box with the distro"
23:31
<vagrantc>
well, "apache" and "samba" are more coherrent pieces of software ... LTSP is more a collection of different pieces of software to accomplish thin-clients (and more)
23:31
<sbalneav>
right.
23:32
<vagrantc>
whiprush: the "out of the box" thing is almost a separate issue.
23:32
<whiprush>
vagrantc: I don't think that should matter.
23:32
<jammcq>
well, LTSP is becoming more of a "specification" than an actual collection of code
23:33
<whiprush>
I think the use case (and I say this as a customer), is "thin clients ootb"
23:33
<jammcq>
there's a bit of glue code that we provide, but more important is the whole concept of booting a thin client
23:33
<whiprush>
and considering the numbers of deployments you guys have
23:33
this is a decent ... bargaining chip with the distros.
23:34
<jammcq>
we are "transitioning" (to borrow a word from a friend of mine), from a mini-distro, to a specification for others to use their bits
23:34
<sbalneav>
For a long time, ltsp was just the stuff that ran in the chroot: you used the stuff that CAME with the server to implement it, i.e. dhcp, nfs, etc. Now we're getting into a situation where we're having to supply stuff for the server. Which is a new ball of wax for us.
23:34
<whiprush>
I played with sunrays for 3 years, and got burned. Now we're doing ltsp, and I can't be the only one.
23:34
if ltsp ends up being some "spec", then whatever.
23:35
for a long time, gnome tried shipping binaries for every distro
23:35
disaster.
23:35
now they do source only.
23:35
if you care about it, you handle it upstream.
23:36
<sbalneav>
yeah, that's where we are. Eventually, we'll be in gnome's position. But what do we do in the meantime. That's what we're trying to solve.
23:36
<whiprush>
yeah that sucks
23:36
you have millions of people in the middle.
23:36
<jammcq>
ultimately, we'd like the distros to be responsible for the binaries. we'll focus on the "techical issues" of thin clients, like local devices, local audio, security. as we find reasonable solutions to those problems, we'll create an implementation, and then hand it over to the distros to productize
23:36
<sbalneav>
yup.
23:37
<whiprush>
jammcq: right, like apache and samba do.
23:37
<jammcq>
yup
23:37
<whiprush>
you guys handle the project, let distros handle the details
23:37
<jammcq>
it really should be better that way, I think
23:38
<whiprush>
dude, if may be out of whack here ...
23:38
make ltsp an fdo standard.
23:38
<jammcq>
although i'd really love it, if we had a fedora guy and a novell guy in here, as passionate about this stuff as vagrantc and ogra are
23:38
<whiprush>
"this is how we do thin clients"
23:38
it's how everyone does thin clients anyway
23:38
<jammcq>
whiprush: we could (and maybe should) do that
23:39
<whiprush>
jammcq: if you push for an fdo-like spec, they'll have to contribute.
23:39
<vagrantc>
i'm just happy a project with name recognition started seeing things my way :)
23:39
<whiprush>
I mean, hell, all of ltsp is based on existing standards.
23:39
<jammcq>
contribute? what... money?
23:39
<whiprush>
formalizing it is ... like, half a step.
23:39
jammcq: people, i mean engineering resources.
23:40
not like, money money
23:40
you guys have been around forever, might as well push this as the "thin client" standard.
23:40
it's all open
23:40
no big deal.
23:40
<jammcq>
tru dat
23:41
<whiprush>
you guys have been working with upstreams that have been willing on making it rock.
23:41
and have kept open invites for what ... a decade now?
23:41
<jammcq>
well, 8 years now
23:41
Jan 99 is when we booted our first diskless workstation
23:42
<whiprush>
dude you know every fdo guy by name probably.
23:42
no one would object.
23:42
It's a defacto standard anyway
23:42
and then
23:42
<jammcq>
it's something i'd prolly look at in the next 6 months or so. too busy to think about it right now
23:42
<whiprush>
when RH and novell get stupid, you have one more trump card.
23:43
<jammcq>
RH an Novell don't seem to be too concerned with fdo standards
23:43
<whiprush>
well then, look at what you're shipping now, and supporting.
23:43
their loss
23:43
<jammcq>
based on that video from fudcon, i'd say there's a decent chance that Fedora will continue in our direction
23:43
no clue on what novell is doing though
23:44
<whiprush>
ok, well, that was my opinion. I'll shut up now
23:44
<jammcq>
heh
23:44* jammcq writes it down, as to not forget it
23:44
<whiprush>
just saying. jammcq, you've been one of the oldest X guys on the planet for years now.
23:44
<jammcq>
hmm, it's prolly not fair to call me an X guy, with follks like jgettys and keithp around
23:44
<whiprush>
if someone were to say "ltsp is now the standard for unix thin clients", no one would say otherwise.
23:45
make it an fdo thing, and rock on
23:45
<jammcq>
heh
23:45
<whiprush>
jammcq: you guys so undersell yourselves.
23:45
<jammcq>
what that means to me is "work" at a time when I don't have the time
23:45
<whiprush>
you guys ship more desktops than anyone
23:45
combined
23:45
more than kde
23:45
more than gnome.
23:46
you guys have been underated for like, 8 years. Screw that! Go tell people what the standard is, you have the users, and the mindshare.
23:47
ok, I'm done now, I promise. :)
23:47
vagrantc, sbalneav, what do you guys think?
23:47
<jammcq>
:)
23:47
<vagrantc>
i don't really even know what an fdo specification is.
23:47
<whiprush>
jammcq: you caught me on a saturday where I had to go into work to fix sunrays, so I am particularly livid. :)
23:48
vagrantc: freedesktop.org
23:48
like, they do things like, window manager hints
23:48
so like, drag and drop works
23:48
<vagrantc>
yeah. i hear about them all the time. people seem to like it.
23:49
so i guess ltsp should jump on the bandwagon
23:49
<whiprush>
vagrantc: so like, when people want something to become universal, they approach fdo.
23:49
things like, avahi for zeroconf.
23:49
poppler for pdf rendering
23:49
that can be used by all projects.
23:49
<vagrantc>
i am unpure, though. i've been playing with other technologies :)
23:49
<whiprush>
it's kind of a defacto standard house
23:49
<jammcq>
we've already got our bugzilla at fdo
23:49
so we're pretty much there already
23:50* vagrantc shrieks at the mention of bugzilla
23:50
<jammcq>
:)
23:50
<whiprush>
hey, you guys still doing bzr? (out of curiosity)
23:50
<jammcq>
bugzilla bugzilla bugzilla
23:50
how's that?
23:50
whiprush: barely using bzr
23:50
<whiprush>
git?
23:50
<jammcq>
nope
23:50
<whiprush>
!
23:50
<vagrantc>
well, debian and ubuntu is using bzr for ltsp work ... i don't know about upstream.
23:51
<jammcq>
it's bzr, we're just not really using much of it
23:51
most of the development bits are sitting on scotty's server at home
23:51
<whiprush>
oh man
23:51
not good!
23:51
<jammcq>
well, yeah, we know
23:52
<whiprush>
bzr/git, for the win man.
23:52
they both do svn plugins too
23:52
so you only need to learn one scm
23:52
<jammcq>
welp guys, i'm about to fall off my chair. gotta get some sleep
23:53
<whiprush>
vagrantc: I promise to make jammcq learn distributed version control.
23:53
jammcq: nite
23:53
<vagrantc>
jammcq: well, obviously you need a better chair if it's letting you fall off and sleep.
23:54
whiprush: good. good.
23:55
<whiprush>
vagrantc: i will try. :)
23:55
<sbalneav>
I'm getting tired.
23:55
I'm gonna hit the sack... Night all.
23:55
<vagrantc>
sbalneav, jammcq: you gonna be around tomorrow ?
23:56
<sbalneav>
yup, for at least the morning
23:56
You gonna be around?
23:57
<vagrantc>
hopefully... though with a little timezone skew, might be hard to make it for your morning.
23:57
<sbalneav>
where are you?
23:57
<vagrantc>
nuevo mexico
23:57
<sbalneav>
ah
23:57
<vagrantc>
MST
23:57
<jammcq>
:)
23:57
yep
23:57
at 2pm they'll kick us out of here
23:57
that's EST
23:57
<sbalneav>
ok well, we'll look for you.
23:58
I'm shuttn' down. Night all.
23:58sbalneav has left #ltsp
23:58* vagrantc will try to make a presence
23:59
<vagrantc>
whiprush: any word on getting LTSP running on the sunrays?