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


Channel log from 13 November 2007   (all times are UTC)

00:10Amaranth has joined #ltsp
00:20yopla has joined #ltsp
00:21
<yopla>
good night!
00:22
the dist-upgrade from edgy to gutsy succeeded
00:25mcfloppy_ has quit IRC
00:30
<yopla>
ltsp is really rapidly changing. What nice features in gutsy distro!
00:31
great work from ltsp team :)
00:32yopla has quit IRC
00:58plamengr has joined #ltsp
01:28rcy has joined #ltsp
01:45jvanrooyen has joined #ltsp
02:00ibt has quit IRC
02:29Egyptian[Home] has quit IRC
02:31Egyptian[Home] has joined #ltsp
02:48
<jvanrooyen>
hi there anybody had problems with hp lazerjet 1100 printer on a thin client?
02:48
I'm trying to install this printer followed all the steps, but now printing at all
02:49
Can anybdy help with this?
02:53Amaranth has quit IRC
02:56Amaranth has joined #ltsp
03:15johnny has quit IRC
03:18exodos_ has joined #ltsp
03:19exodos has quit IRC
03:20exodos has joined #ltsp
03:26Q-FUNK has joined #ltsp
03:32Egyptian[Home1 has joined #ltsp
03:32Egyptian[Home] has quit IRC
03:54CatastrophicToad has joined #ltsp
03:56
<CatastrophicToad>
Hi, I just created a page for LTSP at a new Linux wiki "TheLinuxVault" at http://thelinuxvault.net/wiki/LTSP. Everyone is welcome to edit the page.
04:03CatastrophicToad has quit IRC
04:31mcfloppy_ has joined #ltsp
04:38Jenna has joined #ltsp
04:54rafaelcardoso has joined #ltsp
04:58plamengr has quit IRC
05:00fox2k has joined #ltsp
05:03plamengr has joined #ltsp
05:04Jenna has quit IRC
05:04fernando1 has quit IRC
05:05Guaraldo has joined #ltsp
05:07Jenna has joined #ltsp
05:09fernando1 has joined #ltsp
05:12mcfloppy_ has quit IRC
05:20frownix has quit IRC
05:29masus has joined #ltsp
05:29
<masus>
hi all, where to buy cheap and good hardware ;) for ltsp
05:29
for example terminal clients
05:29
<rafaelcardoso>
where are you from
05:30
<masus>
germany or turkey
05:30
2 locations
05:32
the location is no problem i'll by in mass maybe also from usa
05:32
<rafaelcardoso>
hum search for tk 3670 geode board very good
05:33
<masus>
ok
05:34
<rafaelcardoso>
http://www.mpasystems.com.au/cgi-bin/mpa.pl?05_products+85_Termtek+38_Thin%20Client%20TK3670CE_TK-WT-36-700
05:34IRCzito has joined #ltsp
05:35
<rafaelcardoso>
masus: you need flash card or use pxe boot?
05:35
<masus>
pxe boot
05:36
how much is it ... do u know ?
05:36
<rafaelcardoso>
i´m from brazil here its expensive
05:36
because taxes of importation
05:37
<masus>
how much expensive :)
05:38
<rafaelcardoso>
u$ 500,00
05:38
<masus>
yeah it's expensive
05:49exodos has quit IRC
05:53Avatara has joined #ltsp
06:02J45p3r has joined #ltsp
06:03
<IRCzito>
rafaelcardoso: maybe I can help
06:04
rafaelcardoso: Im from Brasil
06:06
rafaelcardoso: Ow, im wrong, i think you want by sonthing
06:08
<rafaelcardoso>
IRCzito: entra no ltsp-br
06:09
<IRCzito>
sbalneav: good morning
06:13
sbalneav: could you giveme a hand?
06:21Artemka has joined #ltsp
06:21J45p3r has quit IRC
06:23
<Artemka>
ltsp 5 on Edubuntu 7.1 and local_apps working or no? If it is work how can made this?
06:29fernando1 has quit IRC
06:29
<Artemka>
nobody know this?
06:34elisboa has quit IRC
06:34elisboa has joined #ltsp
06:35fernando1 has joined #ltsp
06:42
<Jenna>
know what ?
06:44
Artemka, : things (ltsp) work out of the box in Edubuntu
06:46cliebow_ has joined #ltsp
06:46Artemka has left #ltsp
06:47Artemka has joined #ltsp
06:47elisboa has quit IRC
06:47elisboa has joined #ltsp
06:53jammcq has quit IRC
06:58GodFather has joined #ltsp
06:59subir has quit IRC
07:01Jenna has quit IRC
07:15rjune_ has joined #ltsp
07:16plamengr has quit IRC
07:17rjune_ has quit IRC
07:19rjune_ has joined #ltsp
07:19plamengr has joined #ltsp
07:30plamengr has quit IRC
07:43outofrange has joined #ltsp
07:49ogra has joined #ltsp
07:55masus has quit IRC
08:00rafaelcardoso has quit IRC
08:04plamengr has joined #ltsp
08:14jernst has joined #ltsp
08:16fox2k has left #ltsp
08:20Gadi has joined #ltsp
08:21yopla has joined #ltsp
08:22
<yopla>
Hello, is there any issue with ltsp 5 and 3C905-TX-M netcard?
08:23
<Q-FUNK>
!g
08:23
<ltspbot>
Q-FUNK: "g" is Gadi!!!!!!!!!!!!!!!!!!!!!!!!
08:23
<Q-FUNK>
:)
08:23
<Gadi>
heh, morning, Q-FUNK
08:27frownix has joined #ltsp
08:31
<Gadi>
Q-FUNK: if I test the amd patch on gutsy today, what patch should I test?
08:32
<Q-FUNK>
Gadi: if it stops crashing on X -configure and if DDC will correctly detect the max available rez/depth
08:32
<Gadi>
no, no - I mean, do you have a link to the patch you would like me to use
08:32
I notice in the bug report several new ones
08:34bjohnson has joined #ltsp
08:34
<Q-FUNK>
Gadi: ah, the last one I added as an attachment is the right one
08:34
applies with one line of fuzz against gutsy
08:35
alternately, I built packages with the patch.
08:35
<Gadi>
http://launchpadlibrarian.net/10299432/debdiff_xserver-xorg-video-amd_2.7.7.0-1_2.7.7.0-1ubuntu1.diff
08:35
that one?
08:35
<Q-FUNK>
yup
08:35
<Gadi>
you built gutsy pkg with the patch?
08:36IRCzito has quit IRC
08:37
<Q-FUNK>
yes
08:37
scroll down
08:37
<Gadi>
got it
08:37
Ill try that one first
08:44GodFather has quit IRC
08:44tier1__ has joined #ltsp
08:45
<cliebow_>
ogra:any comments on building ppc tarball from ppc-??gutsy cdimage
08:49
<yopla>
ogra: hello! direct edgy to gutsy works fine
08:51
only 2 packets were broken lisasl2 and libsasl2-modules but second upgrade resovl dependencies
08:51jvanrooyen has quit IRC
08:53
<yopla>
now I try to boot an old pc with 3c905C-TX card
08:53
but it can't fine tftp server :-(
09:00tier1_ has quit IRC
09:02Avatara has quit IRC
09:06frownix has quit IRC
09:09jernst has left #ltsp
09:11
<cliebow_>
wooooo!
09:14
<yopla>
yeaaaaah! ppc master :-)
09:15tier1_ has joined #ltsp
09:16
<cliebow_>
heh..ppc dubber...
09:16
just tossed the cd im this g4 and she up and upgrded hisself
09:19
<yopla>
lol
09:19
<cliebow_>
yep:i can talk bumpkin....
09:21
<Q-FUNK>
Gadi: unless I missed anything, the patch indeed prevents X from crashin under -configure, but here at least I cannot get DDC -reported max resolution to be used.
09:21
<yopla>
i've got trouble with a netcard 3c905c-tx any idea?
09:21
<Gadi>
Q-FUNK: well, Im not successfully getting X up at all - with amd, vesa, anything now
09:21* Gadi wonders if his chroot is funky
09:22
<Q-FUNK>
yikes!
09:22* Gadi removes pkg and reboots client
09:22moquist has quit IRC
09:23
<cliebow_>
yopla:is tftp runing?
09:23
<Q-FUNK>
yopla: exactly what sort of trouble?
09:24
<yopla>
yes it is
09:24
<Gadi>
heh, kernel panic this time
09:24
:)
09:25
but that happens every now and then with this thincan
09:25
<yopla>
it stop with "can't find tftp server"
09:27
<Q-FUNK>
kernel panic?!
09:27
ouch!
09:27
<yopla>
with another pc client it works like a charm.
09:28Egyptian[Home] has joined #ltsp
09:30
<yopla>
any rom-o-matic solution? or dhcpd.conf special opt?
09:31tier1__ has quit IRC
09:32tier1__ has joined #ltsp
09:37tier1_ has quit IRC
09:42
<cliebow_>
yopla:so that is an pxe response you are looking at?
09:43moquist has joined #ltsp
09:43
<yopla>
cliebow: yes it gets IP from DHCP but that's all
09:44
I tried PXE abd DHCP method without result.
09:44
<ogra>
have a look at /opt/ltsp/i386/usr/share/initramfs-tools/scripts/ltsp-nbd ... sounds like a race condition (i.e. thin can being slow ...)
09:44
look for: # mount the readonly root
09:44
add a sleep above
09:44
<yopla>
I wonder if it can read next-server option.
09:45
<ogra>
rerun update-initramfs -u in the chroot and ltsp-update-kernels on the server after changes
09:45
<Q-FUNK>
ok, off to supper
09:45
bbl :)
09:48spectra has joined #ltsp
09:49
<cliebow_>
can you compare pxe versions between a working and nonworking 3com card??i have had mixed results with my bagoshiit network cards..
09:50
<yopla>
Actually my other TEST pc is intel card
09:50frownix has joined #ltsp
09:51* yopla go downstairs look for old floppies to try etherboot images.
09:52
<sbalneav>
Morning all
09:53
ogra!
09:53
Hey, back home?
09:53
<stgraber>
morning sbalneav
09:53
<sbalneav>
Hey stgraber!
09:54
<cliebow_>
Scottie!!!!!!
09:54
<sbalneav>
Hey cliebow_!
09:54
Recovered from our visit yet?
09:55
<cliebow_>
sbalneav, i am upgrading a g4 to gutsy to build a tarball..any comments??
09:55
<sbalneav>
\o/ for Ckuck!
09:55
err
09:55
Chuck
09:55
That's great, it would be fantastic to have tarballs for PPC posted somewhere for easy download.
09:56
<cliebow_>
sbalneav, i get so down when y'all leave....
09:56Egyptian[Home] has quit IRC
09:56
<sbalneav>
Ah, but we'll be back.
09:56
Much like McArthur :)
09:56
Or, McArthur park.
09:57
I understand it's frightening in the dark.
09:57
<cliebow_>
heh!!!
09:57
<sbalneav>
Sorry, it's MELTING in he dark.
09:58
I was thinking of the wierd al yankovic verision
09:58
:)
09:58
<cliebow_>
wierd al rocks!!
10:03Q-FUNK has quit IRC
10:07
<ogra>
sbalneav, yep, back home
10:08
<jcastro>
\o/ LTSP!
10:10
<cliebow_>
jcastro: \0/ jcastro!!
10:12
<jcastro>
how did the rest of the hackfest go? I had to leave early. :-/
10:14
<cliebow_>
jcastro: it went well once it got going...most everyone went out for breakfast..but everything was closed..
10:14
<jcastro>
ah
10:14
<cliebow_>
so didnt get back til afternoon
10:14
i was there quite early and got somne quality time with Gadi and warren
10:15
<sbalneav>
ogra: hey, I'm going to open up my hardy branches on launchpad tonight!
10:15
<jcastro>
pity we had to leave
10:15
<cliebow_>
though never resolved any of my issues
10:15
<ogra>
sbalneav, YAY
10:15
sbalneav, i#d like to discuss some stuff about the structure
10:15
<sbalneav>
Sure!
10:15
<ogra>
like
10:15
dropping the dist and debian dirs
10:16
<warren>
sbalneav, I did some talking with folks, and there is currently no interest in any "upstream" merging of initramfs tools. I am however doing heavy development of our mkinitrd tool, so there is some hope for the distant future if I can morph this toward a common goal.
10:16
<ogra>
moving the code up one level (as it was before) and use bzr export instead of fiddling with cp and tar manually from the Makefile
10:16
but warren has to agree
10:16
<warren>
sbalneav, key to this... our people don't want to use udev in initramfs.
10:16
<ogra>
the current structure has a lot of pointelss complexity we dont need imho
10:17
<warren>
ogra, dropping the dist and debian dirs? what is "dist"?
10:17
<ogra>
warren, do you know about bzr export and its options ?
10:17
<warren>
ogra, no, but willing to see patches.
10:17
ogra, show me how it works through a patch that replaces that part.
10:17
<sbalneav>
Ok, well, otavio wanted the debian dir dropped as well. I don't want to drop them until I'm sure we've got all the goodies out of them we need.
10:18
<warren>
ogra, I wouldn't call it pointless complexity when what is in the Makefile is very simple.
10:18
<ogra>
warren, current structure is: ltsp/dist/server(/)client .... we dont need the dist dir there
10:18
<warren>
ogra, bzr export can create a versioned tarball?
10:18
ogra, NAME-VERSION.tar.{gz,b2}
10:19
<ogra>
if the Makefile just calls bzr export you dont need to cp and tar (bzr creates upstream tarballs, bz2 etc as you like with the export command and cares for bzr status)
10:19
<sbalneav>
I'm assuming we're talking about maintaining the debian dir in a separate bzr tree?
10:19
<ogra>
bzr help export
10:20
<warren>
sbalneav, debian dir would be like our .src.rpm (spec file + patches)
10:20
<ogra>
see the --format option :)
10:20
<warren>
sbalneav, that should be maintained by the distro, not upstream
10:20
<ogra>
debian dir shuld be separate
10:20
<warren>
ogra, so ltsp/Makefile would handle make install and make dist?
10:20
make dist using bzr export
10:20
actually
10:20
make dist would...
10:21
1) ./autogen.sh
10:21
<ogra>
so i can either just merge debian and the upstream branch (which is the furure of ubuntu (no tarballs if possible) or use the tarball with identical content from the release branch
10:21
<warren>
Ubuntu doesn't need the tarball
10:21
<ogra>
we dont need autogen, do we ?
10:21
hmm, getltscfg might actually
10:21
<warren>
Each versioned tarball should have an equivalent branch tag
10:21
<ogra>
but beyond that its only scripts
10:21
<warren>
ogra, regarding autogen, whatever scotty decides.
10:21bjohnson has quit IRC
10:22tier1_ has joined #ltsp
10:22* ogra would happily avoid extra complexity
10:22
<warren>
ogra, during BTS we decided to keep ltsp (server and client) arch dependent because we want the flexibility to add more binary tools later.
10:22
OTOH...
10:22
hmm
10:22
<ogra>
the bzr-releas target in debian rules actually resembles what we need in the Makefile
10:22
*bzr-release
10:22
<warren>
ogra, sbalneav: maybe it makes more sense to put all binary tools into a separate package like "ltsp-tools" so we can keep "ltsp" as entirely arch neutral scripts.
10:23
<ogra>
have a look at debian/rules at the bottom, it should give you an idea what i mean
10:23
warren, that wont work
10:23
<warren>
why?
10:23
<ogra>
ltp-client *needs* getltscfg
10:23
<sbalneav>
I'm happy with whatever makes distros happy, myself :)
10:23
<warren>
ogra, then make ltsp-client require ltsp-tools-client
10:23
<ogra>
lets keep the bits together where they belong together
10:23
*sigh*
10:23
why make it even more complicated ?
10:24
<warren>
ogra, if you have a desire to keep autoconf stuff out of ltsp because it should be only scripts, we need to move all binary tools into another package.
10:24
ogra, I'm not talking slippery slope, I'm talking about an organization that we set now, that has to endure and be maintainable in the future.
10:24
<ogra>
right
10:25
<warren>
If we want "ltsp" to be arch-neutral scripts-only, then we should offload binaries into another package.
10:25
<ogra>
but i dont want to scatter bits that beloing together to much
10:25
<warren>
So what you're saying is that arch-neutral isn't important.
10:25
Maybe I agree.
10:25
sbalneav, any thoughts?
10:25
<ogra>
well, it wont work as " arch-neutral scripts-only"
10:25
you need that little arch depending bit all the time
10:25
<warren>
That's fine
10:26
<ogra>
so i think its rather better to keep the bits in their logical structure
10:26
<warren>
ogra, earlier you complained about auto* in ltsp, now you accept it?
10:26
<ogra>
if its only for getltscfg i'm fine
10:26
<sbalneav>
Well, all of this predicates on ONE binary. My question is: could we easily write a shell script equivalent to getltscfg and avoid the whole issue?
10:26
<warren>
Will it slow it down more?
10:26
<ogra>
if we have to have it for the 20 or so separate dirs we have i'd argue
10:27
<warren>
I haven't got bootchart to work on readonly-root yet, so I don't know what parts are slow.
10:27
<ogra>
warren, i doubt you would see a slowdown
10:27
<warren>
ogra, every little bit counts
10:27
<ogra>
i know
10:27
<warren>
getltscfg is only a config file parser?
10:27
<ogra>
but getltscfg is called once to parse lts.conf
10:27
<warren>
Here's the thing...
10:27
<sbalneav>
Well, what we call getltscfg all over the place, but what we really should do is just call it ONCE, and create a env-variable file that can be .'d when needed.
10:27K_O-Gnom has joined #ltsp
10:27
<ogra>
and to export all the vars
10:27
<warren>
even if we rewrite getltscfg in shell
10:27
<ogra>
should be easy to make a script out of it
10:27
<warren>
we WILL want tiny binaries in the future
10:28
tiny binaries too small to make into their own package
10:28
we need to put them somewhere
10:28
<ogra>
we could define that they need to be separate :)
10:28
<sbalneav>
It's basically a NAME=VALUE file NOW, the only thing that complicates it is the [Default] [MAC] [HOSTNAME] sections.
10:28
<warren>
It is perfectly logical to split out tiny binaries into a separate package.
10:28
<ogra>
but that everything that makes ltsp basically work has to be arch indep
10:28
<warren>
ogra, how does Ubuntu feel about python?
10:28
<sbalneav>
LOL
10:28
<ogra>
i love it :) and ubuntu as well
10:28
<yopla>
wooooo! Whith my first floppy etherboot PXE works great
10:29
<ogra>
but it provided to be slow
10:29
<warren>
Oh good. Red Hat and Ubuntu agree on something.
10:29
<ogra>
thats why we dropped the python driven ldm
10:29
<warren>
oh
10:29
<yopla>
ltsp for gutsy really rocks!
10:29
<ogra>
the speedups in gutsy are caused by two things ....
10:29
new filesystem structure ... (nbd/unionfs)
10:29
<warren>
ogra, sbalneav: do you agree that we will have tiny binaries in the future, and we should have a tools package for it?
10:29
<ogra>
and dropping of python ldm
10:30
warren, i agree, but we should have the ltsp source tree as a functional piece
10:30
<yopla>
ogra: ldm is now in C?
10:30
<ogra>
it should be able to do the basic stuff (booting a client into X)
10:30
every other bit can be separate
10:31
<sbalneav>
warren: Well, what I'd like to investigate is how complex it would be to write a sh/awk replacement for getltscfg. That gets called once. That way, the "ltsp" branch could be arch independent, and then all the arch dependent stuff could be in separate packages.
10:31
<ogra>
i dont want to have a non working tarball as upstream source
10:31
<warren>
sbalneav, if you do it, please profile it
10:31
sbalneav, compare the speed
10:31
<ogra>
if you download ltsp-$VER.tgz you should be able to get a working ltsp
10:32
(if your distro integration is done indeed ;) )
10:32
<warren>
sbalneav, ogra: if that's the case, then maybe we shouldn't care about arch
10:32
sbalneav, ogra: it isn't a big deal
10:32
<ogra>
right
10:33
<warren>
ok, so accept arch-binaries in ltsp
10:33
meaning it needs auto*
10:33
<ogra>
as soon as the source is up on LP i'll branch off and how you some patches for the dist dir stuff ...
10:33
<warren>
ogra, so would it work this way...
10:33
<ogra>
*show
10:33
<warren>
1) check if what you have checked out is current
10:33
<ogra>
it wont need auto if we rewrite getltscfg
10:33
which seems the way scott wants to go
10:34
<warren>
ogra, but what about the future when we will have new binary things?
10:34
ogra, where should that go?
10:34
<Gadi>
sbalneav: http://www.unixreview.com/documents/s=1344/uni1013549197040/
10:34
:)
10:34
<ogra>
they have to be either scripts or be in separate packages (but then not be required features)
10:34
imho you should always get a working ltsp base if you use the tarball alone
10:35
which means ltsp-server and tsp-client binaries ahouls be possible to be generated from that source and they should do the basic bits
10:35
ugh
10:35
<sbalneav>
Gadi: yeah, something like that :)
10:35
<ogra>
s/ahouls/should/
10:36
<sbalneav>
I like how the comments for the awk script are 5 times longer than the code :)
10:36
<ogra>
lol
10:36
<sbalneav>
Ahhh, awk
10:36
Is there nothing it can't do.
10:37
That may be just what we're looking for.
10:37
<ogra>
or sed :)
10:37
or a combo
10:37
<yopla>
how to active nbd swapping in gutsy?
10:37
<sbalneav>
yep.
10:37
<warren>
ogra, if you want "you should always get a working ltsp base if you use the tarball alone" to be true, then we need auto* in ltsp
10:37
<sbalneav>
I'll poke at it tonight, see if I can get jammcq involved.
10:37tier1__ has quit IRC
10:37
<ogra>
yopla, if you have less than 48M in the client it will be switched on automatically ... beyond that NBD_SWAP=True in lts.conf
10:38
warren, what for ?
10:38
<warren>
ogra, for future binaries
10:38
<ogra>
??
10:38
<yopla>
and for 128M we don't need swap?
10:38
<ogra>
yopla, not really
10:38
<warren>
ogra, there will undoubtedly be more binaries in the future
10:38
<ogra>
warren, *inside* the ltsp tree ?
10:38
<warren>
ogra, yes
10:38
<ogra>
not if we define they have to be non arch specific
10:39
its a matter of definition :)
10:39
<warren>
ogra, where else do we put stuff that link to dbus and PolicyKit?
10:39
<sbalneav>
Well, that's just a decision that we can make. We can either say "core ltsp has to be arch independent", or not.
10:39
<warren>
If core ltsp is arch independent, then we need a separate package.
10:39
<ogra>
warren, they are likely not ltsp but ltspfs specific
10:40* ogra cant imagine PolicyKit (which isnt in ubuntu/debian) be a default for anything else but localdev
10:40staffencasa has joined #ltsp
10:40
<sbalneav>
or, perhaps, ldm which is a separate package.
10:40
<warren>
ogra, PolicyKit can add users to /dev/fuse with setfacl on-demand, so we don't have to fuss with fuse groups or setuid.
10:40
<ogra>
same goes for ConsoleKit which will be a ldm feature
10:41
<yopla>
which is specific VAR for auto login?
10:41
<ogra>
warren, we dont have that prob in ubuntu/debian ...
10:41
<moquist>
sbalneav: I'm reasonably close on sshfsplus; it's taken some time to polish off my C tools.
10:42
<sbalneav>
yopla: LDM_USERNAME and LDM_PASSWORD
10:42
<ogra>
warren, and if fedora has it you can add an extra package
10:42
<sbalneav>
moquist: \o/
10:42
<yopla>
they can be in tftproot lts.conf?
10:42
<ogra>
i'm not aware of any distro apart fedora actually using PloKit
10:42
*Plo
10:42
ergh
10:42
Pol
10:42
<warren>
ogra, talking about the future
10:43
<ogra>
warren, it wont get into the debian/ubuntu based distros there are clear statements from either security teams about the design
10:43
<moquist>
sbalneav: I'm planning to create a LP project & bzr branch, and put a package in my PPA. Is that a sound plan?
10:43
<warren>
ogra, really?
10:43
hmm
10:43
<ogra>
warren, but i understand what you mean
10:43
<warren>
ogra, there was a lot of press around UDS and Policykit
10:43
<ogra>
(beyond the specific case here)
10:44
<warren>
OK, so are we declaring that ltsp core will be arch neutral forever?
10:44
<ogra>
did they redesign it to not be setuid wrappers ?
10:44
<sbalneav>
moquist: Don't think we need a ppa for it: it'll just become part of a "localapps" branch.
10:44
<ogra>
debian based distros use groups for everything o you dont need to have setuid bits for most parts
10:44
<warren>
ogra, it isn't setuid on our distro
10:44
<ogra>
one bit has to be, no ?
10:45
<sbalneav>
warren: ogra: lets say this: why don't we make it a GOAL to have base be arch independent, unless compelling reasons dictate otherwise in the future.
10:45
<ogra>
anyway, its not up to me to decide that ...
10:45
<warren>
ogra, not if a service is running as root and communicates over dbus
10:45
<ogra>
all i know is that the security teams in debian and ubuntu refused it until now
10:45
lets not get stuck on polkit :)
10:45
<warren>
ogra, it seems to drop a bunch of privs, and it is further confined here with strict selinux rules
10:45
yeah, ok
10:46* Gadi seconds sbalneav's motion - we should try never to be arch dependent on client or server
10:46
<ogra>
we dont use selinux either ;)
10:46
<warren>
ogra, let's talk about jetpipe as a simple example of structure
10:46
<ogra>
anyway, i think what we define now is not cast in stone and shoulnd apply furter than to the next ful tarball release
10:47
whic is 8.04 for me and the april/may release for you
10:47
debian follows its own schedule here ... we'd need info from vagrant
10:47
<warren>
ogra, jetpipe/Makefile "make dist" can do 1) check to see if changes are not checked in 2) ./autogen.sh 3) make NAME-VERSION.tar.{gz,bz2}
10:47
<ogra>
warren, printing isnt a mandatory feature to get a screen up on te client
10:47
neither is ldm
10:47rafaelcardoso has joined #ltsp
10:47
<ogra>
nor ltspfs
10:47
<warren>
printing?
10:48
<ogra>
jetpipe is a printserver
10:48
<warren>
ogra, no no no
10:48
ogra, I'm talking in general about the structure of the bzr repo
10:48
<ogra>
ah
10:48
<warren>
jetpipe/Makefile "make dist" can do 1) check to see if changes are not checked in 2) ./autogen.sh 3) bzr export to create NAME-VERSION.tar.{gz,bz2}
10:49
jetpipe/Makefile "make dist" can do 1) check to see if changes are not checked in 2) bzr tag NAME-VERSION 3) ./autogen.sh 4) make NAME-VERSION.tar.{gz,bz2}
10:49
?
10:49
<ogra>
sounds perfect
10:49
<warren>
OTOH...
10:50
ogra, should we make it tag in "make dist" or allow the developer to tag manually?
10:50
<ogra>
i dont really care abut tags ... if you want them, add them
10:50
<warren>
ogra, how do you build from bzr without tags?
10:50
ogra, does your builder bzr checkout whatever is at HEAD?
10:50
<ogra>
well, until now ltsp was a native debian package so there was no need to ...
10:50
<warren>
ok well
10:51
<ogra>
it pulled the version from debian/changelog
10:51
<warren>
If you want your package to be based on a specific upstream versioned release
10:51
then you'll need to use tags
10:51
<ogra>
right, feel free t add them then ... as i said, i'm not oppsed
10:51
<warren>
OK
10:51
hmm
10:52
<ogra>
hmm ?
10:52
:)
10:52
<warren>
Should we 'make tag' as a separate step in order to enable people to make versioned tarballs without tagging? (Testing purposes)
10:52
What we do in Fedora is slightly different
10:52
make dist (tags and creates tarballs)
10:52
make test-dist (create tarballs without tagging)
10:52
<ogra>
lets do that then
10:53
<warren>
either way, it needs to ./autogen.sh
10:53
<ogra>
meh
10:53
<warren>
well, I guess it doesn't *need* to, you can do that while building it...
10:53
<ogra>
my point is that most devs that are able to rovide script patches are not able to handle autofoo
10:54
<warren>
ogra, let scotty handle autofoo upstream
10:54
<ogra>
i dont want to make contributing harder than it is now
10:54
they wil need to handle it for testing etc
10:54
<warren>
We *need* autofoo in the binary packages.
10:54
<ogra>
right
10:54
which is fine
10:54
<warren>
We don't need autofoo in ltsp core
10:54
which is fine
10:54
<ogra>
since there is no way around that
10:54
<warren>
only thing Makefile in ltsp core does is:
10:54
<ogra>
bt we have choice for the core
10:55
<warren>
make install [DESTDIR=/somewhere] MODE={fedora,debian,etc}
10:55
<ogra>
why mode ?
10:55
<warren>
ogra, how else do you tell Makefile what distro specific stuff to install?
10:55
<ogra>
is there any ?
10:55
<warren>
ogra, I want make install to install the common things plus fedora specific plugins
10:55
ogra, yes there are
10:56
ogra, installing a chroot is completely different in Fedora
10:56
<ogra>
we use to install all plugins in the plugin dir
10:56fernando1 has quit IRC
10:56
<ogra>
right that should be handlesd by the plugins for ltsp-build-client
10:56
which are triggered by an lsb_release call
10:56
<warren>
I'll take another look at it
10:57
but I am thinking it makes it more maintainable to have distro-specific parts in separate files
10:57
<ogra>
i would like to be able to have a look at fedosa plugins to probably get ideas for my ubuntu plugins without having to fiddle with the tree
10:57
<warren>
some of which are installed depending on mode
10:57
ogra, ok maybe that's doable
10:57
<ogra>
same i think will be the other way around
10:57
<warren>
ogra, i'll look into it...
10:57
I have to go for now.
10:57
<ogra>
that was the initial idea of the plugin theme
10:58
<warren>
Can we see a draft of the new layout before it goes live?
10:58
<ogra>
you can just add links if you find other plugins valuable
10:58
thats an ancient layount
10:58
1.5 years old
10:58
or do you mean the tree layount of the source
10:58
<moquist>
sbalneav: why does sshfsplus belong in an LTSP branch? I see it as generically useful (though nobody but LTSP may ever care). I don't have strong feelings either way; I was just expecting it to stand on its own.
11:02
<ogra>
sbalneav, do you put the branch under ltsp-drivers so we can use the team for ACL management ?
11:02
<sbalneav>
Yes, I will.
11:02
<ogra>
and use its PPA for test building etc
11:02
yay, thanks :)
11:02
<sbalneav>
I'll work that out tonight.
11:02
<cliebow_>
moquist: this is sshfs via fuse on steroids??
11:03
<ogra>
i wonder if i should put the debian dir branch in there as well ...
11:03* ogra sends summoning thoughts to vagrant ....
11:04
<moquist>
cliebow_: well, not steroids. with a statfs call that returns real data.
11:05
<ogra>
sbalneav, hmm, jammcq added a guy called patrice dumas to the team ... do you know who that is ?
11:05
<sbalneav>
Hmm, the name rings a bell, but right atm I can't recall who he is.
11:06
<ogra>
if we use it for ACL it needs to be sure we only have committers we want in there
11:06
(i dont know him ...)
11:07
jcastro, any news about the new logo ?
11:08fernando1 has joined #ltsp
11:08
<ogra>
vagrant wants us to pull out the temes from ldm ...
11:08
*themes
11:08
that means we need a generic ltsp theme so a logo would be handy ;)
11:11mcfloppy_ has joined #ltsp
11:12
<warren>
sbalneav, will you be redoing the bzr layouts and rewriting getltscfg?
11:13
<ogra>
warren, i think we should make patches in our branches and propose the changes and scott has the final say on everything
11:13
so he has not all the work on his sholders
11:14
(and you can merge my suggestion, make changes and he then merges the imrpoved version upstream etc)
11:15
collaborative coding ;)
11:17
i want us all to be able to commit to te upstream branch but it should be clear that nobody merges stuff there without scott nodding it off so he only needs to review our changes but we do the major work and keep his back clear for imprvements
11:17
<sbalneav>
warren: Well, I'll sit down tonight, do some organization, and post to ltsp-devel. Are you on that list?
11:17
I won't touch the current branches till we get consensus.
11:18
I also want vagrantc and otavio's input.
11:18
<ogra>
it requires a tad of self discipline indeed, but eases work for all of us
11:18
<yopla>
when I put LDM_USERNAME and PASSWORD, the server doesn't start
11:19
(X server)
11:19
mmm no, ldm doesn't start
11:19
black screen with X cross
11:19
<ogra>
cut that down furter: the greeter doesnt start
11:19
:)
11:20
<Gadi>
yopla: its trying to log you in. make sure you don't use quoets around the username and password and make sure they are valid
11:20
<ogra>
(and its not suposed to)
11:20
<yopla>
:)
11:21
ldm.log says : In get_userid
11:22
<ogra>
see ~/.xsession-errors for that user
11:22
if it gets beyond the ssh parts
11:23tier1__ has joined #ltsp
11:24
<yopla>
nothing in it...
11:25
<ogra>
then /var/log/auth.log
11:26
<yopla>
ivegor a "-" in username
11:26
is taht a problem?
11:27
<ogra>
does the user work locally on the server ?
11:28
<Gadi>
yopla a dash should be no problem. but make sure you don't put quotes around the username in lts.conf
11:28
<yopla>
I verify
11:28
<ogra>
or the password
11:30
<yopla>
LDM_USERNAME=foy-02 LDM_PASSWORD=blabla
11:30
<ogra>
and foy-02 can log in with pw: blabla on the server locally ?
11:30
<yopla>
the local user works OK. Without those vars, I can log in manually in lts client
11:37
<warren>
sbalneav, ltsp-developer@lists.sourceforge.net right?
11:40
<sbalneav>
warren: correct
11:40tier1_ has quit IRC
11:40tier1_ has joined #ltsp
11:41
<warren>
sbalneav, saw the above about "make dist" and "make test-dist"?
11:42
<sbalneav>
No, I'll page back and look for it.
11:42
<warren>
sbalneav, make dist: check for changes and reject if changes are not committed; bzr tag NAME-VERSION; bzr export; (./autogen.sh if applicable); create NAME-VERSION.tar.{gz.bz2};
11:42tier1__ has quit IRC
11:43
<warren>
sbalneav, make test-dist: (./autogen.sh if applicable); create NAME-VERSION.tar.{gz.bz2};
11:43
sbalneav, test-dist creates a tarball of whatever is in the current directory, without respect for tags or if is even checked in or not. These tarballs are for test purposes.
11:43
<sbalneav>
ok
11:43
<warren>
sbalneav, actually...
11:44
sbalneav, make test-dist: (./autogen.sh if applicable); create NAME-VERSIONtest.tar.{gz.bz2};
11:44
sbalneav, a little more explicit
11:45
<outofrange>
Oh, I see why I can't log in.
11:45
MainThread 2007/11/13 09:45:16.4678 (sabayon-apply): Fatal exception! Exiting abnormally.
11:49
<ltsppbot>
"franzn" pasted "dhcpd.conf" (129 lines) at http://pastebot.ltsp.org/347
11:50
<sbalneav>
out to lunch for a run
11:50
bbiab
11:50
<Guaraldo>
sbalneav: What do you think about putting #ltsp-br as a channel to pastebot?
11:51
sbalneav: Good lunch...
11:56
<yopla>
when i modify lts.conf, I should alway ltsp-update-image?
11:59
<Gadi>
yopla: move the lts.conf to: /var/lib/tftpboot/ltsp/i386/lts.conf
11:59
then, you won't have to update the image every time
12:00
<yopla>
gadi: cool!
12:01
<cliebow_>
yopla:this is gutsy right?
12:01
<yopla>
cliebow_: yes
12:02
is tehre a quick way to disable sound on client?
12:02BGomes has joined #ltsp
12:02
<Gadi>
SOUND=N
12:02
<yopla>
my .xsession-error becomes hughe because of this.
12:04IRCzito has joined #ltsp
12:04K_O-Gnom_ has joined #ltsp
12:04K_O-Gnom has quit IRC
12:05K_O-Gnom has joined #ltsp
12:06
<yopla>
is it LDM_PASSWORD or LDM_PASSWD?
12:07K_O-Gnom_ has quit IRC
12:10
<yopla>
shoud LDM_SESSION be setup?
12:10
I use ~/.xsession
12:19
<jcastro>
ogra: I talked to andreas about it briefly but no one sounded too excited when I mentioned it in Maine.
12:19
ogra: is this something ltsp.org really wants?
12:21edgarin has joined #ltsp
12:21
<ogra>
jcastro, well, would you like the big blue square with the yellow L on the default theme ?
12:22
if we have to provide a default theme it should be nice and modern imho, reflecting the quality of the software ;)
12:22
<jcastro>
right
12:22
anything but the typewriter font. :p
12:22sonjag has joined #ltsp
12:23
<ogra>
heh, yes :)
12:23
but jammcq needs to agree ...
12:23
<jcastro>
are you guys going to do an entire theme?
12:23
yeah
12:23
<ogra>
he holds the trademark
12:23
<jcastro>
I'll talk to him about it tonight at our mug meeting
12:23
<ogra>
if we have a logo i volunteer to work a theme around it
12:23
(doing tht for all the *buntus anyway)
12:23
<jcastro>
right
12:25
<yopla>
hoho. it seems to be locale problem
12:26
file /etc/default/locale isn't set and pam doesn't like it
12:31yopla has quit IRC
12:38
<ltsppbot>
"ogra" pasted "ltsp tree structure change proposal diff" (137 lines) at http://pastebot.ltsp.org/348
12:38
<ogra>
warren ^^^^^
12:39tier1__ has joined #ltsp
12:39
<ogra>
whoops
12:39
<warren>
ogra, you can remove that #SPEC line
12:39
ogra, where did clean go?
12:40
<ogra>
there is missing a bzr-export in the dist: line
12:40
<warren>
oh, I see clean
12:40
<ogra>
i made it more explicit, since you might want to keep former tarballs
12:40
<warren>
ah
12:41
yes
12:41
<ogra>
do you want the spec file in the upstream tarball ?
12:41
<warren>
ogra, oh, so you went the "make tag" as entirely separate instead of "make test-dist" route?
12:41
ogra, no
12:41
ogra, remove the #SPEC line
12:41
not needed
12:42
<ogra>
test-dist: clean check-status bzr-export
12:42
<warren>
ogra, unless we have some other file that contains NAME and VERSION that we want to parse
12:42
<ogra>
dist: clean check-status tag bzr-export
12:42
<warren>
ogra, tag is part of dist
12:42
<ogra>
right
12:42
thats how i understood you
12:43
its not done yet
12:43
<warren>
k
12:43
ogra, so I guess scott will have to work your new Makefile into auto* so it becomes a hybrid of the two
12:44
<ogra>
he just needs to merge my branch
12:44
where would auto* come in there ?
12:44
<warren>
ogra, because auto* generates its own makefile
12:45
ogra, jetpipe, ltspfs, ldm each use auto tools
12:45
<ogra>
i thought we wanted to go towards scripts with getltscfg
12:45
<warren>
ogra, ./autogen.sh creates configure and Makefile
12:45
ogra, yes, autogen wouldn't be in ltsp core
12:45
<ogra>
oh, you want to be in one dir with all the branches and have an ./autogen there ?
12:46
to build all at once ?
12:46
<warren>
huh?
12:46
ogra, the purpose of dist/ was to separate the "make dist" Makefile from "make install" Makefile
12:46
<ogra>
i dont understand where auto* touches the ltsp core Makefile here
12:46
<warren>
auto* doesn't touch the ltsp core Makefile
12:46
<ogra>
ah
12:47
<warren>
ogra, I'm talking about keeping a consistent structure between ltsp, jetpipe, ltspfs, ldm
12:47
so make install and make-dist behave exactly the same between them
12:47
<ogra>
ah
12:47
<warren>
ogra, so maybe it is worthwhile to keep dist/
12:47
<ogra>
i dont see its purpose
12:47
<warren>
so the "make dist" Makefile is out of the tarball. It makes sense, because it is irrelevant to make install.
12:48
Not only irrelevant, but it makes it more complicated.
12:48
<ogra>
i'D rather drop it from all other branches (in case its there as well, i didnt check)
12:48
<warren>
and replace it with what?
12:48
you'll have an even messier and complex Makefile
12:48
ogra, the equivalent of what you're making for ltsp core will need to be written in auto*
12:48
<ogra>
the other way you have two of them
12:48fernando1 has quit IRC
12:48
<ogra>
and a more confusing dir structure
12:49
<warren>
It isn't more confusing
12:49
Just one more level
12:49
Is that really difficult to deal with?
12:49
I change my mind, it makes a lot more sense to keep dist/
12:50
<ogra>
meh
12:50
<warren>
Keep the "make tag dist test-dist" out of the tarball where it can complicate the real Makefile.
12:50
<ogra>
ogra@laptop:~/devel/hardy/ltsp$ wc -l Makefile
12:50
39 Makefile
12:50
<warren>
And there is really no reason why we need to use bzr export.
12:50
<ogra>
can be cut down to 20 lines ...
12:50
i dont see how its more complicated
12:51
<warren>
ogra, stop thinking about only ltsp core
12:51
ogra, the other repos use autotools, which generates its own Makefile
12:51
<ogra>
they are different source trees
12:52IRCzito has left #ltsp
12:52
<warren>
The purpose of making the same STRUCTURE upstream was for consistency
12:52
<ogra>
if someone comes to you with a cool extension that uses quilt, do you refuse him then ?
12:52
<warren>
You can use "make dist" in the base of every repo to make a tarball.
12:52
<ogra>
i dont get that argument
12:52
<warren>
How does dist/ make it more difficult for you?
12:52
<ogra>
i thought the purpose of having separate source trees was exactly to be able to do such stuff
12:52rjune_ has quit IRC
12:53
<warren>
please be more specific
12:53
<ogra>
it just looks very ugly and i dont see how it helps to have to maintain two files and one more dir
12:53
<warren>
sbalneav, I change my mind, I don't see compelling benefits to getting rid of dist/ and using bzr export. It introduces new drawbacks as well.
12:53
It isn't ugly.
12:53
This is closer to how many other upstream projects work.
12:54rjune_ has joined #ltsp
12:54
<warren>
ogra, how difficult is it for you really? During debian package build, bzr checkout the repo, install whatever is in dist/
12:54
ogra, you're making a big deal out of nothing.
12:55
<ogra>
i see us just wasting capabilities of bzr and adding extra code we need to maintain ourselves
12:55
<warren>
ogra, you could even have your debian package build checkout a specific tag from upstream, or from your own branch
12:55
This is NOT complicated.
12:55
<ogra>
beyond that i know no upstream project that puts an extra dist dir in place
12:55
<warren>
ogra, here is a drawback of bzr export
12:56
ogra, you can't ./autogen.sh prior to creating the tarball.
12:56
ogra, typical upstream tarball releases have already done that step
12:58
Just because bzr export exists does not mean we NEED to use it.
12:58
<ogra>
it makes life easier
12:58
and i'm sure there is a way for exporting autogen stuff alongside
12:58
<warren>
Not necessarliy
12:59IRCzito has joined #ltsp
12:59
<warren>
ogra, it makes a lot of sense for upstream LTSP to have a consistent repo structure with all of its modules. So they can use the same commands on each repo to make a tarball release.
12:59
ogra, yes it isn't a problem for ltsp core, but it is for all the other modules.
12:59
ogra, Makefile doesn't even exist to use "make dist" in the pristine state before autogen.sh
13:00tier1_ has quit IRC
13:00tier1_ has joined #ltsp
13:00
<ogra>
still, i dont see the use for two makefiles and one extra dir
13:00
even if we'd use cp and tar manually
13:01
<warren>
I just explained why
13:01
<ogra>
its two more targets in the makefile
13:01
<sbalneav>
Back
13:01
<warren>
You don't even HAVE a Makefile in a pristine checkout
13:01* sbalneav reads up
13:01
<warren>
you aren't listening
13:02
ogra, you are making this very difficult for no good reason
13:02
<ogra>
i see what you write ... i dont see your prob here
13:02
<warren>
ogra, you can easily adapt to this.
13:02
<ogra>
if all modules behave like any other upstream project they wont differ from ltsp core
13:02
<sbalneav>
Is the argument that most upstreams have a dist dir?
13:02
<ogra>
do they ?
13:02
<warren>
sbalneav, it depends on how they do a release
13:02
<sbalneav>
No, they don't
13:02* ogra hasnt seen that
13:03
<warren>
sbalneav, nevermind what other people do, we're trying to decide on a standard for our repos
13:04
<sbalneav>
Wow
13:04
<ogra>
sbalneav, my proposal is to have one Makefile and no dist dir but everything in the toplevel (as we used to have it)... warren wants a dist dir and two Makefiles, one for make dist and the other for the rest
13:04
<warren>
sbalneav, is this an agreeable goal? want to use standard commands on a standard structure with all LTSP repos to make versioned releases
13:04
<ogra>
i simply doe see the reason for maintaining that extra bits instead of adding two targets to the main Makefile
13:05
<warren>
sbalneav, if we have a standard structure, then it becomes complicated to do "make tag dist test-dist" from the same Makefile as "make install" because of autotools in every package except ltsp core.
13:05
<ogra>
warren, i didnt argue about a common standard (even though i wouldnt make that mandatory ... if someone comes with ruby code thats cool i wouldnt reject him because he has a scripted installer)
13:06
i only argue abut having the extra bits
13:06
<warren>
sbalneav, for one, in the pristine checkout you don't even have a Makefile yet.
13:06
sbalneav, meaning you can't easily type "make dist" without running another command, which is inconsistent with ltsp core.
13:07
sbalneav, another problem is that bzr export cannot make a tarball *after* autogen.sh has run.
13:07
<ogra>
please tell me where that would be inconsistent if the structure is the same everywhere ?
13:07
ignore bzr for now
13:07
<warren>
ogra, you are ignoring the fact that Makefile doesn't exist yet.
13:07
<ogra>
in the autotools it has to be added to Makefile.am
13:07
in core its in the static Makefile
13:08
whats the prob here ?
13:08
<warren>
1) checkout ltsp
13:08
make dist
13:08
You have tarballs.
13:08
2) checkout jetpipe
13:08
make dist
13:08
Doesn't owrk.
13:08
<ogra>
??
13:08
why ?
13:08
<warren>
ogra, Makefile doesn't exist yet
13:09
<sbalneav>
I'm failing to see where it says we MUST make dist to make a tarball?
13:09
Can't we just have a
13:09
make-dist.sh in the root
13:09
<ogra>
++
13:10
<sbalneav>
that does the right thing? (either a bzr export, or tar cvf, or what have you?)
13:10
<ogra>
sunds like a compromise
13:10
*sounds
13:10
<warren>
hmm
13:10
that might work
13:10
I don't care what it does
13:10
<sbalneav>
That's me. The king of compromise.
13:10
<warren>
I only care that it follows the same rules.
13:10
<sbalneav>
ok
13:10
<warren>
so we have make-tag.sh
13:11
make-test-dist.sh?
13:11
kind of ugly to have extra files sitting there
13:11
but OK
13:11
<sbalneav>
or, a make-something.sh make-tab
13:11
<ogra>
make-dist --tag
13:11
make-dist --test
13:11
<sbalneav>
make-something.sh test-dist
13:11
yeah
13:11
that
13:11
<ogra>
make-dist release
13:11
<warren>
a little slower to type =)
13:11
<sbalneav>
phht.
13:11
<ogra>
heh
13:12
<warren>
./make-dist (with no arguments) prints help
13:12
<sbalneav>
So, is that acceptible to you two?
13:12
<ogra>
yep
13:12
<cliebow_>
sbalneav, You rock~!
13:12* ogra looks at warren
13:12
<warren>
./make-dist -r or --release check, tag, autogen, tarball.
13:12
<Guaraldo>
sbalneav: What do you think about putting #ltsp-br as a channel to pastebot?
13:12
<sbalneav>
Remember, I'm canadian. We're the peacekeeper people :)
13:12
<warren>
./make-dist -f or --release check, autogen, tarball.
13:13
oops
13:13
./make-dist -f or --force check, autogen, tarball.
13:13
./make-dist -t or --tag
13:14
<sbalneav>
Ok, vote time, so I know where I stand
13:14
sbalneav++
13:14
<ogra>
+1
13:14
<warren>
wait a sec
13:14
<sbalneav>
sleep(1)
13:15tier1__ has quit IRC
13:15
<warren>
./make-dist
13:15
(print help)
13:15
./make-dist -r
13:15
./make-dist --release
13:15
(check; tag; autogen; NAME-VERSION.tar.whatever)
13:15
./make-dist --force
13:15
(check; autogen; NAME-VERSIONtest.tar.whatever)
13:15
./make-dist --tag
13:15
(check; tag)
13:15
./make-dist --clean
13:15
(clean)
13:16
<ogra>
cool
13:16
<cliebow_>
crap..ldap borked bootup againb...
13:16* sbalneav cuts and pastes
13:16
<warren>
oops
13:17
sbalneav, matching -f -t -c for the other commands
13:17
<ogra>
that should be one generic script we can copy around
13:17
<warren>
sbalneav, I just like having something quick to type
13:17
ogra, yes, easily
13:17
<ogra>
so we only need to maintain it once
13:17
<warren>
ogra, if make-dist can read NAME-VERSION from somewhere in the source
13:17
horray
13:17
<ogra>
(i.e. it hould autodetect if autogen is needed or not etc)
13:17
*should
13:17
<warren>
ogra, ++
13:17scrapbunny has joined #ltsp
13:18
<ogra>
and indeed read the release version from somewhere
13:18
<sbalneav>
ok, we'll hope vagrantc and otavio like it.
13:18
<warren>
ogra, easy enough [ -x autogen.sh ] && ./autogen.sh
13:18
<ogra>
yeah
13:18
<warren>
where shall the NAME-VERSION be read from?
13:18
<ogra>
sbalneav, i'm pretty sure they do, i discussed with him yesterday
13:19
NAME should be basename $(pwd)
13:19
<sbalneav>
I'd say name-version should come from configure.ac
13:19
<warren>
VERSION?
13:19
ah
13:19
yeah
13:19
ok, this will be awesome
13:19
<sbalneav>
except in ltsp's case
13:20
<ogra>
sbalneav, right
13:20
<sbalneav>
where we'll just arbitrarily make sh*t up like we always do.
13:20
<warren>
sbalneav, make-dist --clean (or -c) can remove all the files generated by autogen.sh
13:20
sbalneav, ?
13:20
<ogra>
it can call make clean :)
13:20
if Makefile exists
13:20
<sbalneav>
the version number for ltsp, where there's no configure.ac
13:20
<warren>
ogra, it depends how clean you want
13:20
<ogra>
indeed
13:20
<warren>
ogra, make clean wont delete itself =)
13:20
<ogra>
bzr revert ;)
13:21
<warren>
make-dist can conditionally look for the name-version
13:21
<ogra>
sould get you the original state
13:21
<warren>
single script
13:23
<ogra>
heh
13:23
its easier ...
13:23
bzr export ../$NAME-$VERSION
13:23
cd ../$NAME-$VERSION
13:24
[ -x autogen.sh ] && ./autogen.sh
13:24
cd ..
13:24
tar .........
13:24
that way your original branch stays clean
13:24
<warren>
why ..?
13:25
<ogra>
because you run autogen (if there) in the exported branch
13:25
<warren>
.. might not be safe
13:25
you can't make assumptions about where they did a checkout of the repo itself
13:25
safer to create a tempdir within the pwd
13:25
and delete it after the tarballs are creatd.
13:25
<sbalneav>
OK, I got a printer problem downstairs, I'm AFK for a bot.
13:25
err bit
13:25
<ogra>
fine with me, i dont care where the dir is
13:25
<warren>
k
13:26
brb, meeting
13:26
<ogra>
even though you end up with cruft in the dir if something breaks
13:26
that needs to be adressed
13:27
probably we should just use /tmp
13:27
i think its a safe assumption that exists and we have write access ...
13:27
and you prevent polluting the original branch
13:29plamengr has quit IRC
13:31
<scrapbunny>
i had to do a fresh install on my edubuntu 7.10 server and I am now getting this error on my client computer " syntax error in /etc/lts.conf, line=2" my lts.conf is in var/lib/tftpboot/ltsp/i386
13:31
<ogra>
paste it
13:31
!pastebot
13:31
<ltspbot>
ogra: "pastebot" is The LTSP pastebot is at http://pastebot.ltsp.org. Please paste all text longer than a line or two to the pastebot, as it helps to reduce traffic in the channel. A link to the content will be pasted in the channel.
13:33
<scrapbunny>
ogra- sorry was that for me or someone else?
13:33
<ogra>
for you
13:33
paste it to the pastebot
13:33
<scrapbunny>
ok will do
13:33plamengr has joined #ltsp
13:39
<ltsppbot>
"scrapbunny" pasted "XSERVER=i810 X_VIDEORAM=4096 X" (4 lines) at http://pastebot.ltsp.org/349
13:40
<ogra>
scrapbunny, there is no section at all in there
13:40
add:
13:40
[default]
13:40
as the first line
13:41
<scrapbunny>
thanks
13:41
<ogra>
sound should be on by default, no need to specify it
13:41
<scrapbunny>
will try that and check back
13:41
what about local devices?
13:42
<ogra>
is that feisty or gutsy ?
13:42
<scrapbunny>
i had"LOCAL_DEVICE_01=/DEV/HDC:CDROM
13:42
GUTSY
13:42
<ogra>
localdev is enabled by default there
13:42
<scrapbunny>
great thanks
13:42
<ogra>
no need to specify it in lts.conf
13:46
<cliebow_>
ogra: ltsp-server-standalone...claims it is latest version for ppc....should i be looking at cd
13:46
5.0.39
13:46
<ogra>
thats the newest
13:46
<cliebow_>
cool..so ill build-client
13:50fernando1 has joined #ltsp
13:50
<scrapbunny>
that is what i was missing, such a dork :)
13:50
<cliebow_>
happens to the worst of us 8~)
13:51
<scrapbunny>
now i need to bulk add 750 users. this is what i was trying to use : http://www.vcsvikings.org/docuwiki/cgi-bin/moin.cgi/ManagingUsers#fromlinuxbulkmanual
13:52
is there a better/easier way in ltsp 5?
13:55
<moquist>
When an LTSP server reboots and the NBD swapfiles go away, does that necessitate restarting the TCs as well?
13:56
<cliebow_>
scrapbunny:moquist wrote a lot of that
13:58
<ogra>
moquist, yes
13:58
well, the clients should notice theor swap file going away and handle that
13:59
<moquist>
Right, so I really, really don't want to start automating a nightly reboot of the LTSP servers in order to clean up runaway tuxmath & tuxtype.
13:59
<ogra>
but if you use swap because ram is low you will then likely run out of ram
13:59
<moquist>
ogra: OK, so TCs with enough RAM should probably be OK.
13:59rafaelcardoso has quit IRC
13:59
<moquist>
Do they get new swapfiles when the server comes back up?
13:59
<ogra>
beyond the fact that your root is gone as well, yes
13:59
<Gadi>
moquist: but if its gutsy, your rootfs goes away, too
14:00
<scrapbunny>
so moquist- will smbldap-useradd-bulk script still work?
14:00
<ogra>
Gadi, beefy clients should handle tat
14:00
<moquist>
Gadi, ogra: I need to learn more about nbd. Are connections reestablished when the server comes back up?
14:00
<ogra>
there is no reason to read or write after the clients are up
14:00
no
14:00
<cliebow_>
scrapbunny, you are using ldap for authentication??
14:00
<ogra>
you need to reinitiate
14:00
same as with nfs
14:00
<moquist>
scrapbunny: sure, if you install smbldap-tools, which is installed as part of my hackish script thing.
14:01
scrapbunny: hang on
14:01
scrapbunny: I'll point you to a repo
14:01
<ogra>
if the protmap connection changes you eed to remount
14:01
or if the server gets restarted
14:01
<moquist>
ogra: and do the TCs reinitiate?
14:01
<Gadi>
but NFS handles all that itself
14:01
with NBD, it should too
14:02
BUT, if your rootfs img changes, then you can have weird issues as well
14:02
<ogra>
the thing is that unionfs doesnt like if you rebuild the stuff underneath it
14:02
<Gadi>
(ie if you do an ltsp-update-image)
14:03
<ogra>
so even if you could remount after a server reboot that would moan at least
14:03
if not even break
14:03
<moquist>
Hmm. OK.
14:03
Right now we only have feisty in production.
14:04
We really need a solution for these runaway processes. I know the problem is that the upstream projects have bugs, but as long as there's one upstream project that has such a CPU-hogging bug it affects the entire server the same way.
14:04
Right now I log in when they say everything is slow and see runaway processes and kill them manually.
14:05* Gadi agrees with moquist
14:05
<ogra>
we should have post session scripts for ldm ... like the rc.d ones we have now for autostarting stuff
14:06* Gadi has some users on gutsy and even now with tracker removed, I still get black screens and freezings
14:06
<ogra>
so you can put killer scripts in for every broken app and remove them if fixes show up
14:06
<Gadi>
tracking down the apps is sometimes the hard part
14:07
and gets harder when the apps are, say, gconfd or bonobo :)
14:07
<ogra>
beyond that i still miss to se a proper cleanup script that makes sure only apps belonging to the current session get killed
14:07
<Gadi>
as killing those is bad
14:07
moquist's xterminator does a good job
14:07
<ogra>
it kills on a per user base still
14:08
i want to be sure it walks up the processtree and kills really only stuff from that current session
14:08
<Gadi>
right - well processes can always dissociate from the session
14:08
<ogra>
the user might run important stuff in screen on another terminal or so
14:09
just pkill -U isnt the answer here
14:09* Gadi nods
14:09* Gadi feels like this problem has only gotten worse the past few releases
14:09
<Gadi>
perhaps more is registering with bonobo
14:09
<ogra>
according to gnome upstream the dbus, bonobo and gconf issues are fixed now
14:10
<Gadi>
and hence more potential for desktop-munging problems
14:10
lol
14:10
they dont have my users :)
14:10
and prolly dont test much against ltsp
14:10
<ogra>
ltsp shouldnt matter here
14:11
we just run Xsession from the distro
14:11
that script should handle it
14:11
<Gadi>
on a remote box
14:11
<ogra>
well, on a local shell login with forwarded X
14:11
<Gadi>
with users potentially logging in from multiple boxes at different times
14:11
<ogra>
which is clearly not suggested/supported by us
14:12
<Gadi>
well we support logging in, logging out, logging in some place else
14:12
<ogra>
ah, right, that we do
14:12
<Gadi>
and that can be sufficient to cause gconf a headache
14:12
:)
14:12
at times
14:12
specially if old gconfd processes do not die
14:12
<ogra>
if gconfd properly shuts down it doesnt ... :)
14:12edgarin_ has joined #ltsp
14:12
<ogra>
heh
14:12edgarin has quit IRC
14:12
<Gadi>
zactly
14:13
gconf has become a bit to Windows-registry-ish
14:13
:)
14:13
*too
14:13
<cliebow_>
sbalneav, making little indians on powerpc....
14:13* ogra disagrees
14:14
<ogra>
its a well organized dotfile replacement imho
14:14* Gadi shrugs and thinks of the user reporting a "black screen" that just had to be pkilled
14:14
<Gadi>
:)
14:14
<mcfloppy_>
how does a ltsp5 client connect to the server? X :0 --query server01?
14:14
<moquist>
mcfloppy_: ssh -X server
14:14
<ogra>
mcfloppy_, ssh -X
14:14
<moquist>
ogra: beatcha ;)
14:15
<ogra>
or ssh user@server "DISPLAY=$DISPLAY Xsession"
14:15
<mcfloppy_>
ogra sorry, i mean ltsp 4.2
14:15
<ogra>
4.2 uses something along these lines iirc, yes
14:17
<mcfloppy_>
ok
14:17
cause i've now running a 2.4kernel on my 8363 netvista and start manually "X :0 --query server01" after booting
14:18
but i have the impression that these clients are slowlier then the other Dell Optiplex GX1 based clients
14:19
<cliebow_>
the netvistas are pretty fast in 4.1 for a 200 meg client
14:19
<mcfloppy_>
cliebow_ ok... but where is my mistake?
14:19
<cliebow_>
what prob are you having??
14:20
<mcfloppy_>
i use the image from http://linux-2200.berlios.de/ and boot over nfs...
14:20
<scrapbunny>
i have my username file ready and i installed smbldap tools but what folder are the scripts located in?
14:20
<mcfloppy_>
then i start manualy "X :0 --query server01"
14:20
<moquist>
scrapbunny: there are lots more configuration steps to take. Do you actually need Samba & LDAP?
14:21
<cliebow_>
so you are not using 4.1 then??
14:23outofrange has left #ltsp
14:23
<cliebow_>
sbalneav: http;//169.244.3.137/powerpc.img.gz
14:24
<scrapbunny>
no I just want to be able to bulk add the users :)
14:24
and have a stable setup for 32 thin clients to be logged in and running
14:25
<moquist>
scrapbunny: I don't think you should use Samba & LDAP, then.
14:25
<mcfloppy_>
cliebow_ no... i have ltsp5
14:25
<moquist>
scrapbunny: how many users do you have?
14:25
<scrapbunny>
750
14:26
<mcfloppy_>
but the 8363 dont boot with 2.6er kernel
14:26
<cliebow_>
yeah..and you have no udev which ltsp5 depends on..
14:26
<ogra>
and ltsp5 will likely break with 2.4 kernel
14:26
<mcfloppy_>
yes
14:26
<ogra>
for gutsy even worse, there is no squashfs or unionfs
14:27
or nbd
14:27
<moquist>
scrapbunny: hang on
14:27
<mcfloppy_>
so i boot the device and open a xtra x11 forwarding
14:27
but why it is so slow?
14:27
<scrapbunny>
i have the file set up with their Firstname,Lastname,username,gid,shell,home,quota,password where quota is model user
14:27
<ogra>
you have no devices
14:27
<moquist>
scrapbunny: in what form do you have your user data? spreadsheet with first & last names?
14:27
<scrapbunny>
thanks moquist
14:27
<ogra>
no stdin/stdout, /dev/null ...
14:27
no llopback netwoek device
14:27
you simply have no udev
14:28
<cliebow_>
scrapbunny, juust use a 4.1 "chroot"
14:28
<ogra>
which cares for all of them
14:28
<scrapbunny>
i have it in spreadsheet and as a text file with commas
14:28
<moquist>
scrapbunny: webmin has security risks you should know about, but you could install it to upload your users and then uninstall it with minimal risk.
14:29
scrapbunny: wget http://prdownloads.sourceforge.net/webadmin/webmin_1.380_all.deb
14:29
<mcfloppy_>
webmin? brrrr ^^
14:29
<moquist>
scrapbunny: sudo dpkg -i webmin_1.380_all.deb
14:29
scrapbunny: sudo apt-get -f install
14:29
scrapbunny: https://localhost:10000
14:29
<mcfloppy_>
but even bether then confixx ;)
14:29
<moquist>
scrapbunny: there are instructions in there from then on
14:29
<ogra>
sudo gedbi webmin_1.380_all.deb
14:29
;)
14:29
<moquist>
scrapbunny: then sudo apt-get remove webmin :)
14:30
<scrapbunny>
great thanks sooooo much
14:30
<ogra>
saves you the two install steps
14:30
<moquist>
gedbi?
14:30
<ogra>
gdebi :)
14:30IRCzito has quit IRC
14:31
<ogra>
man gdebi ! :)
14:31
<cliebow_>
moquist:i bet you could build a script in about five minutes to bulk add users 8~)
14:31
<moquist>
scrapbunny: ogra's is better. :)
14:31
<ogra>
its the same
14:31
but saves yu to sudo apt-get -f install afterwards
14:31
<scrapbunny>
ogra- so i sudo gdebi webmin_1.380_all.deb and then go to the link moquist gave right?
14:31
<ogra>
right
14:31
<moquist>
cliebow_: Yeah, but I'd be the only one who could use it. :p
14:32
<scrapbunny>
and when i am done sudo apt-get remove webmin ?
14:32
<moquist>
scrapbunny: yeah, recommended.
14:32ace_suares has joined #ltsp
14:33
<ogra>
sudo apt-get remove --purge webmin
14:33
;)
14:33
make sure *all* cruft is gone
14:34
<scrapbunny>
thanks ogra
14:34
<cliebow_>
sblaneav:any sesne tarring up the chroot for powerpc?
14:34
<scrapbunny>
i am guessing the website will work after i install wedmin
14:34
<cliebow_>
/sesne/sense/g
14:35
<ogra>
cliebow_, in gutsy just use the .img file
14:36
<mcfloppy_>
cliebow_, ogra, is there a "faster" or "bether" way as running "/usr/X11R6/bin/X :0 -query 192.168.0.200" after boot?
14:36
<ogra>
its already compressed
14:36
<cliebow_>
yep: i gzipped it and url is
14:36
sbalneav: http;//169.244.3.137/powerpc.img.gz
14:38BGomes has quit IRC
14:39Missionary has joined #ltsp
14:40
<sbalneav>
cliebow_: cool
14:40
<cliebow_>
Finally...something useful to do..8~)
14:41
<Missionary>
can anyone offer any advice on getting LTSP5 to work with my IBM Netvista N2200 thin clients?
14:42
<cliebow_>
Missionary, is there an echo in here?
14:42
<Missionary>
sorry, don't understand.
14:42
<cliebow_>
so far we havnt gotten a successful boot of 2.6 kernel..so 2200 are relegated to ltsp4.1
14:43
Missionary, talk to mcfloppy_ 8~)
14:43
<Missionary>
oh. darn.
14:43
is anyone working on getting 2.6 to boot?
14:44
<cliebow_>
there have been a few promising rumors...
14:44
I have been for a year or so..
14:44
maybe 2
14:45
<Missionary>
pardon my ABSOLUTE ignorance, but does ltsp5 require the this clients to boot 2.6 or can it be made to work with 2.4?
14:45
<cliebow_>
we tried the generic ltsp5 kernel at ltsp by the sea last weekend..
14:45
no dice..
14:45
ltsp4.2 and 5 depend on udev which is not present in 2.4
14:45
<ogra>
ltsp5 uses a lot technology thats only available in 2.6
14:45
<cliebow_>
4.1 works pretty well
14:46
just have to smack the clients up side of the head once in a while to get x
14:46
<Missionary>
should I dump the 2200s (i got a few very cheaply) and stick to something more suitable?
14:47
or try 4.1 instead?
14:47
<ogra>
4.1 it pretty unmaintained
14:47
<cliebow_>
2800 willboot a 2.6
14:48
<ogra>
get the upstream guys to work on netvista support in 2,6
14:48
*2.6
14:48
<cliebow_>
4.1 is totally unmaintained...no local devices..no sound..
14:48
<ogra>
would be the most helpful contribution ;)
14:48
<cliebow_>
i did have rangerpb at ibm struggle with em for a week or two..
14:48
<ogra>
(upstream == kernel.org in this case)
14:48
<mcfloppy_>
hehe Missionary
14:48
:p
14:48
i ve the same problem
14:49
<cliebow_>
ahhh!im afraid the "bios" in 2200 is so wonky..it prohibits much
14:49
<Missionary>
even the latest 2002 version?
14:49
<cliebow_>
wont help..
14:50
it uses no known protocol and my hunch is the bios doesnt create space enough to load any large kernel
14:50
<Missionary>
so - bottom line: N2200s don't/won't currently boot to ltsp5, no matter what I do?
14:51
<ogra>
unless you find the magic solution ...
14:51
<cliebow_>
pretty much right..unless i come up with some breakthrough
14:51
<Missionary>
darn
14:51
<ogra>
cliebow_, snap :)
14:51
<mcfloppy_>
Missionary i have something found with 2.6 and netvista2200
14:51
<Missionary>
anyone want to buy 3 x 2200s?!
14:51
<cliebow_>
Even Gadi couldnt come up with anything 8~)
14:52
Mc:huh???
14:52
<mcfloppy_>
http://projects.kgmoney.net/
14:52
<cliebow_>
hmmm
14:52
<mcfloppy_>
http://sourceforge.net/projects/linux2200-26/
14:52
hehe
14:53
<cliebow_>
i messed a little with this already...
14:53
not to say it wont work...this is the "promising" i spoke of
14:53
<mcfloppy_>
hrhr
14:54Gadi has left #ltsp
14:54* ogra wonders if there is any source published
14:55
<cliebow_>
it gives us little detail...cant tell if it is tagged with an initrd..or expects a seperate initramfs??
14:56
<mcfloppy_>
hmm but what is this for an evil hack?
14:57
<ogra>
who could tell ...
14:57
<mcfloppy_>
with the hexchange from 2 to 1 on position 2c
14:58sonjag has quit IRC
14:59
<cliebow_>
i emailled him...never can tell...
15:02
<mcfloppy_>
hehe
15:03
<Missionary>
can anyone recommend a decent (sheap) thin client that connects easily to ltsp5?
15:03
cheap!
15:04
<cliebow_>
ogra: this mean anything to you??
15:04
To get your own compiled kernel working on the N2200 you have
15:04
to take the vmlinux-file form arch/i386/boot/compressed and
15:04
patch the elf program-header-count to 1 (offset 0x2C).
15:04
<ogra>
no
15:05
and thats bad stuff
15:05
dont poke around in binaries
15:05
<cliebow_>
loks uncompressed huh
15:05
<ogra>
yeah, that as well
15:06
<mcfloppy_>
hrhr
15:06
but i tried this... every selfmade kernel starts up with this hex hack
15:06
^^
15:07bobby_C has joined #ltsp
15:08
<ogra>
mcfloppy_, what about the ltsp5 kernel ?
15:09
<mcfloppy_>
ogra i havn't the original uncompressed vmlinux image
15:09Guaraldo has left #ltsp
15:10Barbara_ghost has joined #ltsp
15:10
<ogra>
hmm
15:10barbara has quit IRC
15:10Barbara_ghost is now known as Barbara
15:10barbara_ has joined #ltsp
15:11Guaraldo has joined #ltsp
15:13
<cliebow_>
i have somewhere.."an" original uncompressed 2.6 image from december of 2005..
15:14slidesinger has joined #ltsp
15:14
<ogra>
objcopy -I binary -O elf32-i386 -B i386 vmlinuz-2.6.22-14-i386 vmlinux
15:14
will uncompress it afaik
15:15Barbara has left #ltsp
15:16
<cliebow_>
hmmm...
15:18
gottaz give it up til tomorrow..
15:18K_O-Gnom has quit IRC
15:18cliebow_ has quit IRC
15:19rjune__ has joined #ltsp
15:20rjune_ has quit IRC
15:23
<barbara_>
ogra, you on?
15:23
i had a question about ltsp and compiz. I am running gutsy and chrooted into ltsp dir installed nvidia drivers and when i try to enable compiz I get "No whitelisted driver found"
15:23
also when i boot my clients they will give me invalid username or password most of the time but if open shell tty and ping my server 10.0.0.1 they will log in fine... any ideas?
15:25Missionary has left #ltsp
15:29plamengr has quit IRC
15:30
<barbara_>
sbalneav, can you help me out with the server issue?
15:34
<mcfloppy_>
ok
15:36iMav has joined #ltsp
15:37
<barbara_>
have you heard of this before ogra ?
15:38
<sbalneav>
barbara_: What kind of switch do you have.
15:38
As for the compiz, no idea, I don't have any interest in it, and have never used it.
15:40
<barbara_>
sbalneav, i have an encore 10/100 switch
15:41
<sbalneav>
So, you say if you ping it, you can log in, but if you don't, they say, what? Workstation not authorized?
15:41
<barbara_>
it won't take the username and pass no matter how many times i type it in but when i go to tty2 and ping 10.0.0.1 it works. otherwise it says username or password incorrect
15:42
<sbalneav>
Well, I'd check /var/log/ldm.log on the workstation, that might yeild some info.
15:42
I have to go now, I'll be on later tonight.
15:42
cheers
15:42
<barbara_>
ok thx cya
15:43
<mcfloppy_>
hmm
15:43
[22:14] <ogra> objcopy -I binary -O elf32-i386 -B i386 vmlinuz-2.6.22-14-i386 vmlinux
15:43
dont work
15:43
^^
15:44
ogra are u here?
15:44Guaraldo has left #ltsp
15:45
<slidesinger>
Hey sbalneav
15:45
Do you know a top notch web programmer?
15:51Q-FUNK has joined #ltsp
15:59
<scrapbunny>
hi again i am trying to use the webmin to add users and I am getting told that the lines do not contain 13 fields. I have the text formated like this create:username:passwd:uid:gid:realname:homedir:shell:min:max:warn:inactive:expire where uid, min, max,warn, inactive and expire are spaces
16:01
<barbara_>
scrapbunny, use ssh it will save you a huge hassle
16:02edgarin_ is now known as edgarin
16:04
<warren>
sbalneav, has anyone tested nfsv3 vs nfsv4 by chance?
16:04
<scrapbunny>
i just switched from one script to webmin so i want to try to make webmin work before i switch again. I just need to know how to format the fields, especially what to put for the ones i am leaving blank
16:05
<ogra>
warren, i didnt get it to work properly back in edgy (6.10) when i tried last ...
16:05
<warren>
ogra, to get which?
16:06
<ogra>
might have changed though
16:06
nfsv4
16:06
<warren>
oh
16:06
I can't get nfsv3 to work here
16:06
with standard mount
16:06
<ogra>
from initramfs ?
16:06
<warren>
I'm using nash to mount nfsv3 without statd (locks)
16:06
<ogra>
i think we use either busybox or klibc mount
16:06
<warren>
ah
16:07
that explains tihngs
16:07
<mcfloppy_>
orga, the hexfile from the vmlinux from your command locks not like a original vmlinux
16:07
<ogra>
but i fiddled with std mount back then as well
16:07
didnt make much differenence in ubuntu
16:07
mcfloppy_, hmm
16:08
i used that command once to convert vmlinuz to an elf file ...
16:08
<slidesinger>
scrapbunny: I use adduser and ssh. Never could get webmin to work the way I wanted it to.
16:09
<ogra>
warren, did you see my last comments after you ran off to your meeting ?
16:09
<warren>
ogra, regarding tmp dir?
16:10
ogra, /tmp/somewhere is fine
16:10
<ogra>
great
16:10
<warren>
ogra, hopefully wit mktemp
16:10
<ogra>
so we dont fiddle with the tree
16:10
<warren>
yeah
16:10
<ogra>
err, mktemp ?
16:11
<scrapbunny>
I got it to work, i had an extra :
16:11
<ogra>
we want a versioned dir, no ?
16:11
<warren>
you don't have mktemp?
16:11
ogra, oh
16:11
ogra, yeah
16:11
ogra, as long as it blows away the old one first
16:11
<ogra>
bzr export /tmp/$NAME-$VERSION
16:11
indeed rmdir in front of that with proper checks
16:11
export will properly strip the .bzr dir
16:12
<warren>
nod
16:12
<mcfloppy_>
hm
16:12
<ogra>
then we can do whatever we want in there
16:12
next export will still be a clean tree export from the original source ...
16:12
<scrapbunny>
thanks again for the help earlier. i think i have all my users added now
16:13
<warren>
ogra, where is ubuntu installing chroots again?
16:13
<ogra>
/opt/ltsp/$arch
16:13
<warren>
Hmm....
16:14
<ogra>
i wouldnt oppose /srv though
16:14
apart from the fact that lots of docs have to be changed
16:15
and that it generates extra work for us to make sure upgrading works
16:16
<warren>
Let's not change /opt/ltsp for now
16:16
<ogra>
well, lets get an army of documenters :) then we can just change things around ...
16:17
docs is usually the piece that makes me decide against a change of defaults ....
16:17
<warren>
#!/bin/bash
16:17
remote=${1%:*}
16:18
test.sh 172.31.100.254:/opt/ltsp5/i386
16:18
remote=172.31.100.254
16:18
how do I extract the part after the :?
16:18
<ogra>
NBD_ROOT_SERVER=$( echo "${nbdroot}" | sed 's/:.*//')
16:18
NBD_ROOT_PORT=$( echo "${nbdroot}" | sed 's/.*://')
16:18
the latter ...
16:19
<warren>
Isn't that counter-intuitive?
16:19
<ogra>
note that ubunt and debian use dash
16:19
not bash
16:19
<warren>
Yeah, this is for mkinitrd which is bash only
16:19
(I asked, they wont switch)
16:19
<ogra>
ah
16:19
<warren>
It is surprising I think
16:19
:.* means everything on the left
16:20
<ogra>
but make sure initscripts that we can use in all distros are POSIX :)
16:20
<warren>
.*: means everything on the right?
16:20
ogra, I've been told that POSIX compliance in initscripts is silly, our developers are told to comply with LSB
16:20
I've also been told that they tested dash a few months ago, it doesn't make our boot faster
16:21
I've been meaning to test that for myself here
16:21scrapbunny has left #ltsp
16:22lns_ has joined #ltsp
16:23
<ogra>
well, there are documented tests from debian and ubuntu
16:23
we dont decide to switch the /bin/sh link out of the blue in ubuntu :)
16:24
it also lowers the ram requirements massively
16:24
<warren>
ah, ${1#*:} is what I wanted.
16:24
<ogra>
note that the default user shell is still bash
16:25
its only system scripts that default to dash and only if they use #!/bin/sh
16:26
most of the ltsp-client scripts are #!/bin/sh though
16:26
ltsp-build-client is still bash iirc
16:27
(so the plugins are bash as well)
16:39Q-FUNK has quit IRC
16:39kl_eisbaer has joined #ltsp
16:42bobby_C has quit IRC
16:43kl_eisbaer has left #ltsp
16:44cliebow has joined #ltsp
16:49
<barbara_>
ogra, i had a question about ltsp and compiz. I am running gutsy and chrooted into ltsp dir installed nvidia drivers and when i try to enable compiz I get "No whitelisted driver found"
16:51slidesinger has quit IRC
16:52
<lns_>
(Ubuntu Gutsy LTSP) Anyone who wants to help me out while I'm onsite is welcome :) I'm trying to get a shell script to run correctly via /var/lib/tftpboot/ltsp/i386/lts.conf (RCFILE_01). This script calls 'amixer' to turn on headphone-jack-sense. When it runs during bootup, the headphone jack sense does NOT get turned on, but the logon wav file is sped up considerably (like playing a 25rpm record at 45rpm). When I run the script manually from a console, it works
16:52
fine.
16:53
It worked fine in Feisty, but now it "broke" - I'm guessing it has something to do with PulseAudio (gutsy) vs Alsa (feisty) but I really don't know where to go to try and troubleshoot it.
17:01
<frownix>
lns_ it runs after all neccesary modules are loaded?
17:02
<lns_>
frownix, seems so... it's not really an issue of whether it loads or not (now), but when it loads during bootup, it makes sound much faster in speed (the record effect)
17:02
<frownix>
sounds weird
17:02
<lns_>
but after the thinclient is totally loaded, CTL+ALT+F1 to a console and manually running the script makes the headphone jack sense turn on.
17:02
you're telling me ;)
17:03
<frownix>
hehe
17:03
<lns_>
i'm purging some alsa/sound packages and reinstalling
17:03
see if that helps
17:03
<frownix>
since it works from cmd, it "could " be that it's run before something else needed is loaded
17:04
tried to boot into cmd only, run the script manually(disabled in rc.sysconf) to see if that works?
17:04
then start into X
17:07
<lns_>
it seems that that would be a logical conclusion
17:08
<frownix>
i'd try the "one step at a time" line, boot to cmd only, run the script, then startx, just to try to pinpoint the problem at bit more
17:13
<lns_>
frownix, how do you boot straight to a shell w/a client?
17:13
i know that sounds like a n00b question..i forgot =)
17:14
<frownix>
hmm, i don't use ltsp now, but you should be able to set it in lts.conf
17:14
let me see if I can find an old file here
17:14
<lns_>
ah n/m =) yeah, it's SCREEN01 = shell
17:14
its ok
17:15
thx anyway though =)
17:15
<frownix>
heh, yeah, i found it too
17:15
<lns_>
=)
17:15
brb
17:16
yack, nm, that doesn't work
17:18
<frownix>
what doesn't work?
17:18
<lns_>
it still booted to X
17:18
crazy
17:18
i'm using gutsy...
17:19
<frownix>
i've never used ubuntu, so I can't help there.., still weird that SCREEN01=shell didn't work
17:19
hmm, should be SCREEN_01
17:19
<lns_>
AH
17:19
yes
17:19* lns_ puts on hold music
17:20
<frownix>
lol
17:20
<lns_>
dude i am SO determined to fix this issue today
17:20
it's holding me back from going to 4 other schools and converting them to LTSP
17:21
<frownix>
i'm sure it's not a serious problem
17:21
it's just a matter or working downwards until it's found;-)
17:26
<lns_>
frownix, agreed =) just very frustrating..but educational!
17:26
so i got it to go to a shell, but how do i manually invoke ldm after i run the script?
17:26
<frownix>
in my very old rc.sysinit sound is started AFTER rcfiles
17:27
no idea, ldm didn't exist when I used ltsp;-), try just to type ldm, or /tmp/ldm
17:27
<lns_>
oh i think i got it
17:27* lns_ holds
17:27
<lns_>
yup, that did the trick
17:28
<frownix>
ok, did it work? normal sound?
17:28
<lns_>
it works great if you boot to SCREEN_01 = shell, invoke the fix script, then run /usr/lib/ltsp/screen.d/ldm and login
17:28
yep
17:28
over headphones
17:28
which is what we wanted
17:29
so lts.conf is loading RCFILE_01 too early (probably before sound libs are loaded?)
17:29
<frownix>
right, so the problem is probably that your script is run before all sound modules are loaded
17:29
right;-)
17:29
<lns_>
=)
17:29
hmm...
17:29
<frownix>
your script can't be more then a few lines?
17:29
<lns_>
1 line
17:29
=)
17:29
<frownix>
ok, just add it to rc.sysinit(not as a script, but put that line in there somewhere AFTER sound is started
17:29
<lns_>
maybe i can throw a sleep in there somehwere, but i dunno..seems kinda band-aidish to do it that way
17:29
<frownix>
if rc.sysinit is still used that is
17:30
sleep won't help
17:30
<lns_>
i don't think it is, but not sure...
17:30
it was highly recommended to me to use RCFILE
17:30
<frownix>
it won't release the script to continue the rc.whatever file
17:30
<lns_>
oh gotcha
17:30* lns_ INAP
17:30
<frownix>
so, sleep will only make you wait longer for fast sound;-)
17:30
<lns_>
lol
17:31
<frownix>
there has to be a sysinit file somewhere, and it shold be in ltsp_root/etc
17:31
go to /whatever/ltsp/etc and grep -R RCFILE *
17:32
shold find some file that are referencing the RCFILE's
17:32
<lns_>
i'm trying it in rc.local, which should run at the end of each runlevel
17:32
<frownix>
it "should"
17:33
let me check the old file again
17:33
<lns_>
btw I haven't seen you here before frownix, you seem versed in LTSP =)
17:33
thanks for the help so far
17:33
<frownix>
i was here a lot many years ago
17:34
been busy with other things lately though
17:34
<lns_>
righ on
17:34
s/righ/right*
17:34
<frownix>
:)
17:34
in my "old" file even rc.local is before "sound"
17:34
sound points to rc.sound though
17:35
so, your line at the end of rc.sound should work equally well
17:36* frownix is still confused over the fact that it used to work with the older dist
17:36
<lns_>
no dice
17:36
i know...it's weird
17:37
i know gutsy overhauled sound
17:37
<frownix>
oh, ok
17:37
<lns_>
so i don't even know if amixer is the right way to go to fix this issue since it uses pulseaudio now
17:37
but...running the script works, so...? heh
17:37
<frownix>
well, it works manually.....so
17:37
it's should just be a matter of finding a place to run it
17:37
are you in /etc of the ltsp root?
17:38
<lns_>
yes
17:38
<frownix>
there should be some rc files there?
17:38
<lns_>
yeah, first one i saw was rc.local
17:38
<frownix>
or rd.d/ or something like that
17:38
err rc.d
17:38
no rc.sysinit?
17:38
<lns_>
rc0-6.d, rc.local and rcS.d
17:38
<frownix>
ah, ok
17:39
what is the default runlevel?
17:39
3? 5?
17:39* sutula interjects a thought here: the sys init scripts aren't run in the logged-in environment, and the env has some "interesting" environment variables set that might be making the audio work there
17:40
<sutula>
lns: You might consider doing a "set" and capturing the environment in both places and see what's different
17:40
<lns_>
wow, no inittab in the chroot
17:40
not sure which runlevel :(
17:40
sutula, hmm
17:40
not totally following you there
17:40
<frownix>
i'd try with level 5
17:40
put your script there
17:41
<lns_>
i could always update-rc.d defaults too, right?
17:41
after chroot
17:41
<sutula>
lns: When you log into a client, there are some environment variables that get set to tell the audio system where your client's audio device is
17:41
<frownix>
heck, put it in 3 as well, it won't hurt
17:41
<lns_>
sutula, ok
17:42prpplague has quit IRC
17:42
<sutula>
lns: My point is that something that works as a user in a script is pretty different from what happens during client boot
17:43* sutula isn't sure how else to help out, so butts out
17:43
<lns_>
sutula, ok that's understandable - but how about when I go to vt1 and log in as root?
17:44
and then run the script - because that's the pure ltsp env, right?
17:44
<frownix>
it "should" be
17:44
so "should" shell be, but a lot might have happend during absence
17:45
<lns_>
well i chrooted and then update-rc.d defaults, so we'll see in a sec
17:45
<frownix>
still, try to add your script to some runlevels, it seems like an easy thing to test
17:45
if that don't work it's env variable digging time;-)
17:45
<sutula>
lns_: I understand...it should be very close
17:46
<lns_>
still no die
17:46
frownix, well when i updated the default runlevels, it put it in them all
17:46
<frownix>
hehe
17:47
<lns_>
i *am* rebuilding the chroot image every time, just to let everyone know =)
17:47
<frownix>
the shotgun way, i love that;-)
17:47
<lns_>
haha
17:47
gotta narrow it down somehow =)
17:47
<frownix>
hehehe, right
17:47
what order are the scripts run at in ubuntu?
17:48
by 01,02 and so on?
17:48
in each runlevel that is
17:48
we have to make sure it's loaded AFTER the sound stuff
17:48
no telling what happens if it's loading before, then again after
17:49
might have the same speed effect as before
17:49
<lns_>
you know what
17:49
it looks like most of the ltsp client stuff is in S20ltsp-client-core
17:49
<sutula>
frownix: high numbers for later
17:49
<lns_>
and in the runlevels i don't even see any Sxx for rc.local
17:50
so it might not even be invoked
17:50
<frownix>
so your script should be a higher number
17:50
ah, ok
17:50
<lns_>
or, maybe i should look at the ltsp-client-core stuff
17:50
<frownix>
call yours S99my_sound then
17:50
<lns_>
for a cleaner solution
17:50
<sutula>
lns_: to be sure it's running at all, echo some test string to a file and see if it shows up
17:50
<frownix>
a one line script can't be done in an ugly way no matter how you do it;-)
17:51
sutula, right
17:51
echo is nice for troubleshooting stuff like that
17:51
<lns_>
ok, lemme try that
17:53
<frownix>
the shotgun didn't work?
17:53
<lns_>
no
17:53
<frownix>
was it named S99something?
17:54
<lns_>
no, it was S20 (which was S20ipaqsoundfix.sh , so technically it should have gone before S20ltsp-client-core)
17:55
but i'll rename and try again
17:56
<frownix>
do add some echo > file in it to make sure it runs like sutula said
17:56
<lns_>
ah thanks, almost forgot
17:57
<frownix>
echo "i'm alive" > /var/lifesigns ;-)
17:57* lns_ puts on hold music again
18:01
<lns_>
ko
18:01
ok*
18:02
this is more interesting than i thought
18:02
<frownix>
ok?
18:02
<lns_>
so it worked, kind of - one thin client executed the script and is using headphones at normal sound speed
18:02
another (differing hardware but same original sound issue) also uses the script, but has sped-up logon wav.
18:02
both did NOT echo anything on the console....
18:03
<frownix>
you just used echo "hello" without > to file?
18:03
<lns_>
yeah..oh is that bad? =p
18:04
<frownix>
might not show up on the console(depending on layout)
18:04BGomes has joined #ltsp
18:04
<lns_>
ok i'm trying again with the output to a file
18:04* frownix is reminded of why he reversed the thin-client when problems like this happens ;-)
18:05
<lns_>
hahaha...yes, fourtunately they have hdds with win2k on them that they're using currently
18:05
i couldn't ask for a more failsafe installation
18:06
<frownix>
haha, win2k as failsafe..now that's funny;-)
18:06
<lns_>
=p
18:06
well, fortunately they have the audio drivers ;)
18:06
ok
18:06
the file did get echo'd to
18:06
so it is running
18:06
<frownix>
right
18:07
<lns_>
but i'm still stumped as to why sound is being sped up on the other clients
18:07
that's just funky
18:07
<frownix>
but only works for one client...it's ...well...hmmm...
18:07
funky was the word..yes;-)
18:07
<lns_>
lol
18:07franzn has joined #ltsp
18:08
<lns_>
it HAS to be something with pulseaudio....i hope it is anyway...i dunno
18:08
<frownix>
i have zero experience with remote sound, but it sounds like a pulseaudio thing yes
18:08franzn has left #ltsp
18:08
<lns_>
i'd love to know how to rip every trace of alsa/esd out of the install and start over installing pa
18:09tier1_ has quit IRC
18:09
<frownix>
the ubuntu guys seems to be sleeping now..they are surely more up to date on this then me, since they wrote it..hehe
18:10
so, pulseaudio is running serverside?
18:10* frownix has no idea how pulseaudio works
18:11
<frownix>
or is it loading the local soundcard?
18:11tier1_ has joined #ltsp
18:11johnny has joined #ltsp
18:11
<lns_>
frownix, i'm not totally sure
18:11
i'm going to try to completely rebuild the chroot
18:12
<frownix>
ouch
18:12
<lns_>
yeah...well =)
18:12
<frownix>
i don't think it'll help much
18:12
<lns_>
hmm
18:12
<frownix>
it didn't work when it was fresh, right?
18:12
<lns_>
nope
18:12
<frownix>
so, it probably won't work on the next fresh build either
18:12
<lns_>
i've just done so many things, not sure where to go next
18:13
<frownix>
does it work without your script? meaning, do they get sound through speakers?
18:14
<lns_>
frownix, w/o the script the headphone-jack-sense is disabled, so they always get audio through speakers, whether headphones are plugged in or not
18:14
<frownix>
but it does work that way?
18:14
<lns_>
noooo
18:14
=) 35 students with speakers blasting... ;)
18:14
<frownix>
i mean, they do have sound that way?
18:14
<lns_>
oh yeah
18:14
<frownix>
lol
18:15
<lns_>
they have sound, just ...all of them....at the same time... =)
18:15
<frownix>
hahah
18:15
i can see the problem with that, i was just trying to to find out if it worked that way;-)
18:15
<lns_>
hehe
18:15
<frownix>
so, we can then say that your script is the problem
18:16
the script is acting on the clients soundcard, i guess?
18:16
or should be at least
18:16Egyptian[Home1 has quit IRC
18:16
<frownix>
it can't be that you are in some way doing the changes on the server instead?
18:17
<lns_>
oh man...
18:17
what the hell
18:17
lol
18:17
<frownix>
it tends to happen ofter on thin-clients that those to are mixed up
18:17
<lns_>
for some reason, pulseaudio isn't even installed
18:17
dude...shouldn't the upgrade like....take care of that?? =)
18:17
<frownix>
hehe, ok, so we can rule out that;-)
18:17
<johnny>
i never got headphone sensing working, i just use the switcher in my mixer
18:18
<frownix>
lns_ so you are basicly using your old setup for sound then, without pulseaudio?
18:19
<lns_>
frownix, right now yes..i'm installing pa right now though
18:19
i don't understand why it isn't there after the distribution upgrade
18:20
<frownix>
beats me, i have not one single clue on how ubuntu does things
18:20
<lns_>
=)
18:20
ok let's try this again now
18:21
<frownix>
lns_ just a thought though..clients that run w2k should be able to run your apps local y...and i belive that ubuntu-ltsp has localapps working more or less on stock setup
18:22
wouldn't it be much easier to go down that road? it would solve the sound problem, and take some load of the server
18:22
(i'm totally aware of that it's not a solution to your problem, but still a thing to consider)
18:22
<lns_>
frownix, the issue is not local apps, but how the headphone-jack-sense isn't being ..sensed
18:23
<johnny>
does that applicaiton really work on all sound cards?
18:23
<frownix>
well, a local app should use local sound, and then the jack-sense should work, or?
18:23
<johnny>
if so.. why isnt' in default
18:23
<lns_>
it should always use "local" sound...there's no sound card in the server at all
18:23
it's how sound is routed
18:23
<frownix>
ah, ok
18:24
<johnny>
yeah.. i do need to setup local apps with firefox.. i don't know how tho :)
18:25* frownix is soooooo out of touch on ltsp that it's embarrasing
18:25
<frownix>
johnny, i'm sure ubuntu has documentation on it
18:25
ltsp.org should have some as well
18:26
<johnny>
not that i've seen
18:26
nothing looks up to date anywhere
18:26
<frownix>
look for local apps in ubuntu docs, it might not be firefox specific
18:27
<johnny>
uggh.. i wish i could find a way to get paid to organize efforts around improving and documenting free software
18:27
<frownix>
a wild guess would be: install firefox in the chroot, ssh to the workstation to start it
18:27
sort of
18:27
<lns_>
ok, i'm gonna bounce the server...might be back, might not...either way, thank you all for your help =)
18:27
<frownix>
firefox -display my_workstation:0.0
18:28
<johnny>
you should read the spec page for it on ubuntu somewhere
18:28lns_ has quit IRC
18:28
<frownix>
lns_ ouch, well, sorry I couldn't help you fix it
18:28
<johnny>
it describes sorta how it works
18:28
<frownix>
johnny, it "shold" be as simple as I said
18:28edgarin has quit IRC
18:29
<johnny>
it should be simpler :)
18:29
<frownix>
on your workstation, ssh to the workstation(since you are actually on the server)
18:29
<johnny>
like.. look for an app locally and then look on the server
18:29
<frownix>
start firefox with the display option
18:29
<johnny>
obviously allow for an override
18:29
it's sad that firefox is so horrible
18:29
<frownix>
what dist are you running?
18:29
<johnny>
ubuntu
18:30
<frownix>
ok, ssh enabled on the clients?
18:30
<johnny>
i hope firefox fixes lots of issues.. otherwise it's epiphany + webkit for all my machiens
18:30
the gconf issue is something to be concerned about
18:31
<frownix>
well, i just find it slower then opera, so I use opera(since i want as efficient programs as possible on the clients)
18:31
<johnny>
opera isn't open source
18:31
not an option for me
18:32
<frownix>
i can appreciate that view. for me efficency is more important
18:32
<johnny>
i'd rather spend more money on hardware
18:32
<frownix>
if open source is better I use that
18:32
<johnny>
than compromise on that
18:33
<frownix>
either way, to run firefox locally should be simple. I might even have a script for it
18:33* frownix checks in the drawers
18:34
<johnny>
you should read the spec page
18:34
<frownix>
i don't have to, since I allready run firefox locally;-)
18:34
<johnny>
sure, but to do it the recommended way
18:34
and a way that will grow in the futre
18:35
to support gconf proxying perhaps
18:35
and such things
18:35moquist_ has joined #ltsp
18:35
<frownix>
my way grows just fine. but then again, my setup is very different from ltsp's
18:35
running an app like firefox locally still is very easy, if the chroot has firefox installed that is
18:36
but, if there's a recomended way that works, sure, that's fine as well
18:36
doesn't sound like it's that easy though(hearing you talk about it)
18:37
you have a link?
18:42
well, its late here..goodnight
18:42frownix has quit IRC
18:51moquist has quit IRC
18:54tier1__ has joined #ltsp
18:55staffencasa has quit IRC
19:10tier1_ has quit IRC
19:10tier1_ has joined #ltsp
19:11mistik1 has quit IRC
19:13mistik1 has joined #ltsp
19:15tier1__ has quit IRC
19:17barbara_ has quit IRC
19:44happywithed has joined #ltsp
19:45
<happywithed>
Hello there
19:45
Has anyone tried to run SecondLife on an LTSP server?
19:46
My clients have good graphics cards, but no hardisk
19:46tier1__ has joined #ltsp
19:46
<happywithed>
SecondLife has a Alpha client
19:47
Does anyone have any ideas how to do this?
19:47
Ogra?
19:53mistik1 has quit IRC
19:55mistik1 has joined #ltsp
19:56
<happywithed>
I guess noone has tried this.
19:56
I wonder if it is possible
20:02tier1_ has quit IRC
20:04kwak has joined #ltsp
20:04kwak has left #ltsp
20:16BGomes has left #ltsp
20:20J45p3r has joined #ltsp
20:28J45p3r has quit IRC
21:11happywithed has quit IRC
21:17ogra has quit IRC
21:17ogra has joined #ltsp
22:18Joris_ has joined #ltsp
22:28iMav has quit IRC
22:28iMav has joined #ltsp
22:31edgarin has joined #ltsp
22:32edgarin_ has joined #ltsp
22:33iMav has quit IRC
22:34Joris has quit IRC
22:37subir has joined #ltsp
22:38edgarin_ has quit IRC
22:55Artemka has quit IRC
23:16daya has joined #ltsp
23:19daya has quit IRC