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


Channel log from 21 April 2007   (all times are UTC)

00:19dtrask has joined #ltsp
00:25dtrask has quit IRC
00:26dtrask has joined #ltsp
00:26dtrask has quit IRC
00:45freet15_ has quit IRC
01:02freet15 has joined #ltsp
01:16freet15 has joined #ltsp
02:21Egyptian[Home] has joined #ltsp
02:48freet15 has quit IRC
03:04freet15 has joined #ltsp
03:17sepski has joined #ltsp
03:39joebob777as7 has quit IRC
03:40joebob777as7 has joined #ltsp
03:46sepski has quit IRC
04:51radoeka has joined #ltsp
05:27radoeka has quit IRC
06:33artista_frustrad has joined #ltsp
07:13J45p3r has joined #ltsp
07:32cyberorg has joined #ltsp
07:33
<cyberorg>
hi jammcq need small addition on ltsp wiki (i dont have access on http://wiki.ltsp.org/twiki/bin/view/Ltsp/DownLoads)
07:38
<ogra>
cyberorg, what do you want added, probably i can find my login data
07:40
<cyberorg>
ogra: i am not even sure where it needs to go, on suse to get ltsp5 working you need to ln -s /etc/X11/xdm/Xsession /etc/X11/ otherwise ldm just restarts X
07:40
may be in troubleshooting
07:41
<ogra>
ah, well at least it work somehow on SuSE ... for redhat even the linking didnt work ...
07:41
<cyberorg>
ok, i have access there
07:41
ogra: debian tarball works perfectly :)
07:42
no changes required anywhere
07:42
<ogra>
we should have a per distro list at the bottom for such extras, pulseaudio needs some server sided config as well
07:42
well, then feel free to edit :)
07:42
debian tarball is some versions behind
07:42
it still uses esd
07:43
<cyberorg>
ok, later i'll try ubuntu one, then i'll start working on native suse
07:43
<ogra>
yay
07:44
do you have a bzr tree already ?
07:44
we're soon start a new codemerge, debian and ubuntu both had releases recently, so we'l merge the upstream code
07:44
would be nice to pull in some native SuSE stuff ;)
07:44
<cyberorg>
ogra: not yet, i would be starting next week, just got tarballs
07:45
<ogra>
poke sbalneav, vagrantc or me if you are ready :)
07:45
<cyberorg>
i dont really understand many of the things, but debian tarball worked out of box
07:45
so expect for using native suse chrooted, what else could be different?
07:46
<ogra>
well, what you need first is the set of plugins for ltsp-build-client to build a client environment from distro tools
07:46
<cyberorg>
may be branding ldm :)
07:46
ogra: i am thinking of using kiwi to build that image
07:46
<ogra>
with the distro packages
07:46
right
07:46
if you can integrate that thats fine
07:47
<cyberorg>
are any core packages in ltsp chroot modified?
07:47
<ogra>
if you once have that image, make sure to have two initscripts, one that configures everything on the client during boot (steal liveCD code from your distro for that ;) )
07:47
and one that starts the ltsp cleint by using the screen scripts
07:48
<cyberorg>
ogra: not now :), next week
07:48
i'll need all the help i can get
07:48
<ogra>
in the ubuntu/debian tarball thats done by /etc/init.d/ltsp-client-setup and /etc/init.d/ltsp-client
07:48
have a look at them and think how you can solve these things for SuSE :)
07:49
<cyberorg>
ok, thanks :)
07:57
ogra: you worked on debian or ubuntu one?
07:58
<ogra>
i'm doing the ubuntu implementation
07:58
<cyberorg>
ok
07:58
<ogra>
(i.e. most of the core development and new features)
07:59
<cyberorg>
hopefully i would be able to use most of your work :)
07:59
<ogra>
if it needs changes, just tell me :)
07:59
or one of the others ;)
08:00
<cyberorg>
yeah, sure
08:00
right now i am looking for a video that shows xen live migration, cant find it :(
08:04bobby_C has joined #ltsp
08:41vagrantc has joined #ltsp
08:44J45p3r has quit IRC
09:19vdittgen has joined #ltsp
09:24vagrantc has quit IRC
09:26vdittgen has quit IRC
09:28vdittgen has joined #ltsp
09:29vdittgen has joined #ltsp
09:40xlyz has joined #ltsp
09:41muh2000_ has joined #ltsp
09:44
<xlyz>
hi
09:57muh2000__ has quit IRC
10:06mhterres has joined #ltsp
10:11
<cliebow>
xlyz hi
10:20
<xlyz>
where shall I install applications? on the server or in the ltsp chroot?
10:20* xlyz being in debian etch
10:21Novell has quit IRC
10:31mhterres has joined #ltsp
10:31
<mhterres>
hey guys
10:31
jammcq, morning
10:31
Guys I need help with some sentence in english
10:32
<cyberorg>
xlyz: if you want to use local app, then in chroot, otherwise install on your server
10:32
<mhterres>
I'm translating some local song to sbalneaves and I'm having problems with that
10:32
anybody can help me ?
10:33privet has joined #ltsp
10:34
<xlyz>
no local apps. maybe ekiga in the future (any experience?). thanks btw
10:34
<mhterres>
If I want to say that It doesn't matter to me if I don't have butter
10:35
can I say: It can lack butter to me ????
10:36
anybody ?? I know this is an off-topic, but I want to keep my promise :-)
10:36
<ogra>
how abut "life can lack butter to me" :)
10:36
<mhterres>
let me show you the first sentence
10:37
It can lack everything to me in life
10:37
rice, beans and bread
10:37
how you understand of that ?
10:37
It sounds me strange :-)
10:37
<cyberorg>
ltsp == language translation service provider?
10:37
L)
10:37
<mhterres>
uhahaha
10:37
<cyberorg>
:)
10:37
<ogra>
*grin*
10:37
<mhterres>
cyberorg, just to send a song to sbalneaves
10:38
but you know, sometimes is complicated local songs to other langagues
10:38
heheeh
10:38
complicated translated
10:38
<ogra>
yes, sounds a bit strange
10:38
<mhterres>
hehhee, I know
10:54
thanks guys
10:54
I'll have to go
10:54
see you monday
10:54mhterres has quit IRC
10:59xlyz has quit IRC
10:59
<cyberorg>
ogra: any way we can get composite desktop (compiz/beryl) on clients?
10:59
if clients has supported hardware
11:00
<ogra>
nope
11:00
it would only work if you would run it as a local app
11:02cyberorg has quit IRC
11:11
<sutula>
ogra: I don't know if you saw my query yesterday, but I'm wondering if there's an easy way to use the debian/ubuntu packaging, but not go through ssh
11:12
<ogra>
sutula, set SCREEN_07=startx in the lts.conf file
11:12
<sutula>
ogra: Where is the lts.conf file?
11:12
<ogra>
that will force XDMCP with all its disadvantages
11:12
in /etc in the chroot
11:13
<sutula>
ogra: By disadvantages, are you primarily concerned with security?
11:14
<ogra>
sutula, well, anybody can read every keystroke with a simple tcpdump
11:14
you can sniff wonderful screenshots from XDMCP traffic without and hassle ...
11:14
*any
11:15
<sutula>
ogra: Right...I was just thinking that I'm within my house, on a private network...the ssh overhead kills performance on old machines
11:15
<ogra>
apart from that the implementation of sounds and local devcide work through ldm ...
11:16
<sutula>
ogra: I'm not seeing an lts.conf in /opt/ltsp/i386/etc ...am I looking in the wrong place?
11:17
<ogra>
depends, do you use the debian tarball ? thats a bit behind, at that time we didnt ship a default lts.conf
11:17
if so, create one with a default section and add the SCREEN_07 variable there
11:18
<sutula>
ogra: Yes, Debian Etch currently...I think I found the example under ...usr/share/doc...
11:19
<ogra>
yup, there is also a complete list of all parameters
11:19
<sutula>
ogra: Do you know if updates will overwrite that? It's not strictly a conf file since it doesn't live under /etc
11:20
<ogra>
there is no code that would overwrite it, no
11:20
<sutula>
ogra: k, thanks much!
11:20* ogra goes back to nfs4 playing ...
11:53freemind has joined #ltsp
11:59rasmuson has joined #ltsp
12:02
<freemind>
hi...
12:02
someone can help with vmware as client???
12:03
or qemu???
12:06
<rasmuson>
freemind: press f12 when vmware boots and it does a pxe boot
12:24privet has quit IRC
12:25privet has joined #ltsp
12:29
<freemind>
tks :-D
12:33
not work...
12:46privet has quit IRC
12:50privet has joined #ltsp
13:03
<stillflame>
ltsp5 question: i'm working on setting up a fat client workstation, and installed icewm and gdm on the chroot (and otherwise went through the workstation instructions). gdm fails on the terminal, first complaining of improper ownership and permissions on /var/lib/gdm, then after i fix that, it complains of an 'internal error' in not being able to start X. am i missing a step? should i have had to do any of these steps?
13:03vagrantc has joined #ltsp
13:03
<stillflame>
ltsp5 question: i'm working on setting up a fat client workstation, and installed icewm and gdm on the chroot (and otherwise went through the workstation instructions). gdm fails on the terminal, first complaining of improper ownership and permissions on /var/lib/gdm, then after i fix that, it complains of an 'internal error' in not being able to start X. am i missing a step? should i have had to do any of these steps?
13:04
(sorry for the double post, i knew vagrantc might have the answer)
13:04
<vagrantc>
hmmm...
13:04
those instructions worked for me back in december
13:04
<stillflame>
they worked for me last month.
13:05freemind has quit IRC
13:05
<stillflame>
okay, i'll keep investigating.
13:09
<vagrantc>
s,december,september,
13:11
stillflame: grep rw_dirs /opt/ltsp/i386/etc/default/ltsp-client-setup
13:14
stillflame: that complaining of ownership and permissions for /var/lib/gdm makes me suspicious.
13:16
stillflame: chroot /opt/ltsp/i386 ls -l /var/lib/gdm
13:17
stillflame: chroot /opt/ltsp/i386 ls -la /var/lib/gdm
13:26
<stillflame>
well, i've changed it since it started, but it is in $rw_dirs, it was originally gdm:gdm 750
13:26
now it's root:gdm 1770
13:27
<vagrantc>
drwxrwx--T 2 0 103 4096 2007-04-02 17:46 .
13:27
gid 103 = gdm on my system
13:27
stillflame: the server and chroot may have a different idea of what the gid for gdm is
13:27
<stillflame>
okay, that's not what was installed by default, but i altered it to be the same.
13:28
vagrantc: yeah, i caught that one early on when it was owned by ssh.
13:28
gdm is what it reports chrooted.
13:28
<vagrantc>
stillflame: and you've got the T part of the permissions ?
13:28
<stillflame>
1770, yup.
13:29
and that error has gone away since i fixed it, now i'm tracking down this 'internal error'.
13:30
<vagrantc>
stillflame: have you tried purging gdm and re-installing ?
13:30
"fixing" things by hand seems likely to result in unreproduceable problems.
13:32
stillflame: what is this setup for?
13:32
<stillflame>
redandblack
13:32
vagrantc: i've rebuilt the chroot and reproduced it.
13:33
if i add a user the the chroot, i should be able to login at the console with it, right?
13:33
^in
13:33
heh.
13:36
<vagrantc>
if that user has a password, yes.
13:38
<stillflame>
hmm. it is also not doing that, then. something is fishy, but i can't get in to grep through the logs.
13:38
<vagrantc>
how did you add a user?
13:39
or, chroot /opt/ltsp/i386 passwd
13:39
should set root's password and allow you to log in
13:40
<stillflame>
i just did: chroot /opt/ltsp/i386 adduser stillflame
13:40
<vagrantc>
that ought to work, yes ...
13:41
stillflame: you could generate an ssh key and install openssh-server
13:42
<ogra>
vagrantc, hey
13:43
i just buit an initramfs with the standard mount and portmap ... next step nfs4 :)
13:43
<vagrantc>
ogra: hi
13:43
<ogra>
*built
13:43
<vagrantc>
nice
13:43
how much bigger is it?
13:44
what's all the hype with nfsv v4 ?
13:44
<ogra>
400k smaller :) because i dropped all base modules additionally :)
13:44
<vagrantc>
heh
13:44
<ogra>
kerberos encryption :)
13:44
nfs4 is safer
13:45
for now i only wanted to have a standard nfs client in the initramfs nfs4 wa just an ide for a next step ... :)
13:45
*was just an idea
13:46
<vagrantc>
sure
13:46
<ogra>
i'm just randomly poking around in the initramfs for speedups ...
13:46
ad found a lot stuff we can drop
13:46sepski has joined #ltsp
13:47
<ogra>
the base set of modules in mkinitramfs is totalnonsense for netbooting
13:48
<vagrantc>
so, MODULES=netboot included too much?
13:48
<ogra>
even though i appreciate that people with usb keyboard can be able to type in busybox if something breaks, i dont really see a reason for having the whole usbstack ...
13:48
same goes for ext2/3 reiserfs, xfs
13:48
etc
13:49
<vagrantc>
i can see usb slightly more than filesystems.
13:49
<ogra>
have a look at /usr/share/initramfs-tools/hook-functions
13:49
the base set is directly above net
13:49
<stillflame>
A parameter was malformed
13:49
Validation error
13:49
Connection closed by 10.10.10.250
13:49
says ssh
13:50
<ogra>
for x in ehci-hcd ohci-hcd uhci-hcd usbhid usb-storage ext2 \
13:50
ext3 isofs jfs nfs reiserfs xfs af_packet atkbd i8042; do
13:50
^^^ is what ubuntu has
13:50
<vagrantc>
stillflame: how many times have you re-installed from scratch? :P
13:50
<ogra>
i assume debian has the same
13:50
<stillflame>
vagrantc: just once now
13:50* jammcq loves seeing ogra pairing down the initramfs
13:50
<vagrantc>
stillflame: inclding the server?
13:50
<ogra>
jammcq, i think nfs4 is possible .... should we spec it for gutsy ?
13:51
<stillflame>
vagrantc: no, i left the server as is. i could try rebuilding it now, i guess.
13:51
<jammcq>
ogra: are you prepared to support kerberos ?
13:51
<ogra>
hmm
13:51
not relly keen, no
13:51
<jammcq>
if you are, then lets go for it
13:51
<vagrantc>
stillflame: things just sound messed up.
13:51
<ogra>
*really
13:51
<stillflame>
vagrantc: yeah they do.
13:51
<vagrantc>
stillflame: but rebuilding without really knowing the problem ...
13:53* vagrantc wonders why on earth all modules are hardcoded rather than variables
13:54
<stillflame>
vagrantc: i can't get in to find out what the problem is. well, i didn't build the server in the first place, so i can't be sure that's not the heart of the problem.
13:54
<ogra>
you mean read from $(ls /etc/modules/blah...) ?
13:54* stillflame reinstalls etch.
13:54
<vagrantc>
ogra: no ... "for x in list of hardcoded modules ; do"
13:55
ogra: like the usb and filesystem modules you highlighted shortly ago
13:56* vagrantc really likes how MODULES=none behaves identical to MODULES=most
13:56
<ogra>
heh
13:56
well, you could populate the hardcoded list with my above command ...
13:56
the modules are pretty well sorted in subdirs :)
13:57
<vagrantc>
oh, i see what you're saying ...
13:57
yeah, that would seem better, wouldn't it.
13:57
/lib/modules/ more
13:58
<jammcq>
just make sure that any optimizing that you guys do favors optomizing runtime performance over anything else
13:59
<vagrantc>
well, we're discussing build-time stuff at the moment... if it ran a little slower but produced a smaller initramfs, we should be fine.
13:59
<jammcq>
sure
13:59
just saying that anything that is executed by the client should be as fast and small as possible
13:59
<vagrantc>
sure
14:00
we should replace some of the python code with shell scripts ...
14:00
<ogra>
smaller initramfs loads faster ...
14:00
<jammcq>
well...
14:00
<ogra>
less modules in the initramfs speedup udev
14:00
<jammcq>
load time turns out to be a very small part of the boot time
14:00
<ogra>
its all related ;)
14:01
<jammcq>
and depending on what you are doing in the shell, python could actually be quite a bit faster
14:01
<ogra>
yeah, agreed
14:01
<jammcq>
but I definately like the idea of fewer modules
14:01
<ogra>
we wont rewrite the world in shell though ...
14:01
<vagrantc>
i think there are some things written in python that would be just as fast if not faster in shell
14:01
<jammcq>
reiserfs isn't terribly useful to a thin client
14:01
<ogra>
well
14:02
<jammcq>
vagrantc: the problem with the shell is it requires launching external binaries for many things, or it creates sub-shells
14:02
<ogra>
i want to have the ability to use laptops ... so reiserfs is important ...
14:02
but surely not for booting :)
14:02
we have all that crap duplicated in the nfsroot anyway
14:02
<jammcq>
at one point, I replaced some complex shell scripts in the initrd with some perl scripts and I was totally surpized at the improved performance
14:02
<ogra>
there is much room for optimization in the initramfs
14:03* vagrantc has had too many reiserfs corruptions in the past two years
14:03* ogra never used reiserfs
14:03* jammcq never used it either
14:03
<jammcq>
we don't even need ext3
14:03
just ext2, eh?
14:03* ogra has seen too many reiserfs corruptions in the past two years
14:03
<ogra>
:)
14:04* jammcq has heard the horror stories
14:04
<ogra>
jammcq, we dont need *any* filesystem modules
14:04
<vagrantc>
nfs
14:04
<ogra>
its the initramfs
14:04
yeah
14:04
<vagrantc>
but really, that should probably go in the "netboot" section
14:04
<ogra>
and af_packet
14:04
<jammcq>
but isn't the initramfs still a certain type of filesystem?
14:04
like cramfs or something?
14:04
<ogra>
sure
14:04
<vagrantc>
it's "initramfs" filesystem
14:04
<jammcq>
ah
14:04
<ogra>
thats compiled in in the kernel
14:05
<jammcq>
ok then, makes that easy
14:05
<ogra>
right
14:05
<jammcq>
with initrd, we really had a filesystem to deal with
14:05
<vagrantc>
yeah, how ugly that was.
14:05
<ogra>
i'm just starting to go through all scripts in the ubuntu initramfs ... and there is a lot we can drop
14:05* jammcq thinks that's a great idea
14:05
<ogra>
i dont think we need thermal before / is mounted, do we ?
14:06
<jammcq>
agreed
14:06
<ogra>
CPUs will run fullspeed anyway during boot
14:06
<jammcq>
we WANT them to run full speed
14:06
<ogra>
and udev will load it later
14:06
<jammcq>
but... how does udev know that it should load it?
14:06
doesn't the kernel or hal generate an event and udev loads the module?
14:06
<ogra>
the kernel sends events for all HW it finds
14:07
<jammcq>
but.... if the kernel sends the even, and udev doesn't find the module, what will happen?
14:07
s/even/event/
14:07
<ogra>
aside udev has udevtrigger and udevsettle to generate events for ventless devices
14:07
if there is no module udev wont do anything :)
14:07
it will return :)
14:08
<jammcq>
ok, but then later (after root is mounted via NFS), how does the module get loaded ?
14:08
<ogra>
udev runs
14:08
"loading hardware drivers" :)
14:08
<jammcq>
again ? or for the first time?
14:09
<ogra>
udev runs once in initramfs to detect the netcard module ... (or for non netbooted systems the HD and filesystem)
14:09
it creates the initial set of devices needed by the kernel and ounts /
14:09
*mounts
14:09
then init kicks in
14:10
udev has an initscript ...
14:10
<jammcq>
ah, so udev starts again?
14:10
<ogra>
so udev runs again and now really loads all modules for all HW it finds
14:11
<jammcq>
ah, sounds perfect
14:11
<ogra>
i dont want to chage the second run
14:11
but i want to drop all stuff we dont need from the first
14:11
<jammcq>
sure
14:11
<ogra>
the second run guarantees that we can run on all HW ...
14:11
the first only boots us
14:12
hmm
14:12
there is apparently a way to use nfs4 without kerberos ...
14:12
i wonder if thats speedier
14:12
(speedier than v3 i mean)
14:13
<jammcq>
but... without kerberos, do you still get the encryption ?
14:13
<ogra>
indeed not
14:13
<jammcq>
I think nfs3 can be fast, I just think there's something broken in nfs3 in Ubuntu
14:13
<ogra>
but v4 can handle bigger package sizes
14:13
<jammcq>
bigger? bigger than 256k ?
14:13
<ogra>
err
14:14
<jammcq>
how big you want?
14:14
<ogra>
i eant rsize wsize values ...
14:14
<jammcq>
yeah, how big? they are 256k now for gosh sakes
14:14
we do fine in ltsp-4 with 32k or 8k
14:15
<ogra>
the default for v4 is 32768 i think
14:15
<jammcq>
I think that's the defalt for nfsv3 too
14:15
since the 2.6 kernel
14:15
used to be 8k with the 2.4 kernel
14:15
<ogra>
v3 is limited to 8k
14:15
v4 ges up to 32
14:15
<jammcq>
I don't think so
14:16
<ogra>
man mount ;)
14:16
<jammcq>
you beleive the docs ?
14:16
<ogra>
oh207, no v2 is limited
14:16
sorry, misread
14:16
<jammcq>
:)
14:16
that's the whole reason we needed to add MOPTS to our initramfs
14:17
to allow people to reduce it back down to 8k or 2k when using 10mbit hardware
14:17
<ogra>
yup
14:17
<jammcq>
gigabit server nick and a 10mbit client nick didn't play very well, when the blocksize was 32k
14:17* ogra just notices he should check the root-path IP bug while fiddling in initramfs
14:18
<jammcq>
what's that bug?
14:18
<ogra>
if you set an ip in the rootpath it doesnt boot ?
14:18
<jammcq>
ah. I thought that was a design change. I'd be very happy if it was fixed to be like previous LTSP versions
14:19
<ltsppbot>
"ogra" pasted "nfsroot code from ubuntu initramfs" (14 lines) at http://pastebot.ltsp.org/107
14:19
<ogra>
have a look
14:19
i dont see an obvoius bug here
14:21
<jammcq>
I'd have to study that code a bit. dunno what ${NFSROOT%%,*} does
14:21
and ${NFSROOT#*,}
14:21
<ogra>
hah, got it
14:21
NFSROOT != ROOTPATH :)
14:22* ogra hugs virtualbox
14:23
<ogra>
/tmp/net-eth0.conf only has gotten ROOTPATH from dhcp
14:23
NFSROOT doesnt exists anywhere in the busybox env ...
14:24
<jammcq>
so does ROOTPATH look like 192.168.0.1:/opt/ltsp/i386 ?
14:24
<ogra>
so i assume "serverip:/path/to/nfs" becomes "serverip:"
14:24
no ROOTPATH is only the path
14:24
<jammcq>
who sets ROOTPATH ?
14:24
<ogra>
the kernel
14:25
<jammcq>
oh?
14:25
that seems odd
14:25
<ogra>
well whatever recieves the dhcp data ...
14:25
i guess the tcp stack
14:25* jammcq is thinking that ipconfig is getting that
14:26
<jammcq>
userspace, not kernelspace
14:26
<ogra>
anyway, NFSROOT=${ROOTSERVER}:${NFSROOT} is clearly wrong
14:26
<jammcq>
yeah, prolly
14:26
simple matter of coding at this point
14:27
<ogra>
nust be s/NFSROOT/ROOTPATH/s and we should be set
14:27
<jammcq>
who is setting ROOTSERVER ?
14:27
same thing?
14:27
<ogra>
yep
14:28
they both get written to /tmp/net-eth0.conf by ipconfig
14:28
<jammcq>
so, if I set in dhcpd.conf: option root-path "192.168.0.1:/opt/ltsp/i386"
14:28
who is splitting the 2 pieces into ROOTSERVER and ROOTPATH ?
14:28
in ltsp-4.2, we do it inside the script in initramfs
14:30
<ogra>
oh207, wow
14:30
option root-path "192.168.0.1:/opt/ltsp/i386" results in 192.168.0.1:192.168.0.1:/opt/ltsp/i386
14:30
<jammcq>
:)
14:30
so, nothing is splitting it out
14:31
<ogra>
i have no clue where NFSROOT comes from though
14:31
<jammcq>
I think all you need to do is detect whether there's an IP address in ROOTPATH, and if so, leave it alone. if not, then prepend the ROOTSERVER to the ROOTPATH
14:31
<ogra>
well, its split in ROOTPATH and ROOTSERVER
14:31
and i cant find NFSROOT in the initramfs :)
14:31
but it gets set
14:32
<jammcq>
if you set 'option root-path "192.168.0.1:/opt/ltsp/i386"', i'm guessing that ROOTPATH will already contain the IP address
14:32
<ogra>
yep
14:32
indeed what yo said will fix it
14:32
but i still would like to know where it comes from :)
14:33
<ltsppbot>
"jammcq" pasted "how LTSP-4.2 sets up NFSROOT" (22 lines) at http://pastebot.ltsp.org/108
14:33
<jammcq>
take a look at our old code
14:34
i'm guessing there's a better way of splitting out the pieces that my line 1 and 2
14:35
prolly some spiffy shell syntax to do it
14:35
<ogra>
well, i dont need to split anything :)
14:35
<jammcq>
well, you just need to detect the ':'
14:35
if it's there, then yer all set
14:38
<ogra>
hmm, actually the code isnt so wrong as it is ...
14:38
<jammcq>
just a little wrong ?
14:39
<ogra>
well
14:39
if [ "x${NFSROOT}" = "xauto" ]; then
14:39
NFSROOT=${ROOTSERVER}:${ROOTPATH}
14:39
we have NFSROOT set to auto
14:39
so it should do it right ...
14:46
ah, not if ROOTPATH is not only the path
14:47* ogra grumbles at the person who choose that variable name ... likely jbailey
14:49
<jammcq>
right, if $ROOTPATH includes the IP address, then you are done
14:54
<ogra>
but it shouldnt be called ROOTPATH then :)
14:55
<jammcq>
agreed
14:55
<ogra>
hmm
15:09
<vagrantc>
dude.
15:09
that is so a bug i filed.
15:09
and recently patched.
15:09
<rasmuson>
Hi. I have a problem with local device support. Usb sticks normally works on the ws, but my new cowon iaudio f2 music player doesn't.
15:11
It is detected on the ws as usb mass storage device, but /etc/fstab is empty. Any ideas?
15:12
<vagrantc>
ogra, jammcq: see http://bugs.debian.org/387808
15:13
i just filed a fix the other day.
15:13
<ogra>
heh, yeah, that looks good
15:14
<jammcq>
there ya go
15:14
<ogra>
still the code below that section ooks wrong
15:14
<vagrantc>
i should've CCed pkg-ltsp-devel
15:14
<ogra>
*look
15:15
<vagrantc>
ogra: trust me, i have gone over it a million times. it's right. at least in the version in debian
15:15
<ogra>
if you set NFSROOT as a kernel parameter it will berak
15:15
<vagrantc>
no.
15:15
<ogra>
*break
15:15
<jammcq>
why would you want to do that anyway?
15:15
<vagrantc>
well, yeah. but you should specify nfsroot=
15:15
not NFSROOT=
15:16
that's what in-kernel nfsroot support uses
15:16
<ogra>
yeah
15:16
<ltsppbot>
"rasmuson" pasted "dmesg after inserting usb device" (18 lines) at http://pastebot.ltsp.org/109
15:16
<ogra>
but since you use NFSOPTS ... i assume people will guess NFSROOT :)
15:17
<vagrantc>
someone else started using NFSOPTS
15:17
they should just use nfsroot=
15:17
<ogra>
wasnt me
15:17* ogra whistles
15:17
<vagrantc>
i think it was the initial implementation
15:17
<jammcq>
rasmuson: does that usb device have a partition table?
15:17
<ogra>
yeah, mailny jbaileays fault
15:17
*mainly
15:18
<vagrantc>
the initial NFSROOT support in initramfs-tools didn't really look at any other existing NFSROOT conventions, as far as i can tell.
15:19
<ogra>
well as the NFSOPTS doesnt now ...
15:19
<rasmuson>
jammcq: I guess so, it works when it is inserted in a normal non thin client. It uses the hole disc though, no partitions. I can mount it manually on the thin client.
15:20
<jammcq>
yeah, that's probably the problem. I think local device support is expecting to mount sdc1, not sdc
15:21
<rasmuson>
hmmm
15:23
<vagrantc>
looking at the ltspfs in debian and ubuntu, i don't see why it wouldn't work with whole devices
15:23
<jammcq>
I think it's just a udev thing
15:23* rasmuson doesn't know if he can change the layout without destrying the mp3 player
15:23
<jammcq>
rasmuson: what version of LTSP ?
15:23
<rasmuson>
4.2 update 3 i think
15:24
latest
15:24
<jammcq>
yeah, that's the problem
15:24
<rasmuson>
?
15:24
<jammcq>
ltsp-4.2 doesn't support filesystems that aren't on a partition
15:24
ltsp-5 *should* work
15:24
<vagrantc>
hopefully :)
15:24
<rasmuson>
is ltsp-5 out?
15:24
<jammcq>
yeah
15:24
it's in Ubuntu and Debian
15:25
<rasmuson>
this is fedora 6
15:25
<jammcq>
well, you can load the ubuntu or debian LTSP-5 on top of fedora if you want
15:25
<rasmuson>
can I install it along with the older version?
15:25
<ogra>
it will work with ubuntus kernel/udev ...
15:25
i cant say anything bout debian ...
15:25
<jammcq>
you'll end up with ubuntu/debian bits being served by a fc6 server
15:26
<vagrantc>
but then they have to build ltspfs for fedora, too, no?
15:26
<ogra>
yes
15:26
<jammcq>
http://wiki.ltsp.org/twiki/bin/view/Ltsp/DownLoads#Ltsp_5_on_Other_distributions
15:26
<ogra>
with all the scripts
15:26* rasmuson looks
15:27
<rasmuson>
That'
15:28
Thats great
15:28
<ogra>
we should take out the fixed paths everywhere where its not security critical to have them think ...
15:28
<jammcq>
umm
15:28
<rasmuson>
I read about ltsp going into fedora 7. Is this still the plan?
15:29
<jammcq>
rasmuson: sadly, ltsp-5 isn't gonna make it in fc7. not enough time to get it done
15:29
lets hope for fc8
15:29
<ogra>
nobody from the fedora guys has contacted any ltsp devs yet
15:29
<rasmuson>
sad
15:29
<jammcq>
oh, there's been contact
15:29
<ogra>
for code merges of for implementation help
15:29
<vagrantc>
we've offered
15:29
<jammcq>
lots of contact, nobody actually doing the work
15:30
the fedora guys really seem to want it, they just don't have the resources to throw at it
15:30
<vagrantc>
i guess it's better there's no work done than lots of work done on a totally incompatible fork :)
15:30
<ogra>
well, there hasnt been any contact on a developer level yet or do i miss anything ?
15:30
<jammcq>
sure there's been contact, Warren Togami was at our hackfest in september
15:30
<ogra>
beyond warren and eric i mean
15:31
<jammcq>
and there's been several private emails between me and some fedora mgmt guys
15:31
they were looking to hire somebody and wanted to know if I had some suggestions
15:31
<ogra>
well, stil i dont see any movement apart from that
15:31
yes, i remember it ...
15:31
<jammcq>
they even wondered if I'd consider the job
15:31
<ogra>
we talked ...
15:31
<jammcq>
and they tried to coax eHarrison, but he wasn't interested either
15:31
<ogra>
do you ?
15:32
:)
15:32
<jammcq>
heh, no
15:32
ogra: you want the job?
15:32
<ogra>
i'm happy, thanks :)
15:33
i like to work for companies whose business models have a future ;)
15:33
RH has a big past though :)
15:33
<rasmuson>
so, should i go with the latest ubuntu packages, or what do you recommend?
15:34
<jammcq>
well, i'm not sure whether local devs would work at all
15:34
they work great when ubuntu is both the server and the client
15:34
<ogra>
well, do you have a tool like alien in RH ?
15:34
would be the easiest way to try
15:35
<jammcq>
he's already got ltspfs, because he's got k12ltsp
15:35* vagrantc suspects fuse-dev incompatibilities
15:35
<jammcq>
but, his server is expecting lbus, but ubuntu's client isn't
15:35
<vagrantc>
yeah, but that ltspfs is pretty different, with lbus and all that
15:35
<ogra>
hmm, right the versions might not match
15:35
jammcq, there are a bunch of other scripts
15:36
<jammcq>
yeah, it'd prolly be a real mess
15:37
<ogra>
ogra@edubuntu:~$ dpkg -L ltspfs|grep bin/
15:37
/usr/bin/ltspfs
15:37
/usr/bin/lbmount
15:37
/usr/sbin/ltspfsmounter
15:37* vagrantc hopes for binaries with longer names for the future of ltspfs
15:37
<ogra>
wel, only one script and a suid binary
15:37
vagrantc, we can start using underscores as well :)
15:37
<rasmuson>
what about a changing what i have? Where should i look?
15:39* ogra thinks about asciiart with ls /bin
15:49
<vagrantc>
rasmuson: switching distro? or what?
15:50
<rasmuson>
vagrantc: no, I'm too lazy for that. I thougt of making a little hack in a script if possible
15:50
<vagrantc>
rasmuson: alien would be your laziest way to go
15:50
oh, right.
15:50
fuse dev compatibility
15:50
nevermind
15:51
rasmuson: just port the ltspfs package to fedora, and then you can ask people to port the rest of ltsp
15:52
<rasmuson>
vagrantc: sorry, i dont think i'm capable of that :)
15:53
<vagrantc>
rasmuson: you might surprise yourself, but i have no idea how difficult packaging is on fedora.
15:53
it can't be that bad, or people would've all switched to other distros by now, no? :)
15:53
<rasmuson>
he he
15:53
ubuntu?
16:01LarryP has joined #ltsp
16:08
<rasmuson>
vagrantc, jammcq, ogra: thanks for explaining the problem. I hope fedora will include ltsp soon.
16:22
<vagrantc>
rasmuson: you should nudge them to do so, as a fedora user
16:23
the more nudges they get from their users, one would hope they prioritize it a little higher somehow
16:28sepski has quit IRC
16:31rasmuson has quit IRC
16:31muh2000_ has quit IRC
16:31joebob777as7 has quit IRC
16:31Egyptian[Home] has quit IRC
16:31ogra has quit IRC
16:31milosz has quit IRC
16:31shogunx has quit IRC
16:31nf1 has quit IRC
16:32rasmuson has joined #ltsp
16:32muh2000_ has joined #ltsp
16:32joebob777as7 has joined #ltsp
16:32Egyptian[Home] has joined #ltsp
16:32ogra has joined #ltsp
16:32nf1 has joined #ltsp
16:32milosz has joined #ltsp
16:32shogunx has joined #ltsp
16:33muh2000_ has quit IRC
16:35muh2000_ has joined #ltsp
17:09LarryP has quit IRC
17:18cliebow__ has joined #ltsp
17:42Roy_BR has joined #LTSP
17:42Roy_BR has left #LTSP
17:42privet has quit IRC
17:46privet has joined #ltsp
17:47bobby_C has quit IRC
18:17J45p3r has joined #ltsp
19:10cliebow_ has quit IRC
19:55cliebow__ has quit IRC
20:28
<stillflame>
vagrantc: reinstalling the server fixed it. weird.
20:32jvin248 has joined #ltsp
20:47ogra has quit IRC
20:48ogra has joined #ltsp
21:19primeministerp has joined #ltsp
22:03J45p3r has quit IRC
22:23itknight has joined #ltsp
22:23
<itknight>
hi all
22:24
has anyone done an installation on Xubuntu
22:29vagrantc has quit IRC
22:36itknight has left #ltsp
23:15Bhaskar1 has joined #ltsp
23:23mistik1_ has joined #ltsp
23:23mistik1 has quit IRC
23:24mistik1_ is now known as mistik1