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


Channel log from 20 April 2008   (all times are UTC)

00:33asac_ has joined #ltsp
00:44asac has quit IRC
00:45asac_ is now known as asac
01:01robbie has joined #ltsp
01:01
<johnny>
warren, are you about?
01:11robbie has quit IRC
01:43justme____ has joined #ltsp
01:44
<justme____>
anyone know why on a fresh install of 8.04 alternate LTSP, that I can't run any applications in the gui as the main user added during installation?
02:08indradg is now known as indradg|afk
02:25deavid has joined #ltsp
02:42Q-FUNK has joined #ltsp
02:46
<Q-FUNK>
ahoi!
02:49deavid has left #ltsp
02:51justme____ has quit IRC
02:51justme____ has joined #ltsp
03:00ogra has quit IRC
03:01ogra has joined #ltsp
03:19Egyptian[Home] has quit IRC
03:20Egyptian[Home] has joined #ltsp
03:23
<Q-FUNK>
hm
03:42mccann has quit IRC
04:12Pascal_1 has joined #ltsp
04:56Pascal_1 has quit IRC
05:28
<Q-FUNK>
ogra: ?
05:34wwx has quit IRC
05:43wwx has joined #ltsp
05:56DonSilver has joined #ltsp
06:03justme____ has quit IRC
06:44
<Q-FUNK>
anybody up for a small backporting challenge?
06:46
I'm trying to backport the -geode libDDC patch to -amd 2.7.7.7. the part for the LX almost applies as-is, but the part for the GX would require cherrypicking more recent commits for extra GX fixes, before the patch applies.
06:49DonSilver has quit IRC
06:53DonSilver has joined #ltsp
06:53
<Q-FUNK>
http://q-funk.iki.fi/debian/pool/x/xserver-xorg-video-geode/geode_libddc.patch
06:53
http://snapshot.debian.net/archive/2008/04/12/debian/pool/main/x/xserver-xorg-video-amd/xserver-xorg-video-amd_2.7.7.7-1.dsc
07:41Pascal_1 has joined #ltsp
07:50Pascal_1 has quit IRC
07:54Pascal_1 has joined #ltsp
07:57Pascal_1 has quit IRC
08:06mccann has joined #ltsp
08:12J45p3r has joined #ltsp
08:42petre has joined #ltsp
08:45indradg|afk has quit IRC
08:45indradg|afk has joined #ltsp
08:53eharrison has joined #ltsp
08:55
<petre>
morning all
08:55
<eharrison>
good morning
08:56
<warren>
hey
08:56
any topics anyone want to talk about today?
08:56
Only thing to report (beyond what's already in rawhide) is ogra/vagrant added code to ltspfs upstream yesterday to allow it work in LDM_DIRECTX mode
08:57
now... I still haven't managed to get it to work on fedora
08:57
we're missing something else
08:57
but I'll push ltspfs-0.5.1 anyway
08:57
<petre>
warren, any issues with encrypted file systems & ltsp on f9?
08:57
<warren>
petre: what is encrypted?
08:58
<petre>
I'm starting a f9 install to try out ltsp5 on; one option, set by default, is encrypted file system
08:58
<warren>
yeah, that one wont bother ltsp
08:58
<petre>
server's fs
08:58
k
08:58
<warren>
that only makes sense on a laptop though
08:59
if you're doing a permanent server don't encrypt it
08:59
you need to type the passphrase in every time it boots
08:59
<eharrison>
I cut my work terminal over to F9/LTSP5. I now have an incentive to work on ltspfs ;-)
08:59
<petre>
that was my thought, but I'm thinking about it in terms of 'simple installation procedure for not-very-technical folks
09:00
but if it's the default in f9, then people will be running into it
09:00
if it's not a problem on the server, then it doesn't matter
09:00
ltsp-wise, that is
09:01
<warren>
it isn't a problem in runtime
09:01
but people might find it annoying that they have to type hte passphrase in MANUALLY at every boot
09:01
<petre>
right, but that's a fedora issue not an ltsp issue
09:02
while I might/would argue it should be off by default, that's a discussion for another #channel
09:05
<warren>
after we enable nbd boot, we could even make the root filesystem over the network be encrypted
09:05
although maybe that's pointless
09:06
<petre>
would that increase the load on the client?
09:06
<warren>
yes
09:06
and you still don't have a protected kernel/initrd
09:06
so it is pointless
09:07
petre: do you have wiki edit access?
09:07
<petre>
yes
09:07
which I've put to massive use making two minor edits in the past week (sigh)
09:07
<warren>
any other topics people want to talk about?
09:07
otherwise we can finish
09:08
<eharrison>
Nothing more than grunt work... my test production server is running well
09:11
<warren>
OK, we're done then. I'll post notes and upgrade ltspfs in F-9
09:11
eharrison: you running with LDM_DIRECTX or no?
09:11
<eharrison>
currently no.
09:12
<warren>
does everyone want to keep this meeting time?
09:13
the person who chose this meeting time never comes anymore =)
09:13
I'm OK either this time or an hour later
09:13* petre likes it
09:13
<eharrison>
the person who chose this meeting time is not a morning person ;-)
09:14* eharrison is ok with this time, but an hour later would be nice
09:14deavid has joined #ltsp
09:15
<petre>
what does the DIRECTX in LDM_DIRECTX mean?
09:16
<eharrison>
warren: LDM_DIRECTX appears to be working fine here
09:16
<warren>
eharrison: with ltspfs?
09:17
eharrison: petre: how about a compromise, 30 minutes later
09:17
<petre>
okay by me
09:17
<warren>
eharrison: so 10:30AM eastern, 7:30AM pacific
09:17
<eharrison>
I have not gotten that far. I left my usb key @ work, I'm looking for another one
09:18
okay, 7:30am pacific it is
09:18
<warren>
ok
09:18
hmm
09:18
I can't build the newer ltspfs yet.
09:18
down for maintenance
09:19
<petre>
grrr, f9 install failed, same place as on dedicated server
09:20
<warren>
?
09:22
<eharrison>
warren: can you post your ltspfs src.rpm somewhere I can fetch it?
09:23
<warren>
I didn't make a 0.5.1 yet
09:23
i'll do it now
09:25
<petre>
warren, I'm planning to do repeated installs of ltsp as part of working on ltsp-server-initialize;
09:25
<warren>
petre: there might be a faster way, like using virtual machines
09:26
<petre>
does it matter whether I'm working on f9 or f8, as I can't seem to get f9 installed anywhere
09:26
<warren>
snap and go back to the point immediately before you ran ltsp-server-initialize
09:26
petre: what are your F9 install error messages?
09:26
<petre>
I'm using vm's now
09:26
<warren>
we need to know exactly how it is failing
09:26
whta kind of vm?
09:26
<petre>
I *knew* you were going to ask that
09:26
vmware
09:27
It fails at some japanese library, but I hit the wrong key before I could write it down
09:27
failed in the same place on my AMD64 server, though, too
09:28
<warren>
are you using the preview release?
09:28
which installer exactly are you usinG?
09:28
<petre>
I think so...not quite sure; where do I look?
09:28
<warren>
what did you download and when?
09:29
<petre>
I downloaded a DVD iso about a month ago
09:30
<warren>
you have something extremely old then
09:32
<petre>
that would be a good thing, as hopefully newer would avoid the problem.
09:32
<warren>
http://torrent.fedoraproject.org/
09:32
<petre>
I'll go get something new now
09:32
<warren>
looks like the only way to download the PR is torrent at the moment
09:33
<petre>
torrent is fine
09:38dan_young has joined #ltsp
09:39
<petre>
downloading now, only ~7 hours to go (sigh)
09:39* petre is tempted by Comcast's 50Mb download offer
09:40
<warren>
speak of the devil
09:40
=)
09:40
<petre>
warren, so is there value in testing on f8?
09:40
<warren>
petre: a lot less
09:40
<petre>
that is, should I bother, or just wait until I've got a working f9
09:40
<warren>
petre: F8 will never have full LTSP support because we can't ship in in F8
09:41
I'll be backporting everything to F8
09:41
and RHEL5 later
09:41
but F9 is the focus
09:42
<eharrison>
hi dan
09:43
<dan_young>
hey, miss anything interesting?
09:44
<eharrison>
we'll start at 7:30am next week ;-)
09:45
<dan_young>
OK by me! ;-)
09:51
<warren>
eharrison: http://people.redhat.com/wtogami/temp/ltspfs-0.5.1-0.fc9.src.rpm
09:51
i'll post the notes to the wiki
09:54
<eharrison>
warren: looks like something is eating the "LTSPFS_TOKEN", all of the other debugging steps are working fine. I'll give the new package a spin and see if it magically fixes it ;-)
09:54
<warren>
eharrison: with LDM_DIRECTX or not?
09:54
<eharrison>
either way.
09:54
<warren>
eharrison: 0.5.1 is specifically to make it work with LDM_DIRECTX
09:54
ooh
09:55
<eharrison>
so far, ltspfs has not been working for me at all
10:05
<petre>
ogra, ping
10:11vagrantc has joined #ltsp
10:17Pascal_1 has joined #ltsp
10:25Pascal_1 has quit IRC
10:34
<Q-FUNK>
warren: what does your symbolic link look like in -geode?
10:34
relative path or absolute path?
10:42
<eharrison>
Q-FUNK: I assume you mean amd_drv.so -> geode_drv.so? If so, it is relative.
10:43
<warren>
Q-FUNK: relative
10:43
<Q-FUNK>
ok
10:44
we must have something different in the ubuntu X, because that fails
10:45
<warren>
what is it?
10:45
<eharrison>
warren: for ltspfs, we are missing a dependency on xorg-x11-utils
10:46
<warren>
eharrison: within the client chroot?
10:46
eharrison: or the server side?
10:46
<eharrison>
client-side chroot
10:46
<warren>
eharrison: is that the only thing we're missing?
10:46
<eharrison>
to be determined. Still not working here, but that may very well be my fault ;-)
10:49
warren: everything is working except having the icon appear on my desktop. It is indeed automagically mounting on the server side
10:50
<warren>
eharrison: perhaps vagrantc/ogra know more about how that's supposed to work?
10:51
https://fedorahosted.org/k12linux/wiki/Meetings
10:51
<vagrantc>
eharrison: up for another ltsp hackfest before and/or after oscon in june?
10:51
<warren>
go ahead and put more details on this page if you find them today
10:52
I have to drive, will be back online in 20-30 minutes
10:53
<eharrison>
vagrantc: yes, I'm sure a hackfest will happen. Jim, Scott, Warren, & probably others will be in town.
10:53
<vagrantc>
oops. july
10:54
<warren>
be back in about 30
10:54
<vagrantc>
eharrison: yeah, that's why i figured i'd try and get some ltsp action going on
10:57cliebow has joined #ltsp
11:11
<eharrison>
vagrantc: in ltspfs (0.5.1), where is the magic that puts the icon on a user's desktop? I see the example for kde, but I don't see how it is supposed to work under gnome
11:11
<vagrantc>
eharrison: "just works" for GNOME, in my experience.
11:12
<eharrison>
okay, back to the code I go ;-)
11:14
<vagrantc>
eharrison: is it mounting in /media/USERNAME/DEVICENAME ?
11:14
eharrison: or /tmp/.USERNAME-ltspfs/DEVICENAME ?
11:15
<eharrison>
/tmp/.USERNAME-ltspfs/DEVICENAME
11:16
<vagrantc>
eharrison: is /usr/bin/lbmount setuid root ?
11:17
<eharrison>
vagrantc: it was not. giving that a spin...
11:17
that was it, thanks ;-)
11:18
<vagrantc>
ta-da!
11:19
i just put in the room use request at freegeek for both the weekend before and after oscon.
11:20
<johnny>
hi folks
11:28Pascal_1 has joined #ltsp
11:28
<vagrantc>
johnny: no way you can make the time to come out to portland in july?
11:29
<johnny>
time is not problem, money is
11:31
at least for now
11:31
i may end up with a "real" job and have money by then
11:32
so my smoketest was almost a success, i am happy .. just need to get a fix or two into the chroot generator
11:36
<Q-FUNK>
ah. is root explicitely disabled in hardy? I created a passowrd, but then when trying to login on console, I get the "password expired. contact admin" message.
11:36
(inside the chroot)
11:37
<vagrantc>
Q-FUNK: i've seen similar issues on debian sid ... i've played with chage to work around it
11:37Pascal_1 has quit IRC
11:48
<Q-FUNK>
Account expires : Jan 02, 1970
11:48
??
11:49
<vagrantc>
yeah.
11:49
:)
11:54dan_young has quit IRC
12:10
<cliebow>
btw:HowdY Eric!
12:11
<eharrison>
hi Chuck!
12:11
<cliebow>
sounds like you have a handle on the ltsp5..
12:12
<eharrison>
yeah, slowly but surely ;-)
12:25vagrantc has quit IRC
12:34Pascal_1 has joined #ltsp
12:38otavio has joined #ltsp
12:41Pascal_1 has quit IRC
13:02eharrison has quit IRC
13:51Shingoshi has quit IRC
13:52DonSilver has quit IRC
14:03cliebow_ has joined #ltsp
14:19cliebow has quit IRC
14:25cliebow__ has joined #ltsp
14:27cliebow_ has quit IRC
14:40likuidkewl has joined #ltsp
14:41likuidkewl has left #ltsp
15:02DonSilver has joined #ltsp
15:28petre has quit IRC
15:47
<warren>
[warren@newcaprica ltspfs-test]$ automake
15:47
src/Makefile.am:8: compiling `cdpinger.c' with per-target flags requires `AM_PROG_CC_C_O' in `configure.ac'
15:47
Makefile.am: required file `./ChangeLog' not found
15:47
hmm
15:47
where is the name "ChangeLog" coming from
15:48
<Q-FUNK>
*sigh* this -geode issue in Hardy is really killing me.
15:49
<warren>
Q-FUNK: ?
15:50
<johnny>
warren, i had that issue too
15:50
i have no idea what is requiring it
15:50
i ended up just touching a ChangeLog file
15:51
<warren>
johnny: it is a hard coded name in automake itself =(
15:52
so the `AM_PROG_CC_C_O' issue is not fatal?
15:52
<johnny>
no
15:52
touch ChangeLog and all is well
15:52
pretty sure i got that one..
15:53jammcq has joined #ltsp
15:53
<jammcq>
hey all
15:53
<johnny>
wtf is that hardcoded on automake
15:54
<warren>
johnny: it is a GNU standard apparently
15:54
we didn't have this problem earlier because ltspfs had a blank ChangeLog file
15:58
<Q-FUNK>
warren: either the symlink is not doing its thing or X core misses some bits to even try the geode driver
15:58
warren: but of course Hardy comes with X 1.4, not 1.5 so misisng bits would not be entirely surprising
16:00
still, all this mess might mean that Hardy will have to be force-downgraded to -amd 2.7.7.7
16:00
(or preferably my own 2.7.7.8, which has leio's backport of the libDDC patch needed by Koolu and other poor sods using the ION 603)
16:01cliebow__ has quit IRC
16:16joebaker has left #ltsp
16:17
<johnny>
warren, i imagine the gnu standard will change now they are going with bzr :)
16:21
<warren>
johnny: Changelog vs ChangeLog has nothing to do with bzr
16:21
johnny: Changelog is what I had hard-coded into mkdst
16:21
yeah, ltsp-trunk already uses ChangeLog
16:22
ldm-trunk as well
16:22
this is a bug in mkdst
16:22
sigh
16:22
<leio>
I don't think the xorg-server bits matter, if Hardys server doesn't even have -amd detection. Something else is doing the detection, maybe ltsp?
16:23
<warren>
Q-FUNK: xorg server has a hard coded list of pci ids to driver name if xorg.conf doesn't exist.
16:23
<johnny>
oh.. i see
16:23
<Q-FUNK>
warren: xf86AutoConfig ?
16:23
<warren>
Q-FUNK: /hw/xfree86/common/xf86AutoConfig.c
16:23
yeah
16:23
<Q-FUNK>
I patched that
16:23
<warren>
X.org 1.5 upstream has the pciid for LX but not GX
16:24
Q-FUNK: what did you change/add?
16:24
<Q-FUNK>
added the amd line from git
16:24
http://launchpadlibrarian.net/13589055/171_xf86AutoConfig_geode_addition.diff
16:24
1.4 lacks it
16:25
<johnny>
warren, so.. are you making tarballs?
16:25
<warren>
note that will work with only LX
16:25
right?
16:25
<johnny>
or just rpm?
16:25
<warren>
johnny: rpm needs the tarballs
16:26
<johnny>
where are they?
16:26
<warren>
johnny: I'm generating tarballs as needed using mkdst --from-tag=NAME-VERSION
16:26
<Q-FUNK>
warren: noted but nobody provided the one for GX, so far
16:26
<johnny>
but are you hosting them?
16:26
<warren>
Q-FUNK: I tried to get the one for GX added into xorg-server but it was rejected
16:26
<Q-FUNK>
warren: are those taken from PCI id numbers or something?
16:26
<warren>
Q-FUNK: because it is the same as NSC
16:26
Q-FUNK: yes
16:27
Q-FUNK: upstream wont accept the GX one until the geode driver supports nsc and cyrix.
16:27
<Q-FUNK>
warren: then it should be adifferent one
16:27
GX != GX1
16:27
<warren>
AMD Geode GX2 is the same pciid as NSC I tihnk
16:27
<Q-FUNK>
is it?
16:27
<warren>
yes
16:27
look in hwdata, you'll see it in the table
16:27
<Q-FUNK>
Gadi would know. he has a few of each.
16:27
ok
16:28
<warren>
I fought to get GX2 added to xf86AutoConfig.c
16:28
but they said no
16:28
<Q-FUNK>
oh well
16:28
<warren>
despite NSC and cyrix being less common than GX2
16:28
<Q-FUNK>
silly, though, since nsc and cyrix currently don't build due to lacking pciaccess support and other things
16:29
<warren>
yeah
16:29
no point in fixing those old drivers
16:29
<Q-FUNK>
no, NSC is extremely common
16:29
<warren>
oh?
16:29
OK, then good thing to not break it
16:29
ajax refused adding GX2 because it would prevent NSC from falling back to vesa.
16:29
<Q-FUNK>
soekeris boards and loads of other stuff was built with the Geode SC which is essentially a GX1 + CS5530a in a single chip
16:30
those became available just before NSC's Geode team was bought by AMD
16:31
so they are marked as NSC but were mostly commercialized by AMD
16:32
e.g. our own dbe60 used an sc2200.
16:32
<warren>
GX2 is still marked NSC in the pci I think
16:32
<Q-FUNK>
yes, because it is their design
16:32
<leio>
warren: the thing is, that LX works without anything in xf86AutoConfig.c per Q-FUNK. So something else seems to do the detection
16:32
<warren>
anyway, only way we'll fix this is by integrating NSC/cyrix into geode
16:32
<Q-FUNK>
which, to complicate things, AMD's marketing droids renamed GX after they bougt the line from NSC
16:33
<leio>
and fixing it for GX is not the issue here currnetly
16:33
<warren>
If you run X without xorg.conf then it picks a driver with xf86AutoConfig.c
16:33
Fedora doesn't attempt to configure xorg.conf
16:33
I don't know if Ubuntu tries
16:33
<Q-FUNK>
leio: but only in LTSP. as standalone, X won't find it unless I tip it in xorg.conf that "amd" or "geode" is the driver to use. then it picks it up just fine
16:34
<warren>
are you sure it isn't trying to run something else to configure X?
16:34
<Q-FUNK>
I'm wondering if LTPS might be doing its own parsing of lspci and other hwdata source
16:34
<leio>
Q-FUNK: right, so for non-LTSP usage you need that patch added to xf86AutoConfig.c, but apparently with LTSP it used to work fine without, so something has the binding to amd driver done other than AutoConfig.c
16:35
<Q-FUNK>
yup. it feels that way
16:35
something in LTSP knows about our PCI ID
16:35
<warren>
It depends if the way ltsp is packaged in Ubuntu is attempting to configure_x or not.
16:35
Fedora never tries it
16:35
Ubuntu might be doing it
16:35
LTSP source has no pci tables in it
16:35
<Q-FUNK>
I think thta starting with this release, it relies on DDC and PCI ID
16:35
<leio>
Q-FUNK: if you run standalone, it doesn't find the driver right? What if you patch "amd" into xf86AutoConfig.c but have geode driver?
16:36
<Q-FUNK>
leio, I could try. lemme see.
16:36
<warren>
running just "X"
16:36
with no xorg.conf
16:37
if boolean_is_true "$CONFIGURE_X" ; then
16:37
${CONFIGURE_X_COMMAND:-/usr/share/ltsp/configure-x.sh}
16:37
fi
16:37
<leio>
Q-FUNK: or actually, what if just even claim "amd" in xorg.conf but have geode driver - does it pick up the symlink and work? Cause if it does then if something else is doing the detection it should be detecting "amd" but work with geode driver fine
16:37
<warren>
amd and geode are working interchangably here
16:38
if [ -z "$CONFIGURE_X" ] && [ -x /usr/bin/X ] && [ -z "$X_CONF" ]; then
16:38
export CONFIGURE_X=True
16:38
fi
16:38
looks like ltsp_config runs this by default
16:38
it tries to run /usr/share/ltsp/configure-x.sh
16:38* warren has to rip that out
16:39
<warren>
ah, it hasn't been working on fedora since it can't write the xorg.conf output file
16:39
and configure-x.sh tries to read a bunch of debian specific config files
16:40
Q-FUNK: configure_x.sh
16:40
LANG=C Xorg -configure -novtswitch :1 > ${TEMPFILE} 2>&1
16:40
it starts with the output of that attempting to configure X before it runs it.
16:40
Q-FUNK: you might try disabling this script and seeing if it helps
16:41* warren driving
16:45
<johnny>
warren, that script doesn't work for me at all
16:46
i put CONFIGURE_X=F in my ltsp-setup
16:49
<Q-FUNK>
moving xorg.conf away or telling xorg.conf to use amd fails on a standalone host.
16:51
but telling xorg.conf to use geode works.
16:51
leio: so it seems that the symbolic link doesn't do its trick.
16:52
that's on the turkish recipe.
16:59
weird how neither the symlink or the magic patch work
17:00
I probably missed that simply because I've always used an xorg.conf on workstations, to make sure I would get the right default keyboard map and bit depth.
17:01
and on LTSP, using -amd, it just works. it fails now for the first time.
17:15
hm. even with the magic number patch, it fails to find -amd
17:16
so now, whether I use -amd or -geode, they still need to be specified in xorg.conf
17:17
except in LTSP/Hardy, where -amd just magically works and -geode simply won't.
17:21vagrantc has joined #ltsp
17:28Q-FUNK has quit IRC
17:42deavid has left #ltsp
18:12DonSilver has quit IRC
18:56
<warren>
sounds like a bug specific to ubuntu then
18:56
X on F9 is using geode or amd
19:03Shingoshi has joined #ltsp
20:24hari_ has joined #ltsp
20:39
<hari_>
does anyone know how to install ltsp5 in ubuntu without $ sudo ltsp-build-client , cause it needs internet connection. thx
21:10hari__ has joined #ltsp
21:16cpunches2 has joined #ltsp
21:17cpunches2 has joined #ltsp
21:18cpunches2 is now known as cpunches
21:21indradg_ has joined #ltsp
21:32leio has quit IRC
21:32hari_ has quit IRC
21:33leio has joined #ltsp
21:38indradg|afk has quit IRC
22:01hari_ has joined #ltsp
22:03J45p3r has quit IRC
22:08hari__ has quit IRC
22:15hari_ has quit IRC
22:27jammcq has quit IRC
22:59rezon8 has joined #ltsp
23:07r3zon8 has quit IRC
23:17abadger1999 has quit IRC
23:18abadger1999 has joined #ltsp
23:33makghosh has joined #ltsp
23:36abadger1999 has quit IRC
23:39abadger1999 has joined #ltsp
23:40cpunches has quit IRC
23:46daya has joined #ltsp
23:48makghosh has quit IRC
23:53makghosh has joined #ltsp
23:56vagrantc has quit IRC