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


Channel log from 26 February 2008   (all times are UTC)

00:17
<lns>
/j #gsoc
00:18
lol
00:38
<warren>
grrr!!!
00:38
# Etherboot PXE (only 5.4))
00:38
elsif substring (option vendor-class-identifier, 0, 9) = "Etherboot-5.4"
00:38
dhcpd.conf seems that it cannot recognize "Etherboot-5.4"
00:38
watching the packet dump the client did indeed transmit "Etherboot-5.4"
00:38
those characters must not be legal to the server or something
00:39
<lns>
warren, i've seen that with certain versions of pxe too
00:39
<warren>
lns, oh?
00:40
<lns>
it's like a single version below what the latest intel one was - and it would hang at tftp
00:40
<warren>
also something changed (could have been in fedora) causing qemu-kvm's network boot to fail the first 4 tries, succeed on the 5th
00:40
<lns>
haha..nice
00:41
<warren>
this was working fine in December
00:41
I don't know what changed
00:41
lns, too bad you can't chain load Etherboot-5.4 or higher
00:41
<lns>
haha!
00:42
that would be interesting
00:42
<warren>
lns, if the BIOS can't handle PXE like linuxbios, then having a newer version of Etherboot still can't PXE
00:42
lns, so it is very annoying that the PXE capable Etherboot has the same vendor class identifier
00:43
<lns>
pxe capable eitherboot?
00:43
ether*
00:43
does pxe chainload in some cases?
00:43nikok has joined #ltsp
00:44
<warren>
lns, Jim has a recipe where you can actually run a newer version of etherboot from an older version
00:44
lns, but it wont help you in cases where you want to do PXE from a non-PXE capable BIOS
00:45BGomes has quit IRC
00:46
<lns>
i haven't come into any etherboot capable bioses before
00:46
i'm not very familiar with the process like pxe
00:47
<warren>
i'm trying to ship a dhcpd.conf that handles any client by default
00:47
which MIGHT have been possible if the stupid dhcp server could substring match
00:47
properly
00:48MacIver has quit IRC
00:48
<lns>
i never understood how that could ever be an issue...seems like such a trivial thing
00:49
but i've heard that older windows nt servers had issues with that too
00:51
<warren>
i'm reading the sources of dhcpd
00:52
<nikok>
any gnome users here?
00:53
<warren>
oh!
00:53
lns, hah
00:53
lns, <daniel_hozac> warren: how would a substring from char 0 to 9 ever match a 13 character long string?
00:53
<warren> daniel_hozac, oh!
00:53
<lns>
=p
00:53
<nikok>
is there some file or files that gets loaded evry time user logs in? a file that keeps track of desktop backgrounds etc...?
00:54
<lns>
nikok, gconf?
00:56
night all
01:17
<nikok>
this might be a little complicated and odd but... what i'm trying to do is to set some sort of a "default profile" that evry thin client loads when they log in, and what makes it a bit complicated is the fact that every one in our ltsp environment uses the same user account to log in... so my goal is that every user can modify their desktop (backgrounds for example) as they want but nothing wouldn't be writed in the profile. after they log out or
01:18
i think u get the idea... but the question is, is this possible? and if it is, where should i start?
02:45plamengr has joined #ltsp
02:48toscalix has joined #ltsp
02:53sep has quit IRC
02:56sep has joined #ltsp
03:03exodos has joined #ltsp
03:12mikkel has joined #ltsp
03:33pdjbarber has joined #ltsp
04:06achandrashekar has joined #ltsp
04:06
<achandrashekar>
hello anyone got a recommendation for thin clients?
04:06
how is the entc1000?
04:13pdjbarber has quit IRC
04:21
<toscalix>
haven't try it
04:33plamengr has quit IRC
04:35alekibango has joined #ltsp
04:50achandrashekar has quit IRC
04:54pdjbarber has joined #ltsp
05:03
<artista_frustrad>
nikok, you could hook up a script at the login manager so that whem the user leaves the .gnome2 directory is erased
05:05pdjbarber has quit IRC
05:35Guaraldo has joined #ltsp
05:35Nubae has joined #ltsp
05:44
<nikok>
artista_frustrad: i tried to add rm -rf /home/user/.gnome2/* to /etc/gdm/PostSession/Default... no notable difference, still remembers the background i choosed @ last login. i've been told that /etc/gdm/PostSession/Default is run every time user logs out... can someone confirm this?
05:57pdjbarber has joined #ltsp
06:01K_O-Gnom has joined #ltsp
06:04mhterres has joined #ltsp
06:11gonzaloaf has joined #ltsp
06:18mhterres has quit IRC
06:19mhterres has joined #ltsp
06:29
<Nubae>
hmmm... klik sounds interesting... virtualisation on the application level
06:29
couldn't that be used to get java working on 64 bit ubuntu?
06:29
maybe I'm killing a fly with a bazooka, but its worth ask
06:29
ing
06:42Guaraldo has left #ltsp
06:56toscalix has quit IRC
06:56Shingoshi has quit IRC
06:56Patina has quit IRC
06:56dberkholz has quit IRC
06:56toscalix has joined #ltsp
06:56Shingoshi has joined #ltsp
06:56Patina has joined #ltsp
06:56dberkholz has joined #ltsp
06:57dberkholz_ has joined #ltsp
06:59wdg has joined #ltsp
07:05loather-work has quit IRC
07:06cliebow_ has joined #ltsp
07:06artista_frustrad has quit IRC
07:06primeministerp has quit IRC
07:06primeministerp has joined #ltsp
07:07artista_frustrad has joined #ltsp
07:08Topslack has quit IRC
07:09Shingoshi has quit IRC
07:09loather-work has joined #ltsp
07:10Topslack has joined #ltsp
07:10dberkholz has quit IRC
07:44slidesinger has joined #ltsp
07:45MacIver has joined #ltsp
07:48jobdrb has joined #ltsp
07:49mccann has joined #ltsp
07:50K_O-Gnom has quit IRC
07:50K_O-Gnom has joined #ltsp
07:52
<jobdrb>
anyone know if we could in ltsp5 use a local harddisk partition ? like in ltsp4.2 local_device_02 =/dev/hda3:hd3
07:53elisboa has quit IRC
08:00
<ogra>
not like that, but you can modify the udev rules inside the chroot to also accept non removable devices
08:00elisboa_ is now known as elisboa
08:08mikkel has quit IRC
08:11otavio_ has joined #ltsp
08:12otavio has quit IRC
08:14ogra has quit IRC
08:19otavio_ is now known as otavio
08:21ogra has joined #ltsp
08:22jammcq has joined #ltsp
08:22
<jammcq>
g'morning kidz
08:28Gadi has joined #ltsp
08:39
<rjune>
jammcq !
08:39
Gadi: !
08:44
how goes it jammcq
08:44
Gadi: can beer be kosher?
08:49otavio has quit IRC
08:51
<Nubae>
on hardy, local cd roms are mounted, but appear on everyone's screens with the following message: You do not have authorisation to view scsi-cdrom
08:51
can I disable that somehwere?
09:05Pascal_1 has joined #ltsp
09:06
<Pascal_1>
hello
09:07
anybody know how to make works ldap authentication and mount samba share with libpam-mount ? i cant make it works
09:12
<Nubae>
there are some howtos if you search the net
09:23
<Pascal_1>
i found some but there is no good solutions for me, if you have some url
09:30
<Nubae>
this is ldap and nis: http://ldots.org/ldap/
09:31
<Pascal_1>
thanks
09:31daduke has quit IRC
09:32
<Pascal_1>
i'll see later the libpam-mount problem
09:32
i try this before thanks
09:40daduke has joined #ltsp
09:47
<Gadi>
rjune: !!!
09:48
Nubae: local cdroms as in on the thin client or the server?
09:48
<rjune>
Gadi: !
09:49
can beer be kosher?
09:49
<Gadi>
rjune: I drink plenty of it :)
09:49
<rjune>
heh
09:49
<Gadi>
most beer is kosher
09:49
<rjune>
just checking.
09:49
<Gadi>
its only if they start blending in grape juice or wines that spoil it :)
09:50
but, who drinks the blended crap anyway?
09:50
so.... when are you buying the next round?
09:52
Pascal_1: http://wiki.ltsp.org/twiki/bin/view/Ltsp/WinIntegration#pam_mount_conf_and_LDM_ssh
10:04
<Nubae>
gadi client
10:05exodos has quit IRC
10:10alumno10 has joined #ltsp
10:10
<alumno10>
hi
10:10
i cant listen sound in clients
10:11
all of them are ebox with sis7019 sound system
10:18
<Nubae>
Gadi, any ideas about local cdroms showing up unwanted on the clients (when put into client computers, not server)
10:18
?
10:23spectra has joined #ltsp
10:29mccann has quit IRC
10:30mccann has joined #ltsp
10:31
<Gadi>
Nubae: LTSP5?
10:32
<Nubae>
yeah, sorry alpha 5 hardy
10:33
<Gadi>
hmm...
10:33
is it a hardy chroot?
10:33
or is the chroot gutsy?
10:34
<Nubae>
hardy and hardy
10:34
<ogra>
Nubae, the desktop switched from gnome-vfs to gvfs ... the ltsp patches are not ported yet
10:35
<Gadi>
huh. there ya go
10:35
an informed answer :)
10:35
<ogra>
its hardy :)
10:35
<Nubae>
ogra, sorry that's too technical for me, what does that mean in practice?
10:35
<ogra>
supposed to be in flux and break here and there
10:35
<Gadi>
Nubae: its alpha
10:35
:)
10:35
<Nubae>
yeah yeah, that I got
10:35
<Gadi>
will be fixed by release
10:35
<Nubae>
ok, its a bug
10:35
<Gadi>
right
10:36
a "porting bug"
10:36
:)
10:36
<ogra>
there is a new technology that doesnt know about ltsp yet, we need to teach it to :)
10:36
but its known and on my todo list
10:36
<Gadi>
ogra is gvfs more, er stable than gnome-vfs was
10:36
?
10:36
er is
10:37
Ive got gutsy users with nautilus screaming out of control at random
10:37
<Nubae>
ok... btw, someone gave me a fix for 64 bit java, that supposedly works in hardy and gutys
10:37alumno10 has quit IRC
10:37
<Gadi>
which I think is a gnome-vfs bug
10:37
<Nubae>
https://help.ubuntu.com/community/Java64
10:38
<ogra>
Gadi, might be, no clue about gvfs ... i can tell you after i ported the patches :)
10:38
<Gadi>
:)
10:38
<ogra>
since that'll require me to dig deep in the dirt of gvfs code
10:38
<Gadi>
gnome-vfs has driven me nuts on several occasions
10:38
happy to hear there's a new girl in town
10:38
:)
10:38
<ogra>
i doubt gvfs is much better yet its still very young
10:39
<Gadi>
and yet used in an LTS release?
10:39
ballsy
10:39
<ogra>
(i personally wouldnt have pulled it into hardy ... )
10:39
but the desktop team thinks its suitable
10:41
<laga>
Nubae: how is java in 64bit broken? seen some odd java issues on amd64 here
10:42
<Nubae>
for me it doesn't work as a browser plugin at all
10:42
on 64 bit
10:43
but supposedly that fix works, I'm still compiling
10:45Pascal_2 has joined #ltsp
10:45staffencasa has joined #ltsp
10:49rjune has quit IRC
10:56mccann has quit IRC
10:56mccann has joined #ltsp
11:00Egyptian[Home] has quit IRC
11:03indradg has joined #ltsp
11:04Egyptian[Home] has joined #ltsp
11:05pdjbarber has quit IRC
11:05rjune has joined #ltsp
11:06Pascal_2 has quit IRC
11:09dberkholz_ is now known as dberkholz
11:09
<jobdrb>
HELLO ALL! IS THERE A WAY TO USE A LOCAL HD PARTITION IN LTSP-5, LIKE IN 4.2 (local_device_02 =/dev/hda3:hd) ?????
11:10
<Gadi>
heh, maybe if you yell loud enough:)
11:10
<laga>
yes, more caps makes for better answers
11:10
<Gadi>
seriously, tho: its all in the udev rule
11:11
look in: /opt/ltsp/i386/etc/udev/rules.d/88-ltsp.rules
11:12
if you use debian, chances are you may have vagrantc's patch to not allow non-removable devices
11:12
<warren>
high definition?
11:12* Gadi laughs WITH warren - honest
11:12
<Gadi>
:)
11:12* warren just woke up
11:13* Gadi hands warren a beer and a fedora
11:13
<Gadi>
(to hide his bloodshot eyes)
11:14
<warren>
and slashdot tells me that Fish Can Count to Four
11:14
<Gadi>
maybe the remedial fish
11:14
my fish beat up your honor student
11:15
<warren>
Gadi, Jewish education is that good huh?
11:15mccann has quit IRC
11:15
<Gadi>
lol
11:16mccann has joined #ltsp
11:16
<Gadi>
only the best education for our fish
11:16
:)
11:16
<warren>
Gadi, but then your fish has to walk on Saturdays/
11:16
?
11:17wdg has quit IRC
11:17
<Gadi>
yeah, but at least they can walk and count passed four at the same time ;)
11:18
<Nubae>
or past four ;-)
11:18klausade has quit IRC
11:20vagrantc has joined #ltsp
11:25otavio has joined #ltsp
11:28klausade has joined #ltsp
11:28
<Gadi>
Nubae: right
11:28
:)
11:28
fingers faster than brain
11:28
:)
11:32cesar_ has joined #ltsp
11:33klausade has quit IRC
11:33
<cesar_>
hello people
11:34
good afternoon for me!
11:34
:-)
11:37artista_frustrad has quit IRC
11:41klausade has joined #ltsp
11:43dtrask has joined #ltsp
11:44* dtrask loves snow, but this is getting ridiculous!
11:55mccann has quit IRC
11:59mccann has joined #ltsp
12:03
<warren>
dtrask, ?
12:10
<dtrask>
warren: yeah?
12:10
<warren>
snow
12:10
<dtrask>
warren: I'm teaching so I'm back and forth
12:11
<warren>
ah
12:11
<dtrask>
hows the snow there?
12:11
are you in MA?
12:12
This storm and then 2 more behind it...Sat-Sun and Mon
12:13
oh well....skiing will be good this weekend
12:15
<Nubae>
is there a problem with firefox 3 and printing?
12:16
<ogra_cmpc>
Nubae, i dont think it was ported over yet look at launchpad there are likely bugs
12:16
note that ff3 only entered main last weekend
12:17
its very likely that not all parts are done yet
12:18
<warren>
dtrask, not snowing now in somerville, MA
12:19* ogra_cmpc feels like something moved him to a rainforest this winter ... no snow, only rain and storm in mid-germany
12:19
<ogra_cmpc>
and i think it was only three timews below zero yet
12:21
<Nubae>
sunny blue skies here (Southern Spain)
12:21
<laga>
ogra_cmpc: hey. do you know anything in LTSP that'd mess with the power button? i don't get that "do you want to shutdown?" dialog window on my fat client when i push the power button
12:22
<ogra_cmpc>
laga, thats a thing of hal, gnome-power-manager and gnome-session
12:22
<Nubae>
from mailing list: While I realize that the advise from this list had been to pay someone
12:22
from Canonical if there are problems, this is, of course, completely
12:22
unacceptable.
12:23
funny
12:23
<ogra_cmpc>
make sure hal and g-p-m are running
12:23
<laga>
ogra_cmpc: ok, i was just wondering if there's anything LTSP-specific. like capturing the event and sending it to the server
12:23
<ogra_cmpc>
Nubae, which one ?
12:23
<Nubae>
I guess the expectation now is that open source must be free
12:23* ogra_cmpc has no mailer on this machine
12:23
<ogra_cmpc>
laga, nope
12:24
<Nubae>
edubuntu-users
12:24
<ogra_cmpc>
heh, i'll answer that
12:24
<laga>
ogra_cmpc: great, thanks.
12:24
<ogra_cmpc>
i'f i'm back in the office (might not be today anymore though)
12:24
<Nubae>
that was i reference to the usb problem in gutsy ;-)
12:25
amazing the outcry that one little bug can create
12:25
<ogra_cmpc>
well, preople ofthen dont get: ubuntu != canonical
12:26cyberorg has quit IRC
12:27
<rjune>
heh
12:27
ogra_cmpc: how many times do you tell people that
12:28
<ogra_cmpc>
once a week at least
12:28
fighting windmills and trying to overcome myths
12:28
<Nubae>
the thing I dont get is, its damn cheap to get support from canonical
12:29
and people complain on top of it
12:29
<ogra_cmpc>
Nubae, development is a totally different thing though
12:29
i dont know the prices we take for dev stuff, but i suspect its quite expensive
12:30
<Nubae>
would canonical not advise on how to fix that bug though?
12:30
is that really development?
12:30
<ogra_cmpc>
but then the development we do for money is usually *very* specific
12:31
sure the support team would advise how to work around the bug
12:31
(if there is a way to)
12:31
<Nubae>
they could even login to the machine and do it themselves
12:32
anyway I'm off home, goodnight...
12:34Egyptian[Home] has quit IRC
12:41vagrantc has quit IRC
12:49Egyptian[Home] has joined #ltsp
13:03cyberorg has joined #ltsp
13:19Egyptian[Home] has quit IRC
13:20Egyptian[Home] has joined #ltsp
13:42dtrask has quit IRC
13:43Egyptian[Home] has quit IRC
13:46Egyptian[Home] has joined #ltsp
13:50otavio has quit IRC
13:51alekibango has quit IRC
13:53Big-K has quit IRC
14:00scrovy has joined #ltsp
14:00
<scrovy>
hi
14:01
I have a question related to mounting an NFS root filesystem
14:01otavio has joined #ltsp
14:01
<scrovy>
does anybody understand about this topic?
14:04alekibango has joined #ltsp
14:05scrovy has left #ltsp
14:21otavio has quit IRC
14:34Gadi has left #ltsp
14:39toscalix has quit IRC
14:41mccann has quit IRC
14:41otavio has joined #ltsp
14:46cliebow_ has quit IRC
14:46otavio has quit IRC
15:02alekibango has quit IRC
15:02mhterres has quit IRC
15:04bobby_C has joined #ltsp
15:13vagrantc has joined #ltsp
15:21K_O-Gnom has quit IRC
15:39Egyptian[Home] has quit IRC
15:50Stevecar has joined #ltsp
15:53cesar_ has quit IRC
16:01mccann has joined #ltsp
16:15jammcq has quit IRC
16:38slidesinger has quit IRC
17:21mopey is now known as babys4sale
17:26J45p3r has joined #ltsp
17:29babys4sale has quit IRC
17:31bobby_C has quit IRC
17:44Shingoshi has joined #ltsp
17:48ogra_cmpc has quit IRC
17:48ogra_cmpc has joined #ltsp
17:50ogra_cmpc has quit IRC
17:50ogra_cmpc_ has joined #ltsp
17:52ogra_cmpc_ is now known as ogra_cmpc
17:56staffencasa has quit IRC
18:14mccann has quit IRC
18:25J45p3r_ has joined #ltsp
18:25J45p3r has quit IRC
18:26ogra_cmpc has quit IRC
18:33cliebow has joined #ltsp
18:36vagrantc has quit IRC
18:47mccann has joined #ltsp
18:55dtrask has joined #ltsp
18:56
<dtrask>
warren: Is it snowing there now? ;-)
19:00
<cliebow>
not here/....
19:09ogra_cmpc has joined #ltsp
19:28spectra has quit IRC
19:29
<dtrask>
yet....it's coming! ;-)
19:34J45p3r_ has quit IRC
19:40mccann has quit IRC
19:51dtrask has quit IRC
19:56otavio has joined #ltsp
20:22jammcq has joined #ltsp
20:32otavio has quit IRC
20:48eric has joined #ltsp
20:48
<eric>
hi all.
20:48
Anyone care to help me out? I have edubuntu server installed and a wyse thin client on the same network...
20:48
unsure how to make the wyse client get a desktp from the server...
20:54eric_ has joined #ltsp
20:57Stevecar has left #ltsp
21:02cliebow has quit IRC
21:07Stevecar has joined #ltsp
21:08eric has quit IRC
21:18eric_ has quit IRC
21:19chupacabra has quit IRC
21:20chupacabra has joined #ltsp
21:29vagrantc has joined #ltsp
21:31eric_ has joined #ltsp
21:36
<vagrantc>
warren: unless i have no idea what your vmclient script does, i think it would be totally useful to include on the server
21:37
<warren>
vagrantc, even if it has requirements that are irrelevant to server?
21:38
vagrantc, I put it at the base in a new directory because it is clearly another package
21:38
<vagrantc>
warren: ah, you're making it a third package?
21:39
<warren>
vagrantc, ltsp-vmclient is the package name for fedora
21:39
<vagrantc>
warren: i might just include it in the example scripts like the qemu-ltsp script i already distribute with the ltsp-server package
21:39* warren looks
21:39
<vagrantc>
warren: i put that in the documentation directory
21:39
<warren>
oh
21:40
<vagrantc>
it used the user-space networking and commandline arguments to configure it
21:40
<warren>
vagrantc, in my case I'm able to make a functional package that can be installed and used
21:40
<vagrantc>
warren: sure. wouldn't be hard to do on debian either...
21:41
warren: seems very... hardcoded.
21:41
<warren>
vagrantc, perhaps we should generalize vmclient
21:41
so that it uses different VM types if available
21:42
<vagrantc>
that would be good. the qemu-ltsp script also accepted a few commandline arguments...
21:42ogra has quit IRC
21:42ogra has joined #ltsp
21:43
<warren>
user-space networking can do PXE?
21:43
My solution uses brctl
21:43ogra_cmpc has quit IRC
21:43
<warren>
which is very nice because you can add an arbitrary ethernet interface to that virtual bridge and it instantly has access
21:44
<vagrantc>
warren: no, i load the kernel directly, so it didn't actually test the dhcp/tftp part of the process
21:44
yes, i've used brctl
21:44
<warren>
isn't the dhcp/tftp part of the process really useful to test?
21:44
<vagrantc>
yes, it is useful to test, but that doesn't mean it's useless to test the rest without configuring an actual bridge...
21:45* vagrantc finds that it is often good to have a variety of ways to do things
21:45
<vagrantc>
and to know what the advantages and limitations of each of those ways are
21:46
as far as i can tell, setting up a brctl method requires root access, no?
21:46
<warren>
that's true
21:46
<vagrantc>
the user-space networking doesn't.
21:47
some versions of qemu don't support PXE booting.
21:47
it was introduced sometime within the last year, if i'm not mistaken
21:47
<warren>
and some didn't support it because they broke it =)
21:48
<vagrantc>
and some versions of qemu didn't work even if you loaded the correct etherboot off a virtual CD image or something ... it just didn't work.
21:48
but i really like that it's easier to test the whole process now.
21:49
<warren>
I'll put vmclient into a k12linux specific directory for now then
21:49
I don't have time to work on that part now
21:49
I'm hitting a deadline next week
21:50
<vagrantc>
ah well. in the future, i'd like to work with you to make it more generalized and flexible.
21:50
<warren>
sure
21:50
I'd like that
21:50eric_ has quit IRC
21:51ogra_cmpc has joined #ltsp
21:53mccann has joined #ltsp
21:55
<vagrantc>
warren: no need to move it to a fedora-specific directory i think ...
21:57chupacabra has quit IRC
21:59
<vagrantc>
warren: does fedora's udev use "udevinfo" or "udevadm info" ?
21:59
or neither? :)
22:00
<warren>
vagrantc, F8 has udevadm
22:00
vagrantc, want me to look at F9?
22:00
<vagrantc>
warren: i think udevadm is the newer one
22:00
warren: so i'd wager that F9 has it
22:00
<warren>
RHEL5 has udevinfo
22:01
<vagrantc>
there's some code in ltspfs ... i think i'll just re-write it to support both
22:01
a pretty simple if statement ...
22:01
debian 4.0 uses udevinfo, and debian unstable still is using udevinfo but might switch someday and may as well be future-proof. :)
22:02
<warren>
[root@newcaprica ~]# udevadm info
22:02
missing option
22:02
?
22:02
<vagrantc>
uh-oh.
22:02
<warren>
I don't know how udev works.
22:03
<vagrantc>
i'm looking at http://bugs.debian.org/456327 and trying to figure out how to support all cases.
22:03
<warren>
let me send you a copy of the man page
22:03
or rather
22:04
would you prefer the source?
22:04
<vagrantc>
warren: what version of udev are you using?
22:04
<warren>
F8 has udev-118-1.fc8
22:04
F9 has udev-118-5.fc9
22:04
<vagrantc>
http://bugs.debian.org/cgi-bin/bugreport.cgi?msg=5;filename=tmpoP0wxM;att=1;bug=456327
22:04
was the patch from ubuntu
22:05
<warren>
ah
22:05
F8 had 116, we issued a udev update a few days ago.
22:05
<vagrantc>
debian's using udev 0.114 and i think ubuntu is using 0.117
22:05
well, debian etch is 0.105
22:06* vagrantc wonders how heavily patched the ubuntu version is
22:06
<vagrantc>
or the debian version, for that matter
22:06* warren wonders if Ubuntu's udev pushed patches back upstream =)
22:06* warren nudges jcastro
22:07
<vagrantc>
in the bug report, it is claimed that "tools merged into a single binary upstream"
22:07
warren: your upstream version is basically ... 118 ... 0.118 ... ?
22:07* warren looks at the source
22:08
<warren>
vagrantc, it says 118
22:08* vagrantc wonders where debian/ubuntu get the 0. from
22:09
<warren>
+export $(/sbin/udevadm info -qenv -n ${DEVICENAME})
22:09
vagrantc, what is this command supposed to do?
22:09
what is an example of DEVICENAME?
22:09
<vagrantc>
warren: do you have both udevadm and udevinfo ?
22:09
<warren>
vagrantc, apparently yes
22:09* vagrantc looks at the ltspfs code
22:10
<warren>
vagrantc, ah, udevinfo is a symlink to udevadm, probably for backwards compat purposes.
22:11
<vagrantc>
warren: ok, i think DEVICENAME is basically the /dev/hda, /dev/sda type stuff
22:11
warren: but "udevadm info" doesn't work ?
22:12
<warren>
[root@newcaprica ~]# udevadm info
22:12
missing option
22:12
according to the manpage it expects an option
22:12
<vagrantc>
ah
22:12
udevadm info -qenv -n /dev/hda
22:12
(or whatever device you have)
22:12
<warren>
[root@newcaprica ~]# udevadm info --version
22:12
118
22:12
<vagrantc>
should spit out a bunch of variables
22:13
<warren>
ltspbot, pastebot
22:13
<ltspbot>
warren: "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.
22:13
<warren>
oh wait
22:13
I don't want to paste this info
22:13
<vagrantc>
heh
22:13
<warren>
somebody could write code that targets me
22:13
=)
22:13
<vagrantc>
well, does it spit out a bunch of NAME=VALUE type things?
22:14
<warren>
yeah
22:14
<vagrantc>
ok, that seems like it's working, then.
22:14
<warren>
ok
22:14mccann has quit IRC
22:14
<warren>
udevinfo -qenv -n /dev/sda
22:14
<vagrantc>
by the way, udevinfo --version on debian etch/4.0 spits out 105 ...
22:14
<warren>
this does the same thing
22:14mccann has joined #ltsp
22:15
<warren>
looks like a back compat symlink
22:15
<vagrantc>
sure. so i'm guessing udevadm is the future, and it would be nice to keep a little backwards compatibility.
22:15
just in case the distro doesn't implement the backwards compatible symlink
22:16
warren: is udevadm in /sbin ?
22:16johnny_ has joined #ltsp
22:16
<warren>
yes
22:21
<vagrantc>
committed a patch that should support both older and newer udev versions
22:22
<warren>
ah, just figured out how our readonly-root mode works
22:22
/etc/rwtab contains dirs, files or empty dirs to map to bind mounts in tmpfs
22:22
/etc/rwtab.d/ allows you to define more
22:22
<vagrantc>
nice.
22:23
i wonder how hard this would be to port to debian
22:23
<warren>
doesn't look complicated at all
22:23
<vagrantc>
conceptually seems that way ...
22:24
i suppose some of the directories *might* be different ...
22:25
<warren>
hmm, i guess no reason to have LVM in my client chroot...
22:25
<vagrantc>
heh
22:25
<warren>
vagrantc, no problem, just define your own default rwtab
22:25
vagrantc, not optimizing, it is spewing error messages during startup distracting me from the more important error messages
22:25
<vagrantc>
warren: got a source or vcs URL for the readonly root stuff?
22:28
<warren>
vagrantc, http://koji.fedoraproject.org/packages/initscripts/8.64/1/i386/initscripts-8.64-1.i386.rpm
22:28
vagrantc, unpack that
22:28
vagrantc, see /etc/rc.sysinit and /etc/rwtab*
22:28
<vagrantc>
warren: it's integrated directly into your initscripts?
22:29
hrm.
22:29
<warren>
vagrantc, yes, readonly-root mode is an official mode of Fedora
22:29
since FC3 or so
22:29
<vagrantc>
yes, i remember the announcement
22:29
i got many people saying "that sounds like what you're doing with debian"
22:30
warren: ok, so how would i unpack this rpm ? ... with the "rpm" command? ... my rpm is pretty rusty. not that it was ever very sharp.
22:32
i know how to do the equivalent with .deb's in muscle memory
22:32
<warren>
vagrantc, do you have rpm2cpio?
22:33
vagrantc, something like mkdir something; cd something; rpm2cpio whatever.rpm | cpio -id
22:34
<vagrantc>
warren: got it.
22:34
thanks
22:34* vagrantc missed the -d bit
22:35
<warren>
interesting...
22:35
vagrantc, does debian use dash btw?
22:36
screen_session line number 17
22:36
TTY_NUM=${SCREEN_NUM#0}
22:36
<vagrantc>
warren: not by default. i tried it with a thin-client the other day and it knocked a ridiculous amount of time off the boot session.
22:36
<warren>
on fedora TTY_NUM becomes "--"
22:37
<vagrantc>
weird.
22:37
<warren>
i'm not even sure how that syntax is supposed to work
22:38
#0 means what?
22:38
OH!
22:38
-- is used as a parameter separator
22:38
but in this case it is actually using it
22:39
<vagrantc>
SCREEN_NUM=07 and output 7 ... or SCREEN_NUM=12 and output 12
22:39
i think that's what it's trying to do
22:40
ubuntu's used dash for ages
22:49
<warren>
vagrantc, I asked our people about dash, they insist that it doesn't help much at all
22:50
vagrantc, instead we seem to be focusing on reducing redundancy. an intel guy pointed out a few places where we were doing things stupid hundreds of times in a row during our boot.
22:50
F9 switched to upstart
22:50chupacabra has joined #ltsp
22:50
<vagrantc>
warren: in debian, it knocks off a very, very noticeable amount of time.
22:51
<warren>
so far we've discovered that it makes our shutdown 2 seconds faster
22:51
<vagrantc>
warren: someone actually did a bunch of bootspeed analysis on debian a couple years ago, and it knocked something around 30 seconds off of a 2.5 minute boot time
22:52
<warren>
vagrantc, I dunno then, one of our most respected engineers told me that he personally did testing and dash did not improve things much at all, he said.
22:52
maybe fedora is different in some regard
22:53
<vagrantc>
maybe so
22:54
warren: maybe they tested on too fast equipment?
22:54
where the differences were harder to notice
22:54
<warren>
vagrantc, I dunno
22:54* warren sits on an old quad core box
22:54
<vagrantc>
nor do i. but i cannot deny what i see with debian.
22:55* vagrantc mostly uses an 866MHz machine
22:56
<warren>
vagrantc, when I have some breathing room I'm going to test dash again
22:57
Oh
22:57
I see the problem
22:57
<vagrantc>
when i noticed how much of a difference it made, i started thinking of how to make it the default at least for ltsp
22:57
ubuntu's got it easy in that regard
22:58
<warren>
my client's boot is already very fast
22:58
the slowest part by far is udev
23:00
vagrantc, and we discovered that bash is really stupid in cases like
23:00
vagrantc, [ -e file1 -a -e file2 ]
23:00
vagrantc, it checks for the existence of both files even if the first doesn't exist.
23:00
<vagrantc>
heh
23:01
<warren>
checking for the existence of files is real slow
23:01* vagrantc always uses [ ] && [ ]
23:01
<warren>
seek times
23:01
etc.
23:01
yeah, me too
23:01
it is easier for the human brain to parse too
23:02Stevecar has left #ltsp
23:03* vagrantc just added an additional check for the existance of a file
23:03
<vagrantc>
in the udevadm/udevinfo stuff
23:03
but the binary gets used immediately after you check for it's existance ...
23:03
so it can't be too bad.
23:05
<warren>
is /var/run/pulse anywhere in the scripts?
23:06
<vagrantc>
don't know
23:06
<warren>
W: main.c: Warning: home directory of user 'pulse' is not '/var/run/pulse', ignoring.
23:06
not sure what this is coming from
23:06* vagrantc would recommend grep
23:06
<warren>
oh
23:07
<vagrantc>
:)
23:07
<warren>
that's hard coded in our pulseaudio binary
23:07
<vagrantc>
heh
23:07
<warren>
vagrantc, does your binary have any home directory hard coded?
23:07
<vagrantc>
no idea.
23:07
<warren>
our pulseaudio daemon creates a pulse user with home directory /
23:07
this might be a packaging bug
23:11
<vagrantc>
wow. debian's m68k buildd's just got very much faster.
23:14* vagrantc suspects some distcc going on
23:21oh207 has quit IRC
23:21oh207 has joined #ltsp
23:25Egyptian[Home] has joined #ltsp
23:42
<warren>
vagrantc, do you know the command that lets you run another command on a target VT?
23:43
vagrantc, I've used it in the past to do "thatcommand 9 /bin/bash" for example, to get full job control on VT9
23:44
<vagrantc>
openvt ?
23:45
<warren>
yeah!
23:45
I found it just as you said that
23:46
failing to login as root on the client no matter what I do in the chroot
23:46
so just bypassing the login
23:53
<vagrantc>
SCREEN_NN=shell doesn't work?