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


Channel log from 19 September 2007   (all times are UTC)

00:56captain_magnus has quit IRC
00:56captain_magnus has joined #ltsp
01:07plamengr has joined #ltsp
01:08plamengr has left #ltsp
01:09
<loather-work>
god, fedora has so much unneeded cruft in their base install
01:10ari_stress has quit IRC
02:06Daggett has joined #ltsp
02:18highvoltage has joined #ltsp
02:20mikkel has joined #ltsp
02:23subir has joined #ltsp
02:36Egyptian[Home1 has joined #ltsp
02:37Egyptian[Home] has quit IRC
02:47Amaranth has joined #ltsp
02:53captain_magnus is now known as captain_maGNUs
03:06Egyptian[Home] has joined #ltsp
03:07Egyptian[Home1 has quit IRC
03:16Amaranth has quit IRC
03:22mistik1 has quit IRC
03:29Amaranth has joined #ltsp
03:39plamengr has joined #ltsp
03:52putty_thing has joined #ltsp
03:56Amaranth has quit IRC
03:59Egyptian[Home] has quit IRC
03:59Egyptian[Home] has joined #ltsp
04:10putty_thing has quit IRC
04:22Egyptian[Home] has quit IRC
04:23Egyptian[Home] has joined #ltsp
04:25Amaranth has joined #ltsp
05:11muh2000 has quit IRC
05:37Clown-MG has joined #ltsp
05:52Avatara has joined #ltsp
06:18muh2000 has joined #ltsp
06:25plamengr has quit IRC
06:26EoinH has joined #ltsp
06:26plamengr has joined #ltsp
06:27
<EoinH>
Anyone here have experience with LTSP across wireless?
06:33mikkel has quit IRC
06:40ogra has quit IRC
06:43Amaranth has quit IRC
06:44ogra has joined #ltsp
07:01
<EoinH>
Anyone here have experience with LTSP across wireless?
07:01Blinny has joined #ltsp
07:04
<ogra>
apart from "it doesnt work because there is no wireless netboot protocol like PXE or etherboot ?"
07:08
<EoinH>
The machines are booting OK. We have a wireless link between buildings.
07:08Amaranth has joined #ltsp
07:14muh2000 has quit IRC
07:22jernst has joined #ltsp
07:25Guaraldo has joined #ltsp
07:27jernst has left #ltsp
07:41matte_ has joined #ltsp
07:42matte_ is now known as asdrubale
07:46
<ogra>
EoinH, that doesnt solve the issue of the need to have them netboot and retrieve kernel and initramfs from the net
07:46
there is simply no protocol to do that in wlan
07:49
<EoinH>
Sorry, I have not explained well. We have two buildings. The LTSP server and three clients are in one. All working perfectly. A point to point wireless links another building with two clients. The two remote clients boot OK, but fuse, local devs and on occasions nfs fail.
07:49
<ogra>
ah
07:49
so they are normal PXE clients
07:49
but use a WLAN link
07:49
indeed thats different :)
07:50
localdev and nfs use normal TCP connections ...
07:50
<EoinH>
Yes the remote clients are identical boxes with built in PXE. They work OK when hard wired in.
07:51
<ogra>
do you filter anything on the access points ?
07:51
<EoinH>
Not to my knowledge.
07:51
<ogra>
what bandwith do you have on the wlan ?
07:52
<EoinH>
The APs are bog standard out of the box configured Linkssys and Buffalo. The run 'g'. Not tested the speed.
07:52
<ogra>
well, that would be my first look :)
07:53vpro has joined #ltsp
07:53
<ogra>
under 10M things will surely get bad
07:53
make some speed tests between the buildings :)
07:53
<EoinH>
Pings are 99%+ What tool wold you suggest for testing the wireless speed.
07:54
<ogra>
an nfs mount and dd'in´g a file
07:54
measure the time for a 100M file and compute the outcome :)
07:54
do the same in the LAN without wlan involved and you see the speed difference
07:55
<EoinH>
sorry offline a tick
07:56
<asdrubale>
hi, not all pci-lan support pxe and netboot is very slow, what's best way for to do work anything terminal server?
07:58
what are our experiences with many pcs all differents?
07:59
have you changed all pci-lan cards?
08:02
many network cards support pxe, but they haven't a rom.. what's the best solution?
08:03
<vpro>
does usb-sticks work in ltsp4.2 and can I have separate access server and aplication server?
08:04
this works fine in ltsp5 but I don't know how it works in ltsp4.2
08:08
asdrubale: what do you mean by "they haven't a rom"?
08:13OvaKill has quit IRC
08:13F-GT has joined #ltsp
08:20Guaraldo has quit IRC
08:20Guaraldo has joined #ltsp
08:21asdrubale has quit IRC
08:21asdrubale has joined #ltsp
08:22fernando196 has quit IRC
08:24fernando196 has joined #ltsp
08:25subir has quit IRC
08:28bobby_C has joined #ltsp
08:32primeministerp has joined #ltsp
08:40Gadi has joined #ltsp
08:47elisboa has joined #ltsp
08:50captain_maGNUs is now known as captain_magnus_Z
08:51fernando196 has quit IRC
08:54
<sbalneav>
Morning all
08:54
<Guaraldo>
morning sbalneav...
08:55
<jammcq>
Scotty !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
08:55
<sbalneav>
Hey jammcq!
08:58tux_440volt has joined #ltsp
08:59EoinH has left #ltsp
09:01tiagovaz has joined #ltsp
09:04ogra has quit IRC
09:04ogra has joined #ltsp
09:10
<asdrubale>
what's the best place where to install etherboot? flash on ide? rom on network card?
09:10
or buy a network card with rom already installed?
09:11
<Blinny>
asdrubale: I've had good luck with ROM on network card.
09:14elisboa has quit IRC
09:14elisboa has joined #ltsp
09:14
<asdrubale>
Blinny, i dont understand: if possible you say more about it
09:16
<Blinny>
asdrubale: In that case, you should just buy a network card with ROM already installed.
09:18
I just got a back of 10 Intel PXEs off of eBay for $25
09:18* sbalneav can hardly wait for his book to arrive.
09:18
<sbalneav>
I ordered "Python & XML" last night from Amazon.
09:19
<Gadi>
the snake wins
09:19
<Blinny>
You ruined the suspense!
09:19* Gadi only read the last page
09:19
<Gadi>
:)
09:22ogra has quit IRC
09:24
<asdrubale>
Blinny, if i have a p1, p2 that supported boot from lan?
09:26mistik1 has joined #ltsp
09:27
<Blinny>
asdrubale: Booting from LAN is an i386 thing that has been around for ages. If you put a bootable card in there, then yes.
09:28
<asdrubale>
Blinny, thank a lot!!
09:29* asdrubale inglese macheronico ownz!
09:30ogra has joined #ltsp
09:46tux_440volt has quit IRC
09:52monteslu has joined #ltsp
09:58MRH2 has joined #ltsp
10:02vagrantc has joined #ltsp
10:03
<sbalneav>
Morning vagrantc
10:03asdrubale has quit IRC
10:04
<vagrantc>
howdy
10:04* vagrantc checks the status of ltspfs in debian
10:06
<ogra>
seems its gotten through :)
10:08
<vagrantc>
well, it's through one of the many phases of being uploaded ...
10:09
<rjune_>
Ahoy Maties
10:10staffencasa has joined #ltsp
10:10
<MRH2>
hi I can;t seem to get any more connections to the terminal server - anything I can restart without kicking everyone else off?
10:10
<sbalneav>
Arrrrgh
10:11
MRH2: Which version of LTSP are you running?
10:12putty_thing has joined #ltsp
10:12
<MRH2>
4.2 - not a config issue - guess is that xdmcp / gdm has gone bad
10:12
at the server side
10:12
<sbalneav>
Well, or you've maxed out the sessions
10:12
<MRH2>
nope
10:13
<sbalneav>
How do you know you haven't?
10:13MagicStorm has joined #ltsp
10:13
<MRH2>
i've killed off all the screensavers as i've seen then cause the occasional odd behaviour
10:13
<sbalneav>
Check for idle gdmlogins
10:13
<MRH2>
if i kill gdm on an active session what happens
10:13
<sbalneav>
Also, when you kill them off, a session isn't released immediately
10:14
MRH2: Well, that user gets kicked off, obviously.
10:14
If you kill the root gdm, then they all get kicked off.
10:15
<MRH2>
yeah thought so
10:15
<Gadi>
try killing the gdmgreeters
10:15
<sbalneav>
right.
10:15
if you've got any of them, they're using sessions up.
10:17
<vagrantc>
wow, the m68k build beat out all the others for ltspfs ... that's unusual
10:20
<MRH2>
nope no gsdm-greeter processes
10:20
gdm
10:21Avatara has quit IRC
10:22mikkel has joined #ltsp
10:23
<MRH2>
i am going to take a gamble on xfs
10:26
<sbalneav>
How many people are logged in?
10:27
<MRH2>
not many
10:27
nope on xfs
10:27
i have a gnome-session process i can;t kill that might be the cause
10:27
<Gadi>
it should be gdmgreeter
10:27
no dash
10:28
dont restart xfs
10:28
it will wreak havoc with existing connections
10:28
you can reload xfs
10:28
but not restart
10:29
hmm... fakeroot is throwing up errors all of a sudden
10:29
grr...
10:29
<ogra>
vagrantc, do we have any automatic code for encrypted swap if cryptsetup is installed ?
10:30
cryptsetup is in main in ubuntu now ... i could add it to the default
10:30
<vagrantc>
ogra: i don't think there's any automatic code, but it would be easy to add.
10:31
<MRH2>
ain't no gdmgreeter running
10:31* vagrantc checks
10:31
<ogra>
!meep .. wrong answer ... beta freeze is tomorrow :)
10:31
<ltspbot`>
ogra: Error: "meep" is not a valid command.
10:31
<ogra>
pfft ...
10:31* ogra pokes ltspbot`
10:31
<ogra>
vagrantc, that means no for me :)
10:32
i'll leave it in recommends for now
10:32
<sbalneav>
ogra: Merge in the fixes for cdpinger I did?
10:32
<ogra>
not yet
10:32
i'm just on the ltsp branch
10:32
merged that though
10:32
<sbalneav>
k
10:32
Any changes you've done? Sync required?
10:32putty_thing has quit IRC
10:33
<ogra>
i switched deps from mknbi to mkelf
10:33
nothing else yet
10:33
<sbalneav>
ah, ok
10:33putty_thing has joined #ltsp
10:33
<ogra>
oh, and dropped nfs-kernel-server to be a recommends
10:33
<sbalneav>
Let me know when you're done, I'll merge
10:33
<ogra>
no need to have it :)
10:34
<sbalneav>
BenC get the squashfs bug... squashed?
10:34
Ba-dump dump
10:35
<vagrantc>
ogra: i'll be sure to get the auto-enabled encrypted swap in debian's next upload ... it's a good idea.
10:36tiagovaz has quit IRC
10:36
<Gadi>
have any of you deb roller seen this? fakeroot, while creating message channels: Invalid argument
10:36
This may be due to a lack of SYSV IPC support.
10:36tiagovaz has joined #ltsp
10:37
<sbalneav>
Weird
10:37
<MRH2>
anything else I can check before i have to restart X for everyone?
10:37
<sbalneav>
No, never seen that Gadi.
10:37
MRH2: The usual suspects, ram, filled up /tmp, etc
10:38
<MRH2>
i guess i need to think about upgrading... lol
10:38
a restart i shall go
10:38
<ogra>
vagrantc, cool
10:39
<vagrantc>
ogra: then ubuntu can pull it as part of the sync with debian
10:39
<ogra>
Gadi, all the package builders use fakeroot ... if that woud be broken we'd not have any binaries :)
10:39
<Gadi>
ogra: I realize that
10:40
its gotta be something that crept up on the system
10:40
it worked in the past
10:42* mistik1 thinks Gadi's system is tired of be faked ;P
10:42
<Gadi>
heh - howdy mistik1
10:42
<mistik1>
hey man
10:47
<Gadi>
huh - just learned a new command: ipcs
10:47
:)
10:47
<jammcq>
oh yeah
10:47
<Gadi>
if only I knew how to weild its power
10:48
<jammcq>
years ago, I wrote a command to do that, then someone pointed out one already existed.
10:48
<Gadi>
jammcq: do you know how I can use it to delete elements in the list?
10:48
or is that not a good thing?
10:48staffencasa_ has joined #ltsp
10:48
<jammcq>
list?
10:48
<Gadi>
I've got ipcs owned by users who don't have procs running
10:49
yeah, when I run ipcs, I get a huge list of Semaphore arrays and such
10:49
<jammcq>
ipcrm is what you want I think
10:49
<Gadi>
ah, great!
10:51ws101 has joined #ltsp
10:52tux_440volt has joined #ltsp
10:55
<rjune_>
mistik1 !
10:55
<mistik1>
hey rjune_
10:56
<ws101>
hello, I use ltsp5 on debian testing but cdrom and floppy are mounted twice
10:57
<vagrantc>
ws101: yeah, i was noticing that yesterday
10:57
well, actually with the versions from sid
10:57
<ws101>
example: cdrom and cdrom(1) are the same media
10:58
<vagrantc>
it's because of the hair-brained way we do mounting with ltspfs
10:58
ws101: please file a bug on ltspfs
10:59* vagrantc thinks it's better to have a setuid root binary that creates a directory in /media than to mount in /tmp and then a setuid binary to bind-mount it into /media/FOO
11:00
<vagrantc>
ws101: one of the icons is the /tmp mount, and one of the icons is the mount in /media/USER
11:00
which are effectively the same ...
11:02
i feel like we've gone over this before, and i never remember why we haven't fixed it.
11:02
<Gadi>
ipcrm saves the day
11:03
<jammcq>
yayyyyyy!!!
11:03
3 cheers for ipcrm
11:03
<Gadi>
jammcq: you're my hero.... again
11:03
<sbalneav>
vagrantc: what if a sysadmin makes /tmp a separate filesystem from /?
11:03
<jammcq>
awe shucks
11:03* Gadi stares dreamily to the West
11:03* jammcq gets freaked out a little
11:03
<vagrantc>
sbalneav: i don't understand why we're using /tmp at all.
11:03MRH2 has quit IRC
11:04
<sbalneav>
ok, well, where ELSE would we mount things?
11:04
<ogra>
vagrantc, hmm, i wonder why you see it at all
11:04
you shouldnt see /tmp
11:04staffencasa has quit IRC
11:04
<vagrantc>
sbalneav: uh, /media/USER ?
11:04
<ogra>
vagrantc, security ...
11:05
<vagrantc>
insecurity!
11:05
<ogra>
if you use /media you need to have *all* code in the suid binary
11:05
<vagrantc>
creating files in /tmp with predictable paths is a long-standing known security bug, and should be avoided.
11:05
<ogra>
if you only do the bind mount it only needs to be a oe liner in there
11:05
not with fuse filesystems
11:05
not even root can look into it by design
11:06
<sbalneav>
I was told by people at the time I cam up with that, that having users create dirs in / directly could result in a dir in root owned by the user, which then the user could use as an attack vector for a DOS attack by filling up /
11:06
<ogra>
feel free to change the mount path if yu fee better then
11:06
<vagrantc>
someone can re-direct the mount using a symlink, move the symlink aside, and watch the person write files into a directory they can read.
11:06
<ogra>
no
11:06
try it
11:07
nobody but the owner can touch a fuse fs
11:07
<jammcq>
ah, that's putting your money where your mount is
11:07
:)
11:07
<vagrantc>
yes, and by moving aside the symlink, they can effectively move the location of the mount so that they're not touching the fuse filesystem
11:07
<sbalneav>
Well, by default, unless the user adds options to /etc/fuse.conf
11:07
<ogra>
which user has write access to /etc/fuse.conf ?
11:08
<sbalneav>
only root.
11:08
<ogra>
right
11:08
if your user can be root something else is wrong :)
11:08
<vagrantc>
and every single user has write access to /tmp, and if your mount paths are dependent on files in /tmp, you can get fucked.
11:08
<ogra>
vagrantc, i'm not opposing to change the mount path if it makes you feel better ...
11:08
as i said before
11:09
<sbalneav>
I was just pointing out other users CAN see into fuse mounts, if root's gone and enabled that.
11:09
<vagrantc>
i still don't understand the double-mounting ...
11:09
<ogra>
security
11:09
<vagrantc>
the setuid binary can create /media/USER, and that's all it does.
11:09
which we do anyways.
11:09
<sbalneav>
One supposes that having ltspfs be hal aware would solve all this?
11:09
<jammcq>
if normal users can create things in /media, doesn't that just move the /tmp problem to /media ?
11:10
<ogra>
normal users cant
11:10
<vagrantc>
jammcq: no, because the only way they can create the directory is through a specific binary.
11:10
<jammcq>
ah, but if the binary is setuid, then normal users CAN
11:10
<vagrantc>
which we are smart and write well.
11:10
we're ALREADY doing that part.
11:10
<ogra>
well, thats why we kept it separated :)
11:10
<jammcq>
and if you add a truckload of code to that binary, then it's more vulnerable
11:10
<vagrantc>
we're just doing a bunch of other stuff as root.
11:10
"for security"
11:11
<ogra>
we create/delete the dir and symlink it
11:11
werr
11:11
bind mount it
11:11
<vagrantc>
i'm proposing to reduce the number of things done as root.
11:11
<ws101>
is there a workaround?
11:11
<ogra>
how would that reduce anything
11:11
<sbalneav>
ws101: no, not at this time.
11:11
<ogra>
you need to create/delete the dir and link or mount something
11:12
<sbalneav>
That's what's being discussed.
11:12
<ogra>
apart from that i have never seen that behavor in ubuntu ever
11:12
<vagrantc>
ogra: because instead of creating the directory in /media/USER and bind-mounting is as root, we'll just be creating the directory in /media/USER and mounting directly there.
11:12
<ogra>
vagrantc, try it send a patch :P
11:12
<vagrantc>
well, it's a symptom of a bad design.
11:12
<sbalneav>
vagrantc: I started out with that, and was told by Martin Pitt that that was against Debian Policy.
11:13
<ogra>
in the face of virtual hal devices thats moot anyway
11:13
<vagrantc>
sbalneav: where ?
11:13
<ogra>
as soon as you will have gnome-mount support for ltspfs
11:13
<sbalneav>
vagrantc: I don't know. I don't know debian policy.
11:13
You say it isn't against debian policy, Martin says it is.
11:13
<ogra>
sbalneav, are you sure you dont mean creating symlins in $HOME
11:14
<vagrantc>
i REALLY don't understand why creating a directory + bind mounting is any more policy compliant than simply creating the directory.
11:14
<sbalneav>
I don't know either. I just do what I'm told.
11:14
<jammcq>
:)
11:14charlie_21 has joined #ltsp
11:15
<ogra>
i'm pretty sure its not a policy violation ...
11:15
<vagrantc>
debian policy is a big thing, and i'd at least like a pointer to which part of it is violated before blindly accepting it.
11:15
<ogra>
but pitti was eager to keep them separate for security reasons
11:15* vagrantc stops repeating vagrantc
11:16
<ogra>
just send code :P
11:16
<vagrantc>
ws101: please file a bug
11:16
ok, well, i'd best be off so i can fix this insanity :)
11:16vagrantc has quit IRC
11:17* ogra still doesnt see any insanity :P
11:17
<ogra>
beyond noever having seen that bug in ubuntu
11:17
<sbalneav>
I'm happy to do either. Looks like someone took his ball to go off to play by himself :)
11:18
<ogra>
i wonder if its debians udev or debians gnome-vfs which is broken
11:18
<sbalneav>
No idea
11:18
<ogra>
likely gnome-vfs
11:18
blindly using hardcoded crap for floppy and cdrom devices ...
11:19
i bet as soon as the ltspfs mount would be called cdrom-1 or so it would be fixed
11:19
<sbalneav>
I don't mind doing it vagrants way, but I don't want to get into a situation where vagrant's on once side, and pitti's on the other, and I'm in the middle makeing and reverting changes.
11:19
<ogra>
you wont
11:23maulanar has joined #ltsp
11:23
<ogra>
ws101, if you hack up cdpinger in the client chroot to use cdrom1 or so on the server for mounting i'm sure you wont see it twice
11:25maulanar has quit IRC
11:25doggkatcha has joined #ltsp
11:25
<vpro>
ogra: you have made the debian package for ltsp4.2?
11:29
<ogra>
vpro, i never did anything with 4.2
11:30
<sbalneav>
vpro: I think it's mainly been abandoned. Really, it was just the installer bits. Not sure if it's even maintained by anyone in Debian anymore.
11:30
<ogra>
the maintainer had promised to drop it for sid long ago
11:30
if its still in there thats pretty odd
11:31
vagrant and me asked him
11:31
<sbalneav>
ok, going out for lunch with my folks, be back later.
11:31
<ogra>
sbalneav, will have something to sync for you by then
11:31
(uploading right now)
11:39ogra_ has joined #ltsp
11:40plamengr has left #ltsp
11:43
<Gadi>
ogra, sbalneav: what do you think about an ldm rc.d/ script to sessreg the user (ie to wtmp)?
11:45
<ogra_>
not for gutsy anymore
11:45
but yes, sounds good
11:45
<Gadi>
right
11:45
gutsy++
11:45
<vpro>
I meant the one in feisty
11:45
<Gadi>
;)
11:45
<ogra_>
beta freeze is tomorrow
11:45BadMagic has quit IRC
11:45BadMagic has joined #ltsp
11:46
<vpro>
I was just wondering if it included everything that i need to use local devices such as usb-sticks
11:47tux_440volt has quit IRC
11:48
<Gadi>
I wonder if I freeze the beta I have at home, if I could revive him with medical science sometime in the future...
11:48ws101 has quit IRC
11:48
<Gadi>
he's on his way out... poor old fish...
11:51
<charlie_21>
hi all
11:51
any Idea how much power consuption does the ltsp term 170 have
11:51
<jammcq>
power... like in watts ?
11:51
<charlie_21>
yes
11:51
<Gadi>
2.2 gigawatts
11:51
<jammcq>
hmm, seems liek that's in the specs
11:51
<Gadi>
oh no wait
11:51
<jammcq>
1.21 Gigawatts, gadi
11:52
<Gadi>
oh right
11:52
<charlie_21>
thanks
11:52
<monteslu>
jigawatts
11:52
<Gadi>
doh
11:52
<charlie_21>
I am trying to calculate power consuption
11:52
for these devices
11:53
<jammcq>
hmm, i'm not seeing the wattage specified in the specs
11:53
<charlie_21>
I know
11:53
I looked at the spec sheets
11:53
and nothing
11:53* jammcq makes a quick phone call
11:54ws101 has joined #ltsp
11:55ogra has quit IRC
11:55ogra_ is now known as ogra
11:57
<ws101>
ogra: Debian Bug report #443200
11:58Daggett has quit IRC
11:58
<ogra>
ws101, vagrants thing ... i dont do debian :)
11:58
<jammcq>
charlie_21: there's nothing published about the T-170. The LTSP-1220 is similar, and it's 30watts
11:59
<ogra>
ws101, we dont have that problem in ubuntu apparently
12:00
<charlie_21>
Thanks much
12:00
<ogra>
joebob777as7, https://lists.ubuntu.com/archives/gutsy-changes/2007-September/008665.html if you use ltsp-build-client from that version, your compiz stuff should work
12:01
<ws101>
ogra: I confirm, I didn't have that problem on ubuntu
12:01
<ogra>
should be on the server within the next 2h
12:04
<vpro>
any idea why that with some thin clients the automatic mounting of usb-sticks work but with others it does not. Though I am able to do it manually?
12:05
<ogra>
define "manually"
12:06
<vpro>
mount on thin client and then use ltspfs on server
12:07
<ogra>
any tryces of the device in dmesg of the client ?
12:07
if you plug it in
12:08
<vpro>
the normal
12:08
that it has found a usb-device
12:10
<ogra>
does it list the partitions properly ?
12:10
ltspfs implementation can only work on partitions atm
12:11
<vpro>
yes it does
12:11
<ogra>
intresting
12:11
vpro, which version is that ?
12:11
<vpro>
ltsp5 feisty
12:12
<ogra>
and the user is in the fuse group ?
12:12vagrantc has joined #ltsp
12:12
<ogra>
vagrantc, and ? fixed ?
12:12* ogra hides
12:12
<vpro>
well it should be because I can do with other thin clients
12:13
<ogra>
with the same user ?
12:13
<vpro>
yes
12:13
<ogra>
did you do any tweaks to your lts.conf ?
12:14
<vagrantc>
ogra: ?
12:14
<ogra>
"<vagrantc> ok, well, i'd best be off so i can fix this insanity :)"
12:15
<vagrantc>
ogra: ah, no. breakfast came first
12:16
<vpro>
ogra: sorry, false alarm. The user I used for testing did not belong to fuse, problem with ldap :(
12:16
<ogra>
aha
12:16Guaraldo has quit IRC
12:17Guaraldo has joined #ltsp
12:17Guaraldo has left #ltsp
12:17
<Gadi>
vpro: I may get slapped for this, but I find when using network auth, its easiest to set fusermount to all executable and add a udev rule to make perms on /dev/fuse 666
12:17
(ie circumvent the whole fuse group thing)
12:18
<vpro>
ok :)
12:18
well I can use ldap aswell
12:19
but was it possible to separate ltsp access server from the aplication server and still have these local devices working?
12:20
<Gadi>
you mean have the thin client connect to a server other than the one that boots it?
12:20
<vpro>
yes
12:20
<Gadi>
sure
12:21
use LDM_SERVER
12:21
in lts.conf
12:21
<vpro>
in ltsp4.2
12:21
<Gadi>
ah, 4.2
12:21
yeah, should work, as long as you have lbussd and friends on the server
12:21fernando1 has joined #ltsp
12:21
<ogra>
didnt you say "<vpro> ltsp5 feisty" ?
12:21* ogra is confused
12:22
<vpro>
ogra: different problem :)
12:22
<ogra>
heh, ah
12:23
<vpro>
I'm trying ltsp4.2 to make the boot faster
12:23
<ogra>
try gutsy ;)
12:23
gutsy is on par with 4.2 wrt bootspeed
12:23
<vpro>
I have these small thin clients that goes your by the name Term 1000 PXE
12:24
<ogra>
ah
12:24
yeah, they are not great ... even on gutsy they take their time to boot
12:24
and dont use them with more than 1024x768
12:24
<vpro>
ltsp4.2 boots them quite fast
12:25
<charlie_21>
is it because of ssh
12:25
<ogra>
charlie_21, nope
12:25
<vpro>
that does not affect the boot time?
12:25
<ogra>
ssh has absolutely nothing to do with booting
12:25
<charlie_21>
oh ok
12:25
<vpro>
Thought so
12:25
<charlie_21>
I see
12:26
<vpro>
only for session speed
12:26
<ogra>
ssh is only used after logging in
12:26
right
12:26
<charlie_21>
ok
12:26
<vagrantc>
and can be disabled
12:26
<ogra>
gutsys ldm has an option to switch off encryption for the graphics transport
12:26
while keeping it for the password auth
12:26
so you got XDMCP with safe auth
12:27
<jammcq>
umm, actually, you have no XDMCP at all
12:27
you have X with safe auth
12:27
<ogra>
yeah
12:27
well, speed wise i meant :)
12:27
<jammcq>
and... it's a brilliant idea, btw
12:27
<ogra>
yeah, lets praise Gadi :)
12:27
<jammcq>
for those who want the speed of X with the security of ssh logins
12:28* Gadi loves when he gets to milk a 2-line patch for a whole release cycle
12:28
<Gadi>
:)
12:28
<ogra>
right, there is no usecase for XDMCP anymore as long as you have the control over both sides
12:28
<jammcq>
Gadi: we're all embarrassed that we didn't think of that one
12:29
one of those things that was so obvious we all missed it
12:29
<Gadi>
15 mns of fame, jammcq, 16 mins of fame
12:29
or something
12:29
:)
12:29
<jammcq>
hey, enjoy it while you can
12:29
bask in the glory of it
12:29
<Gadi>
I know - doesnt come around too often
12:32
<elisboa>
I use ssh -X, with no display manager
12:32
But it's a different case, since we don't need to login our users
12:33* vagrantc tosses in a bit more praise for Gadi
12:33
<ogra>
sbalneav, which revisions are the ones i need to merge for ltspfs ? 11 and 12 ?
12:37* ogra uploads with these two fixes added
12:38
<ws101>
vagrantc: Debian Bug report #443200
12:38
<vagrantc>
ws101: yeah, got it.
12:38
ws101: good thing i know exactly what the problem is, or i'd have to email you back and ask for more information :P
12:38
<ogra>
vagrantc, its gnome-vfs i'm pretty sure
12:39
<vagrantc>
ogra: well, the bug is that we're mounting it twice. ubuntu has patches that probably hide the bug.
12:39
<ogra>
we patched out the hardcoded grepping for mounts with the mane cdrom and floppy a while ago in ubuntu
12:39
vagrantc, no
12:39
its a gnome-vfs bug, really
12:40
s/mane/name/
12:40* vagrantc still thinks mounting it twice is a bug.
12:40
<ogra>
you might call it two bugs thought
12:41
well, find a sane and secure way to do it with one mount :) or even better implement virtual-hal-devices :)
12:41
then you wont need to care :)
12:42charlie_21 has quit IRC
12:42
<ws101>
i see an error on dmesg
12:47
<sbalneav>
Back from lunch
12:48
<ws101>
not in dmesg, sorry, in the first console:
12:49
error: /media/ws101 is not mounted
12:50
error: can't bindmounter under /media/ws101/floppy0: alredy mounted
12:51K_O-Gnom has joined #ltsp
12:51
<sbalneav>
vagrantc: I'm happy, btw, to do it your way, mounting directly in media. Like I say, when I coded that up originally, I was told it was bad mojo. If that's changed, I'm cool.
12:51
<vagrantc>
i'd just leave too hear something more specific than "bad mojo"
12:52
s,leave,like,
12:53
<ogra>
ws101, you are not logged in mulitple times i suppose
12:53
<ws101>
no
12:53
<sbalneav>
Well, pitti'd be the guy to talk to. Mine and ogra's original idea was to patch pmount, and do it there. That was a no-go as well :)
12:54
Yes, 11 and 12, btw
12:56
<ogra>
sbalneav, pmount is long gone
12:56
we dont use it since feisty anymore
12:56
<ws101>
there is also "fusermount failed to access mount point /tmp/.ws101-ltspfs/floppy0: permission denied
12:58
"
12:59Clown-MG has quit IRC
13:00
<vagrantc>
ws101: is your user in the fuse group ?
13:01
<ws101>
yes, i correct mount hard disk and usb pen, cdrom and floppy twice :)
13:06ccherret1 has joined #ltsp
13:12Amaranth has quit IRC
13:13ws101 has quit IRC
13:13Amaranth has joined #ltsp
13:13ws101 has joined #ltsp
13:13rcy has quit IRC
13:17ccherrett has quit IRC
13:18cliebow_ has quit IRC
13:31Amaranth has quit IRC
13:32bricode has joined #ltsp
13:37rcy has joined #ltsp
13:40Q-FUNK has joined #ltsp
13:45
<bricode>
With gutsy do you need to do an ltsp-update-image after an ltsp-update-sshkeys?
13:45
<sbalneav>
bricode: yes
13:46
<bricode>
I suppose that's why it doesn't work then.
13:46
Are there any plans to be able to mount the squashfs image to make changes easily?
13:48Clown-MG has joined #ltsp
13:49* Gadi is curious, as well, whether a mounted squashfs rootfs can be updated without requiring all thin clients to reboot...
13:49
<sbalneav>
No
13:49
<vagrantc>
gah. seems like the new X in sid isn't working today.
13:50
<Blinny>
Is there an allowance already in place for periodically clearing /tmp out?
13:50
<bricode>
Gadi: That sounds like it's asking for trouble.
13:50
Gadi: Had a question about that Kiosk LTSP plugin. Does it run stuff server or client side?
13:50
<Gadi>
bricode: those who do not seek trouble may never find it...
13:51
or something
13:51
<sbalneav>
No, you'd have to reboot anyway, and I don't see a need to make changes directly to the squashfs. Why would you want to?
13:51
<Gadi>
the answer to your second qu is client side
13:51
<bricode>
Gadi: Adventure, excitement, a Jedi craves not these things.
13:51
<vagrantc>
Gadi: well, you don't have to reboot, but the changes won't take effect
13:51
<sbalneav>
I mean, the way it is now, you can easily switch back and forth between nbd and nfs
13:51
<vagrantc>
unless nbd-server is stupid
13:51
which, come to think of it, it might be
13:52
<sbalneav>
if we start making changes to the squashfs directly then they'll get out-of-sync
13:52
<bricode>
sbalneav: That's true, but if you're only doing a small update to the squashfs, it takes a while to recompress.
13:52
<sbalneav>
what constitutes "a while"?
13:52
<loather-work>
harrr, ye scurvy sea dogs!
13:52
<bricode>
sbalneav: Ideally it would just mount it and do some quick sync function instead of an entire recompress.
13:52
<sbalneav>
In my dual core box, it's 45 seconds.
13:52
<vagrantc>
with sqaushfs, it's a read-only filesystem, so to make changes to it we have to re-create it.
13:53
<bricode>
sbalneav: I haven't timed it on the Geode LX box I'm working a proof of concept on, but it's *much* longer than that.
13:53
<klausade>
I have a client that uses floppy etherboot to boot, and something keeps reading the floppy every few minutes. I commented out the fd line in /opt/ltsp/i386/etc/udev/ltspfsd.rules, which made the annoying automatic reading of floppy stop. but it seems to have broken using the cdrom also. What is the correct way of disabling this constant reading of the floppy?
13:53
<sbalneav>
Yeah, well, if you want to serve off the smallest box you can find, then you pay the price :)
13:54
<bricode>
vagrantc: Well crap. I forgot that was the case.
13:54
<sbalneav>
You could BUILD the image elsewhere, on a fast box.
13:54
<bricode>
sbalneav: True.
13:55
<sbalneav>
klausade: It'll be Nautilus, doing that.
13:55
<vagrantc>
bricode: use NFS for development, and then switch to squashfs once you have a working setup.
13:56
<bricode>
vagrantc: Humm. Good point.
13:56
<sbalneav>
yep
13:56* bricode bangs head
13:56
<highvoltage>
moo.
13:57ccherret1 is now known as ccherrett
13:57
<sbalneav>
Foo!
13:58
<ogra>
loather-work, arrr, i totally missed its talklikeapirateday
13:58
yo, ho ho
13:58
<Gadi>
mmm..... rum....
13:59
<loather-work>
avast! how could ye be forgettin' talk like a pirate day!? HARRR!
13:59
<ogra>
http://www.talklikeapirate.com/ !
13:59
<Q-FUNK>
greetings from Istanbul
13:59* vagrantc yearns for the return of the purple corsair
13:59
<loather-work>
HARR, Greetings from the seven seas!
14:00
<highvoltage>
sbalneav: how are you doing?
14:02
<klausade>
sbalneav: using kde.
14:06jammcq has quit IRC
14:06
<sbalneav>
klausade: Then it'll be konq
14:07
<vagrantc>
ogra: "asoundconf set-pulseaudio" is invalid syntax in debian
14:08
<ogra>
oh
14:08
i wasnt aware
14:08
thats bad
14:08
<vagrantc>
yeah, well.
14:08
<ogra>
i know pitti wrote it ages ago
14:08
i thought it had made its way into debian
14:09
<vagrantc>
i can document the method for setting it as the default card globally, mentioning the drawback, and wait for a better option
14:10
<ogra>
ask the alsa guys to take our patch :)
14:11
<vagrantc>
well, sure. but in the meantime, at least i know how to get something almost as cool in debian :)
14:11Blinny has quit IRC
14:13
<klausade>
sbalneav: thanks for the pointer. I'll have a look at what makes konq do this.
14:13
<ogra>
sbalneav, ?? how can konq or nautilus trigger seeks on the client ?
14:15
<klausade>
ogra: beats me aswell, but when sbalneav says to look at konq, I do as he says :-)
14:15
<sbalneav>
ogra: both konq and nautilus run the "stat" system call periodically
14:15
<ogra>
well, there is no technical reason :)
14:15
sbalneav, but that would require the mount to be there
14:16K_O-Gnom has quit IRC
14:16
<sbalneav>
Remember: if you've got a floppy, the mount's ALWAYS there, because we can't detect media for floppies
14:16
<ogra>
but not cdroms
14:16
<sbalneav>
a floppy is ALWAYS "mounted".
14:16
right, not cdroms.
14:16
because we use cdpinger
14:17
<vagrantc>
there's no way to make a correlary cdpinger ?
14:17
<sbalneav>
floppy pinger?
14:17
<vagrantc>
yeah
14:17
<ogra>
floppyd ?
14:17
<sbalneav>
no. The only way to try to detect media in a floppy is... to try reading a floppy.
14:17
the media detection systems in floppies are notoriously broken.
14:17
<vagrantc>
so we'd shift the floppy polling client-side, which basically gains us nothing
14:17
<sbalneav>
right.
14:18
you'll see that with hal as well
14:18
if you leave a floppy in the drive, some floppys will "seek" every once in a while with hal
14:18
I've got one here that does it.
14:19
Floppies, in short, are kind of broken. It's a really old standard.
14:19
<loather-work>
do people still even use floppies?
14:20
<sbalneav>
klausade does, obviously :)
14:20
Lots of people do.
14:20
<loather-work>
i mean, there's no bulk DMA transfer, no MMIO range for them; it's all interrupt-driven and demands all of the CPU's attention.
14:20
i wish they would just go away :(
14:21
come to think of it, i don't think a single machine in this office has a floppy
14:21
<sbalneav>
Not everybody lives in North America, you know. Lots of people in developing nations cant afford to go buy themselves a 2 gig memory stick :)
14:21
<klausade>
sbalneav: I don't "use" the floppy, it's in the drive because the client doesn't have a pxe-card, so I ise etherboot on the floppy.
14:22
<sbalneav>
Well, isn't that using it?
14:22
You use it to boot.
14:22
<elisboa>
klausade: here, we use the floppies to store users data
14:22
<klausade>
sbalneav: sort of.
14:22
<highvoltage>
there's some local schools here that still use floppies too
14:22
<klausade>
elisboa: that is what I mean, I don't use them to store users data.
14:23bricode has quit IRC
14:23
<highvoltage>
although they are still on ltsp 4.2 and mtoolsfm and floppyd :/
14:23Amaranth has joined #ltsp
14:23
<elisboa>
klausade: yes, you use them just to boot
14:23
<sbalneav>
loather-work: See? Lots of floppies still out there :)
14:23
<loather-work>
anyhow, i've got my fedora chroot all pared down
14:24
sbalneav: that saddens me :(
14:24
<ogra>
cool
14:24
<loather-work>
ogra: as soon as i can find mknbi, i'll try and boot this terminal
14:24
<ogra>
fedora should have it ...
14:24
dont you have PXE clients ?
14:25
<loather-work>
well yeah, but pxe != etherboot :(
14:25
oh. you mean the terminals... no, they're all etherboot.
14:25
<ogra>
use syslinux/pxelinux
14:25
ah, k
14:25
then you need mknbi
14:26
or better mkelf-linux
14:26
<loather-work>
I would have bought pxe terminals, but the guys who did the setup here got etherboot terminals for whatever reason
14:26
<ogra>
well, we support both oout of the box in ltsp5
14:28
<klausade>
funny thing is that I have a huge box full of pxe-cards here, but I just dont have the time to dig out the clints and replace the nics.
14:28* klausade don't like getting his knees dirty.
14:29
<loather-work>
ogra: yea, another reason why i need to get mknbi :) so everyone else has support too!
14:29
<rjune_>
klausade: I could make a crack about that, but sometimes it's just too easy
14:30
<loather-work>
yeah, the 'down on your knees' jokes are abundant :)
14:30ws101 has left #ltsp
14:34richguit has joined #ltsp
14:35vagrantc has quit IRC
14:35Amaranth has quit IRC
14:40
<klausade>
soo, I noticed that running df -h on the server triggered a read on the floppy in the client, and this: ltspfs 62M 36K 62M 1% /tmp/.foobar-ltspfs/floppy0
14:44Gadi has left #ltsp
14:48mikkel has quit IRC
14:50Q-FUNK has quit IRC
14:52MRH2 has joined #ltsp
14:53
<MRH2>
hi just wondering on the server side - any issues with mounting / with noatime
14:54
ext3
14:55Guaraldo has joined #ltsp
14:55Guaraldo has left #ltsp
14:55
<sbalneav>
Well, other than the fact you may break a bunch of things which depend on atime, no
14:56
<MRH2>
what sort of things would depend on atime?
14:56
<sbalneav>
I mean, if you're doing a flash-based filesystem or something embedded like that, it's a good idea
14:56
backups
14:56
some modes for the find command.
14:56
etc.
14:57
I guess the question is: why do you WANT to disable it?
14:57
What are you hoping to accomplish by disabling it.
14:57Guaraldo has joined #ltsp
14:57
<MRH2>
well i thought i'd see if it would make any difference at all to performance
14:58
<sbalneav>
On a normal system with regular hard drives, it won't make any difference at all.
14:58
but, try it and see. You won't hurt much by trying it for a short time.
14:59
<MRH2>
i'll give it a go - just don't want to break anything
14:59
<sbalneav>
Well, that's always a possibility.
14:59
<MRH2>
(important)
14:59
<sbalneav>
I'm assuming you're going to do this on a test system, right?
14:59
Not a live system?
15:00
<MRH2>
course not
15:00
<---living on the edge
15:00
<sbalneav>
So, you're going to do it on a live system?
15:01
<MRH2>
yeah that's why i thought i would ask if wnyone else does this
15:01Guaraldo has left #ltsp
15:02bricode has joined #ltsp
15:02
<sbalneav>
Well, I'd never make a change like that on a live system.
15:02
I'd test it first somewhere.
15:03highvolt1ge has joined #ltsp
15:04Guaraldo has joined #ltsp
15:10elisboa is now known as ELISBOA-OFF
15:10ELISBOA-OFF is now known as elisboa-off
15:10
<MRH2>
ok seeing as you have ur doubts i'll give it a test somewhere
15:10
thanx
15:16highvoltage has quit IRC
15:21Clown-MG has quit IRC
15:35MRH2 has quit IRC
15:49tiagovaz has quit IRC
16:12Guaraldo has left #ltsp
16:15timday has joined #ltsp
16:29
<sbalneav>
Headin' home. Be on tonight.
16:41richguit has left #ltsp
16:41bobby_C has quit IRC
16:42J45p3r has joined #ltsp
16:44bricode has quit IRC
16:46mistik1 has quit IRC
16:48timday has quit IRC
16:49timday has joined #ltsp
16:51mistik1 has joined #ltsp
17:01bricode has joined #ltsp
17:02bricode has quit IRC
17:03jammcq has joined #ltsp
17:04
<jammcq>
hola
17:04bricode has joined #ltsp
17:20
<timday>
\quit
17:21timday has quit IRC
17:47MagicStorm has quit IRC
17:59captain_magnus_Z has quit IRC
18:00captain_magnus has joined #ltsp
18:21bricode has quit IRC
18:33bricode has joined #ltsp
18:44
<cliebow>
arrrrrgh:
18:46
<loather-work>
aye aye, ye scurvy scoundrel!
18:48vagrantc has joined #ltsp
18:58staffencasa_ has quit IRC
19:15MagicStorm has joined #ltsp
19:17vagrantc_ has joined #ltsp
19:31vagrantc has quit IRC
19:33drwahl has joined #ltsp
19:33
<drwahl>
can ltsp thin clients render 3D?
19:37
<cliebow>
it is not athin clients forte..
19:38
<drwahl>
from what i can see, it looks like it is possible though... how would one go about getting 3D acceleration on the thin client?
19:40
<cliebow>
better catch the ubermenschen on that...sbalneav...et al
19:40
<vagrantc_>
if it's LTSP5, not really any different than how you'd configure it for a desktop machine.
19:42
<drwahl>
hmm, i'm not sure which version it is.... I'm running debian Sid and i got ltsp-server from debians repositories
19:42
ya, looks like 5.0.27
19:44
"not really different than... a desktop machine" i assume you mean a standard desktop. if that is the case, then to my understanding, i would go to the thin client, download the latest driver, install them, edit xorg, and then restart X. correct?
19:45
<cliebow>
drwahl, vagrantc_ is your man..he is the debian maintainer
19:45
<drwahl>
ok, here is another question. i try to log in as root and it seems to be wrong username/password. is this a security feature i can get past some how?
19:55
<vagrantc_>
drwahl: can you ssh in as root ?
19:55
<drwahl>
ya
19:56
<vagrantc_>
drwahl: are you talking about logging in through the GUI interface ?
19:56
<drwahl>
i hit Ctrl+Alt+F1 to bring up a command line
19:56
and it prompts to login
19:56
and i try root, but it doesnt seem to like it
19:57
<vagrantc_>
yeah, there's no root password by default
19:57
<drwahl>
ah, ok
19:57
<vagrantc_>
if you want a shell, you can set SCREEN_07=ldm and SCREEN_08=shell
19:57
and go to tty8 for a shell
19:57
or, if you must, set a password: chroot /opt/ltsp/i386 passwd
19:58
but that password is available to the network via NFS, so don't use an important one
19:58
<drwahl>
ah, ok
19:58
that makes sense
19:58
maybe i will just stick with the more secure practices
19:58
<vagrantc_>
heh
19:59
actually, root's password is disabled by default ... which is different than no root password
19:59
:)
19:59
finally figured out how to get remote sound working without too much tweaking in the past few days :)
20:00
<drwahl>
really? i have been working on that myself
20:00
how did you go about getting it working?
20:00
<vagrantc_>
drwahl: http://bugs.debian.org/src:ltsp
20:01
drwahl: there's a bug report in sound, near the end i posted how to get it working
20:01
it's an alsa -> pulseaudio configuration thing
20:01
<drwahl>
interesting...
20:01
ok, i have another question. can i have the clients automatically mount their local harddrives?
20:10
<vagrantc_>
if you install ltspfs on the server, and add your user(s) to the fuse group, it will happen automatically.
20:12vagrantc_ has quit IRC
20:53drwahl has quit IRC
21:07J45p3r has quit IRC
21:09chris_punches has joined #ltsp
21:29Shingoshi has quit IRC
21:29sbalneav has quit IRC
21:29Lumiere has quit IRC
21:29str4nd has quit IRC
21:29vlt has quit IRC
21:29ufo_ has quit IRC
21:29Shingoshi has joined #ltsp
21:29sbalneav has joined #ltsp
21:29vlt has joined #ltsp
21:29ufo_ has joined #ltsp
21:29Lumiere has joined #ltsp
21:29str4nd has joined #ltsp
21:42moquist_ has joined #ltsp
21:48parsley has quit IRC
21:55
<sbalneav>
Evening all
21:57
<jammcq>
Scotty !!!!!!!!!!!!!!!!!!!!!!!!!!!
21:58Guaraldo has joined #ltsp
21:58Guaraldo has left #ltsp
21:58moquist has quit IRC
21:58
<sbalneav>
Evening jammcq!
21:59
<jammcq>
how's it goin, eh?
21:59
<sbalneav>
Goin! Testing, testing testing
21:59
We'll see if the patches I pushed up to ogra made it to the mirrors
21:59* sbalneav crosses fingers
22:02
<jammcq>
hey, how much snow you got up there?
22:02
<sbalneav>
None, yet.
22:03
Whoohoo
22:03
Ogra's a good man. Now I just gotta wait for the 120 megs of updates to download.
22:03
<jammcq>
:)
22:03
<sbalneav>
argh. Gotta download a daily build too.
22:07moquist_ has quit IRC
22:15moquist has joined #ltsp
22:16chris_punches has quit IRC
22:23Q-FUNK has joined #ltsp
22:24
<sbalneav>
Hey Q-FUNK
22:27
<Q-FUNK>
hey Scott!
22:28elisboa-off is now known as elisboa
22:29
<sbalneav>
Q-FUNK: Anyone you know of had a chance to look at Bug #140051?
22:34
<Q-FUNK>
both bryce and I did, but we haven't figured out what might cause it. Jordan is also looking into it now.
22:34joris_ has joined #ltsp
22:35
<sbalneav>
Wish I knew more about debugging X
22:36
If I could even get an idea as to which ROUTINE it's cacking in, I could at least read the code, and play around a bit.
22:37
Time for my daily chroot rebuild :)
22:47joris has quit IRC
22:51joebaker has joined #ltsp
22:52elisboa is now known as elisboa-off
23:07Q-FUNK has quit IRC
23:26chris_punches has joined #ltsp
23:37MagicStorm has quit IRC
23:47spectra has quit IRC
23:54chris_punches has quit IRC
23:54
<joebob777as7>
sbalneav, have any issues with black screen after login?