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


Channel log from 24 April 2008   (all times are UTC)

00:21vagrantc has quit IRC
00:35asac_ has joined #ltsp
00:37subir has joined #ltsp
00:41tiger74 has joined #ltsp
00:46asac has quit IRC
00:46asac_ is now known as asac
01:07deavid has joined #ltsp
01:12empiric has joined #ltsp
01:15
<empiric>
any one did heartbeat with ltsp?
01:15
for failover with another ltsp server
01:16
<daduke>
empiric: not yet, but we might at some point. How far have you gotten?
01:16
<empiric>
well doing HA on my ltsp server
01:17
also need drbd to replicate my /home partition
01:18
<daduke>
empiric: what's your cluster resource(s)? IP address? anything else?
01:18
<empiric>
well am doing it internally
01:18
will share with you
01:18
am doing setup right now
01:18
if any help is provided it would be great
01:19
<daduke>
empiric: I've built a debian/drdb/heartbeat setup, but for xen, not for ltsp
01:19
<empiric>
letok
01:19
can you share
01:20
<daduke>
empiric: without having thought about it for more than 2 min I should think that the IP address is the only cluster resource you'd need.
01:20
<empiric>
hmm think soo
01:21
mainly it is ip
01:21
<daduke>
empiric: do you have /home locally on the server?
01:22
<empiric>
yes
01:22
its where all ltsp users stores files
01:23
<daduke>
if you don't need seamless takeover drdb + ext3 should be fine. what distro you on?
01:31
<empiric>
kubuntu
01:31
its ext3
01:31
wellseam less is goodd
01:34
<daduke>
for seamless you'd need a real cluster fs. GFS or OCFS2. I've spent a fair amount of time trying to get OCFS running on an etch heartbeat cluster. the problem is: OCFS uses disk based heartbeating (their own). but to avoid split brain hell, OCFS heartbeating has be integrated into HA heartbeating. Only SLES has managed to do that. I've tried porting their patches to debian, but it didn't work in the end - too many changes in too many places
01:43Pascal_1 has joined #ltsp
02:01
<Pascal_1>
hello
02:02
i'm trying to make eepc ltsp client, does anybody try this? my actually problem is a "kernell panic" can't open /tmp/net-eth0.conf
02:04subir has quit IRC
02:14BGomes has quit IRC
02:36Q-FUNK has joined #ltsp
02:46
<empiric>
daduke then how i do drbd + ext3
02:48
<daduke>
!pastebot
02:48
<ltspbot>
daduke: "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.
02:51
<daduke>
empiric: I haven't written the real docs yet, but my note are here:
02:51
<ltsppbot>
"daduke" pasted "preliminary cluster docs" (77 lines) at http://pastebot.ltsp.org/515
02:52
<empiric>
thanks
02:52
will see taht
02:55
<daduke>
empiric: it will give you drbd'd filesystems that you can mount on one of the cluster nodes at a time.
02:58
<empiric>
hmmm
02:58
ok
02:58
it means i can not use both my ltsp servers at the same time
02:59
one should be online and passive
03:00
<daduke>
empiric: if you want that, I think you'll need something like NFS homes. that's what we do here.
03:00
<empiric>
means a 3 third machine which act as a NFS
03:02
<daduke>
empiric: otherwise, you might consider something like our xen cluster: domUs that can migrate between the cluster nodes. make one of them your NFS server.
03:03subir has joined #ltsp
03:11ogra has quit IRC
03:12ogra has joined #ltsp
03:15toscalix has joined #ltsp
03:18tiger74 has left #ltsp
03:30toscalix_ has joined #ltsp
03:30toscalix_ has joined #ltsp
03:41toscalix_ has quit IRC
03:43toscalix has quit IRC
03:55exodos has joined #ltsp
04:11mikkel has joined #ltsp
04:33daya has joined #ltsp
04:49Pascal_1 has quit IRC
04:49Q-FUNK has quit IRC
05:12gordslater has joined #ltsp
05:14gordslater has left #ltsp
05:25Pascal_1 has joined #ltsp
05:36
<ogra>
oh how i love that pre release excitement :)
05:46Q-FUNK has joined #ltsp
05:51basanta has joined #ltsp
05:52
<ogra>
laga, did you see the ubuntu mediacenter thread on -devel-discuss ? someone is heavily pushing for elisa there
05:53basanta has quit IRC
06:13generic has joined #ltsp
06:13
<generic>
hi all
06:13
i want
06:13
to do my ltsp on HA
06:13
i ahve 2 ltsp servers
06:14
what service should be in HA
06:15daya has quit IRC
06:16empiric has quit IRC
06:19
<ogra>
generic, like that ? https://wiki.edubuntu.org/EdubuntuDHCPload-balancingFailover
06:21TelnetManta has quit IRC
06:25
<generic>
ogra i need to have dhcp in ha
06:26
right
06:26
right now ma testing heart beat on real=192.168.1.1:80 gate
06:26
port 80 which is web
06:28captain_magnus has quit IRC
06:28
<generic>
well
06:28
for lTSp to be on HA
06:28
what i need to do
06:28
<ogra>
well, for an ubuntu r debian setup with ltsp5 the above should suffice, many people use it that way
06:29
<generic>
the link is not opening
06:31
<ogra>
the wiki is a bit loaded atm, we'Re about to release ubuntu 8.04 within the next hours
06:31
sorry for that
06:31
it will ooad eventually, give it some time
06:32captain_magnus has joined #ltsp
06:37subir has quit IRC
06:42hersonls has joined #ltsp
06:49
<laga>
ogra: i thought that the ubuntu media centre edition is already using elisa?!
06:49
ogra: i never heard good things about it.. i might read the thread later, right now it would make my head explode ;)
06:50
<ogra>
well, they want to have it included as default install option
06:50
thought you wanted to know about that
06:51
<laga>
ogra: AFAIK, elisa is backed by fluendo. if you want to play some codecs, you've got to pay
06:51
<ogra>
yeah
06:51
for some at least
06:52
<laga>
i don't think it can do TV either
06:54
no, it can't. well, i can't prevent them from using silly software ;)
06:55
<ogra>
well, you could chime in with the working mythbuntu solution :)
06:55
and get more hackers
06:57
<laga>
mythbuntu is not particularly user friendly (unfortunately), and ubuntu seems to be about that
06:57
oops.
06:58
i meant to say "mythtv". mythbuntu certainly tries very hard to be userfriendly :)
06:58
<ogra>
make it userfriendly :) get the mediacenter guys involved to help :)
06:58hersonls has quit IRC
06:58* ogra will test mythbuntu diskless next week if he has the release stuff off the sholders
06:59
<ogra>
finally time to instal my TV card and have a silent VCR in my living room :)
06:59
<laga>
great :)
07:00
the documentation for mythbuntu diskless is almost finished. but i guess you won't need it :)
07:00
<ogra>
well, lest see what bugs i find :)
07:00
<laga>
scary.
07:00
;)
07:00Guaraldo has joined #ltsp
07:04hersonls has joined #ltsp
07:06Blinny has joined #ltsp
07:06hersonls has quit IRC
07:07viking-ice has joined #ltsp
07:07hersonls has joined #ltsp
07:07Q-FUNK has quit IRC
07:07Q-FUN1 has joined #ltsp
07:07Q-FUN1 is now known as Q-FUNK
07:08Pascal_2 has joined #ltsp
07:08Pascal_1 has quit IRC
07:08hersonls has quit IRC
07:09* ogra wonders if #ubuntu-release-party will pass the 700 ppl mark before the release announcement
07:09hersonls has joined #ltsp
07:09* laga wonders if parties on IRC aren't a little bit odd
07:09
<ogra>
nah
07:10
share the pre-release excitement over the world :)
07:10
have you seen that channel ?
07:10
its a madhouse :)
07:10
<laga>
hm
07:10
<ogra>
nearly 700ppl waiting in awe for a mail :)
07:10* laga ponders trolling ;)
07:11
<ogra>
they will kick you
07:11
<laga>
yeah
07:11
<ogra>
over a certain threshold at least :)
07:11
<laga>
i'm already running the latest & greatest. so i've had enough excitement ;)
07:11
<ogra>
woah, #ubuntu nears the 1700 ppl mark
07:13
<laga>
dear binary printer driver, please stop using all my spare cpu cycles, therefore making my laptop run warm which causes some uncomfortableness on my laptop. love, laga
07:14makghosh has joined #ltsp
07:19TelnetManta has joined #ltsp
07:22
<Q-FUNK>
hehe
07:22
it's download time
07:22
<laga>
yup
07:23
<ogra>
yeah, there we go
07:24
<laga>
gentlemen, start your torrent clients
07:25
<ogra>
ha, now #ubuntu-release-party has 705
07:26
damned i would have won 5€ if it hit that 10 mins ago
07:31
<tarzeau>
it's really funny
07:34
<Q-FUNK>
heh
07:34
<tarzeau>
do you remember the billennium unix second party?
07:35slidesinger has joined #ltsp
07:35
<tarzeau>
i still got the logs, it was soooo uber-fun
07:35slidesinger has quit IRC
07:35slidesinger has joined #ltsp
07:45daya has joined #ltsp
07:50makghosh has quit IRC
07:51otavio has joined #ltsp
07:53makghosh has joined #ltsp
07:55tux_440volt has joined #ltsp
08:16daya has quit IRC
08:19vagrantc has joined #ltsp
08:22K_O-Gnom has joined #ltsp
08:22Gadi has joined #ltsp
08:26K_O-Gnom has quit IRC
08:30
<warren>
Q-FUNK: question from a former OLPC engineer, have you considered openfirmware instead of coreboot?
08:30
Q-FUNK: they found it a lot easier to work with
08:37
<Q-FUNK>
warren: of course they did. they are engineers. they want a hackable boot prompt. :)
08:37
warren: dilinger is considering porting it to the thincan. no idea if he ever will.
08:38* ogra dances around warren with a release in his hands ...
08:38
<ogra>
... now you
08:38
:)
08:38
<Q-FUNK>
:D
08:39
ogra: we're cheating. they're only a few days away from it. :)
08:39
warren: when are you releasing again? tomorrow?
08:39
<ogra>
heh
08:39
<Q-FUNK>
I remember you said this week, but not when exactly
08:40
<ogra>
Q-FUNK, they postponed to may i heard
08:40
rumors at least :)
08:42
<warren>
Q-FUNK: we delayed our release until may 13th I think
08:42Bengoa has joined #ltsp
08:42
<warren>
Q-FUNK: more stabilization
08:42
Q-FUNK: we're using crazy newer versions of stuff than Ubuntu across the board
08:43
Q-FUNK: I'm disturbed that mkelfImage requires hard coding boot prompt options into the elf image.
08:43
this completely defeats my desire to decouple update-kernels from boot options
08:43
<ogra>
warren, but you dont have a 5 year support setup for that release, we were pertty conservative because of that
08:43
<warren>
I'm seeing if I can possibly chain load anything to emulate PXE instead.
08:44
ogra: nod.
08:45
<Q-FUNK>
ogra: ...not that it prevented me from shooting myself in the foot with amd->geode.
08:45
<ogra>
Q-FUNK, but thats something we can get sorted
08:46
<Blinny>
Who do I need to bribe to get a module updated to the stable release in ubuntu-kernel? Hardy didn't include the stable version.
08:47
<ogra>
did you plan to come to prague ?
08:47
Blinny, #ubuntu-kernel .... we have a mechanism to updae modules
08:48
<warren>
sigh, chain loading wont work
08:48
<Blinny>
ogra: Cool, thanks.
08:58ray_ has joined #ltsp
08:58
<ray_>
hi, I had some questions about ltsp ... and I was wondering if someone may be able to answer them ...
08:59K_O-Gnom has joined #ltsp
09:00
<pscheie>
ray_, ask away
09:00
<ray_>
basically, I've been trying to set up a set of client machines to use as disk based thin clients ... and running a barebones version of linux on them, and then starting up rdesktop to connect to a windows server ...
09:01
it works great ... but with one problem ... i can't easily connect local devices such as floppy disks, and USB drives etc to the server ... since rdesktop has already been invoked ...
09:02
i guess my question, as it relates to ltsp, is ... can ltsp do this sort of thing with rdesktop, or will that be a problem there as well ...
09:02
<Gadi>
ray_: on ur disk-based clients, do: rdesktop -r disk:drives=/media ....
09:03
then, in Windows go to My Computer "drives on <hostname>"
09:03
<Q-FUNK>
ogra: ah, thanks for reminding me. May... 25, was it?
09:04
<ray_>
yea ... that was my goal ... only problem is this type of situation ... :user walkes up to the terminal and logs into the windows terminal ... and then puts their floppy disk, or flash drive in the system ... rdesktop has already been started, so i don't know any information about the drives they put in ...
09:05toscalix has joined #ltsp
09:05
<ray_>
i was messing around with ivman to automatically mount the usb disk ... but the only problem is ... sometimes there may not be a disk, or it may be named with a different name depending on the partitions / drive brand etc ..
09:05
<Gadi>
ray_: sure it does
09:05makghosh has quit IRC
09:05
<Gadi>
it will appear magically in a folder under rives
09:05
*drives
09:05
it just appears in windows as a folder and not a drive
09:06
ie, redirect the whole /media directory
09:06
the folders are created by Linux
09:06
<ray_>
yea ... i got that part ... but only after mounting the floppy disk manually at the command prompt ... and THEN running the command for rdesktop ...
09:06
oh ...
09:06
you can do that? ...
09:06
<Gadi>
no, launch rdesktop fullscreen from within xfce or some such
09:07
something with a volume manager
09:07
the volume manager is part of the desktop
09:07
these days
09:07
<Q-FUNK>
ogra: can you think of any way to convince vorlon to allow 2.9.0 into 8.04.1? it's a new upstream, but it closes bugs and relieves the tension with a key canonical partner (koolu).
09:07
<ray_>
i was hoping to go as bare bones as possible ... with no way to break out to a windows manager ....
09:07
<Gadi>
well, you can, but it gets trickier with a lot more hacking
09:07
there's no "canned" way
09:08
but, XFCE is very light
09:08
<ray_>
it seems that installing ivman mostly did the trick ...
09:08makghosh has joined #ltsp
09:09
<ray_>
so ... with the rdesktop command ... would you just say something like "rdesktop -r disk:MEDIA=/media
09:09
"
09:09
<warren>
edubuntu.org wiki is really slow
09:09
<ray_>
i don't know if that would take care of the floppy mount ...
09:10
<Q-FUNK>
anyone has any idea for liquidating our old stock of DBE60 as Etherboot terminals? they're already priced dirt cheap, but people seem to shun them becuase they are built on an older Geode variant.
09:10
<ray_>
yea ... the ubuntu wiki is really slow (or not responding at all) i noticed ...
09:11
<Gadi>
ray_: Linux takes care of all mounting
09:12
but only if you have a volume manager
09:12
<daduke>
Q-FUNK: unfortunately XGA isn't enough for our needs
09:13
<ray_>
yea ... i'll give that a try .. i know you can run nautilus in the background to take care of automounts ... i messed around with that a little bit in fluxbox ..
09:14deavid has quit IRC
09:14
<ray_>
gadi ... do you know about the delayed writes, or how to "unmount" the vol. safely? ...
09:14
<Gadi>
not nautilus - gnome-volume-manager
09:14
ah, well for that we came up with ltspfs
09:15
you need some sort of automounter
09:15
<ray_>
maybe that was it .. i was using nautilus somewhere in that equation with fluxbox : )
09:15
i thought i had read somewhere that fuse was being used ...
09:16
<Q-FUNK>
I was thinking that someone who is really strapped on cash and has access to a bunch of 15" VGA to setup a class of LTSP terminals could use them, but there doesn't seem to be too much interest
09:16
<ray_>
so it would just idle out at 2 seconds ... make the writes to the disk, and let you take out the disk without much worry ...
09:17Guarald1 has joined #ltsp
09:17
<Gadi>
right
09:18
<ray_>
is there a way to take advantage of that ... without going the full ltsp route? ...
09:19
<ogra>
Q-FUNK, no, but we can add a patch to the existing package i'm pretty sure autofoo is fine as long as you point out only the actual patch in the SRU and note it needs an autocrap run
09:19PMantis has quit IRC
09:19
<ray_>
i would love to use ltsp ... but in my environment ... i don't have access to the network equipment, so i can't do the PXE and tftp stuff ...
09:20
<Gadi>
ray_: are the users *only* using rdesktop?
09:20Guarald1 has quit IRC
09:20BGomes has joined #ltsp
09:21
<ray_>
basically ... yea ... the situation here is that I'm setting up these stations for users here in a library ... and the'll connect to a windows server running terminal services ...
09:21
all they should see is a login prompt ...
09:22
<Gadi>
have a look at thinstation
09:22
<Q-FUNK>
ogra: in that case, we already have a patch. where could I upload it for proposed-updates?
09:22
<ray_>
for their windows login ...
09:22
<Gadi>
im sure they tackle this issue, too
09:22
and they are more geared towards boot-from-disk
09:22
<ogra>
Q-FUNK, talk to pitti in -devel
09:22
Q-FUNK, it needs the paperwork first to not get rejected
09:23
warren, al our sites get hit pretty hard on release day usually ... probably not the best day to look up info on one of the ubuntu wikis
09:23
<ray_>
it's just surprising there isn't already a solution to do this with rdesktop ...
09:23
or ... maybe there's a commercial one out there ...
09:23
<Q-FUNK>
ogra: then again, I have a sneaking suspicion that X core also needs patching, as LTSP no longer detects our dbe61 correctly, since the switch from amd->geode
09:24
<Gadi>
well, I know there's a *commercial* one out there
09:24
:)
09:24
<ogra>
Q-FUNK, well, 8.04.1 is due in july, plenty of time to figure out the bits and pieces and to get them tested
09:24
<ray_>
yea : ) ... exactly ... bah! ... it's bad enough we're stuck connecting to a windows box ...
09:25
but ... like we talked about ... it sounds like redirecting the entire folder for /media is the way to go right now ...
09:25
i wonder if there's a way to speed up the delay for writing to the USB vol ...
09:27
<ogra>
Q-FUNK, i see already uploads to -proposed on hardy-changes btw
09:27
<warren>
I really want to avoid hard coding bootprompt options into my elf image
09:28
I'm looking into chain loading a PXE capable loader instead, but I guess that isn't possible to do in a generic way.
09:28
(works with any network card)
09:28
<Q-FUNK>
ogra: ok, good to know. wanna sponsor pick up some goodies from my PPA and sponsor them, if I prepackage?
09:28toscalix has quit IRC
09:29
<ogra>
Q-FUNK, next week, if i'm dont with the last classmate bits
09:29
weekend will still keep me busy with hardy :/
09:29
<Q-FUNK>
ok
09:29
yes, same thing here. mostly debugging why X 1.4 is still such a mess
09:30hersonls has quit IRC
09:31
<ray_>
by the way ... thanks for all the help ...
09:31hersonls has joined #ltsp
09:36ray_ has quit IRC
09:36mccann has joined #ltsp
09:37elisboa has quit IRC
09:37elisboa has joined #ltsp
09:43
<Blinny>
ogra: Is aliasing a non-LTSP subnet address to a LTSP gig NIC a bad idea? Back when I used to do IP aliasing heavily on Sun Sparcs the system would bog down under heavy load b/c every packet was inspected for source/destination/etc.
09:44
Erm, well, I suppose that's a question for the room. Apologies for direct addressing.
09:46cliebow has joined #ltsp
09:49
<warren>
vagrantc: ogra: have you tried to get older Etherboot to chain load gPXE in order to avoid mkelfImage?
09:50
<ogra>
nope
09:54cliebow has quit IRC
10:00
<vagrantc>
warren: not i
10:00
warren: i have chain-loaded etherboot though
10:00
PXE -> etherboot
10:00
and etherboot -> etherboot
10:00
<warren>
yeah, that's all the docs I can find online
10:00
how do you do etherboot to etherboot?
10:01
<vagrantc>
well, i had different versions of etherboot so i could distinguish the two
10:01
<warren>
any URL describing how to etherboot -> etherboot?
10:03
<vagrantc>
never found any documentation really ... you just need to use the .lilo or .zlilo etherboot and pass that in the filename
10:03
instead of the nbi.img or kernel or whatever
10:03
<warren>
ahhh
10:03pimpministerp has joined #ltsp
10:04
<vagrantc>
warren: but typically etherboot only has a limited set of supported network cards ... what's gPXE ?
10:04
<ogra>
is that grub2 stuff ?
10:05
<warren>
vagrantc: gPXE seems to be the newer etherboot
10:05
<Q-FUNK>
yes and no
10:05
etherboot doesn't need a real bios to load it
10:05
but gPXE does, because it depends upon traditional bios features
10:06
so we cannot build an image with coreboot and gPXE
10:06
<vagrantc>
so chain-loading gPXE isn't going to work around mkelflinux, then?
10:06* warren REALLY wants to avoid hardcoding bootprompt args in the elf image
10:06
<warren>
Q-FUNK: our former OLPC engineers suggested openfirmware is a lot less trouble than coreboot + etherboot
10:07
<vagrantc>
warren: have you looked into passing them via dhcp?
10:07
<warren>
vagrantc: any example of how?
10:07
<vagrantc>
warren: i'll look around
10:08
i think it was possible with mkelf-linux ... hopefully mkelfimage also has the ability to do so
10:08
<ogra>
warren, former ? did fedora drop the ball ?
10:08pimpministerp has joined #ltsp
10:08
<warren>
ogra: we have engineers rotating in and out of it
10:09
<ogra>
ah
10:09
<warren>
one of the earlier engineers was there when they were still fighting linuxbios
10:09
they found openfirmware to be far less trouble
10:10
<Q-FUNK>
it's not less trouble. it's just easier to thinker with.
10:10
it's essentially a geek toy.
10:10gonzaloaf has joined #ltsp
10:10
<Q-FUNK>
make your bootloader cook and do the dishes= OF
10:10
<warren>
don't sparc and apple use it?
10:11
<ogra>
apple did
10:11
they dont anymore
10:11
<warren>
oh?
10:11
<ogra>
they use x86 now ...
10:12
no point in that
10:12
<Q-FUNK>
they use EFI
10:12
<ogra>
right
10:12
<Q-FUNK>
OF is nicely thought, but it's really a geek toy.
10:12pimpministerp has quit IRC
10:13cliebow has joined #ltsp
10:13Arauto has joined #ltsp
10:14
<Q-FUNK>
but it's not much different than running grub2 on top of coreboot
10:14
and coreboot 3 works nicely for that
10:14tux_440volt has quit IRC
10:15
<warren>
sigh
10:16
<ogra>
warren, you need to regenerate the image anyway for changes why not go with it like that fr one release and we figure out something nice over time
10:17
<warren>
ogra: for what kind of changes?
10:17
<ogra>
well, an eherboot image merges vmlinuz and initrd into one file
10:17
<warren>
I'm not against that part
10:18
<ogra>
changes to initrd mean regenerating the image anyway
10:18
<warren>
changes to initrd are only necessary on kernel upgrades
10:22makghosh has quit IRC
10:23
<vagrantc>
warren: in dhcpd.conf: option option-129 "$BOOTPROMPT_OPTS";
10:23
warren: should work with etherboot, at least
10:23
http://www.etherboot.org/doc/html/userman/a838.html
10:24
<warren>
vagrantc: thanks
10:24
<vagrantc>
warren: you may also need to specify the magic etherboot option: option option-128 E4:45:74:68:00:00;
10:24
<warren>
OK, so this means I have two independent places to put boot arts...
10:24
<vagrantc>
otherwise it might ignore 129
10:24
warren: that appends to an existing boot prompt option, rather than overwriting it entirely, though.
10:25Pascal_2 has quit IRC
10:26
<vagrantc>
warren: oh, the exact syntax is broken with 3.x of isc dhcpd ... there's a note at the bottom about it
10:26
warren: but you can find examples of the new syntax in ...
10:27pimpministerp has joined #ltsp
10:27
<vagrantc>
warren: ltsp-trunk/server/configs/k12linux/dhcpd.conf :)
10:27
<warren>
we have 4.0 isc dhcpd
10:27
<vagrantc>
egads!
10:27
<Q-FUNK>
that's out already?
10:27
is FC9 codenamed raw hide?
10:28
<warren>
I have no idea
10:28* vagrantc wonders if isc decided to break anything major
10:28
<warren>
we have *the* ISC dhcpd lead developer on staff now
10:28staffencasa has joined #ltsp
10:28
<ogra>
cool
10:29
<warren>
it is more like, nobody else cared for a long time, and he was the last person to touch it, so now he's doomed to own it.
10:29
<ogra>
heh
10:29
<warren>
looks like we've been building dhcp-4.0.0 since january
10:29* ogra wast aware 4.0 was production ready
10:29
<warren>
it seems OK here
10:29
<ogra>
*wasnt
10:30
<Q-FUNK>
warren: ... mellon?
10:30
<warren>
?
10:30
I don't know the full story
10:30
<Q-FUNK>
the guy
10:30
<warren>
Cantrell
10:30* ogra thought Q-FUNK was handing out fruit
10:30
<Q-FUNK>
:D
10:31
<cliebow>
heehee!
10:31
<laga>
ogra: when did the media centre discussion happen on ubuntu-devel? i can't see it in the archives.
10:32
<ogra>
laga, devel-discuss
10:32
iscussion agenda for UDS-Intrepid around the integration of Elisa media center
10:32
<warren>
vagrantc: so actually the syntax in my ltsp-dhcpd.conf should already work
10:32
<laga>
ogra: argh. yeah, i've just seen it. i didn't realize i have to scroll down.
10:33
<vagrantc>
warren: i believe so
10:33
<cliebow>
ogra:this mean anything to you?looks like panel starts to laod then dies..
10:33
<Blinny>
Does LinuxMCE use LTSP to boot their diskless media centers or do they do something on their own?
10:33
<cliebow>
_cairo_format_from_pixman_format: Assertion `NOT_REACHED' failed.
10:33
<vagrantc>
warren: and here i spent a while searching the whole of the intarweb ... only to look in my ltsp-trunk checkout :)
10:33
<Blinny>
Hmm
10:34
<warren>
vagrantc: eh
10:34
vagrantc: all we have is:
10:34
option option-128 code 128 = string;
10:34
option option-129 code 129 = text;
10:34
option option-221 code 221 = text;
10:35
<vagrantc>
warren: and some commented examples of how to use option-1299
10:35
er, 129
10:37
<cliebow>
t'hell with it..dist-upgrade..
10:37
<Blinny>
heh
10:38
<cliebow>
get a whole new set of issues to work with..tired of the old ones ;-]
10:42
weeeee!
10:44
<Gadi>
option 221?
10:44
VPN?
10:45
<vagrantc>
Gadi: most of the option numbers aren't standardized
10:46
<warren>
oh this really sucks
10:46
vendor class identifier "Etherboot-5.4" on both a PXE capable client and PXE incapable client
10:46
I might not be able to get both to work out of the box with the same dhcpd.conf
10:46
<vagrantc>
warren: you're chainloading etherboot?
10:47
<warren>
vagrantc: no
10:47
vagrantc: trying to boot the thincan using an elf image I created with mkelfImage
10:47
but not getting that far even
10:47
it is trying to load pxelinux.0
10:47
<vagrantc>
oh.
10:47
<warren>
because its vendor class identifier is not unique
10:47
<Gadi>
vagrantc: /me wonders what warren uses 221 for
10:47
<warren>
Gadi: I have no idea where that came from
10:48
<vagrantc>
yes, i don't know if there's a way to distinguish PXE capable etherboot from non-PXE capable etherboot
10:48
<ogra>
cliebow, use update-manager *dont* dist-upgrade
10:48
<warren>
Gadi: copied from eharrison
10:48
<laga>
ogra: btw, why? does update-manager add specific hacks?
10:48
<Gadi>
ah - maybe its a Mac thing
10:48
:)
10:48
yaboot, et al
10:48
<ogra>
laga, tons, yes
10:48
<vagrantc>
warren: but etherboot since 5.4 has defaulted to PXE support ... this is not the case with the thincan ?
10:48
<laga>
ogra: that's bad :/
10:49
<ogra>
laga, every failure that showed up in a test during the last three months is covered by it
10:49
<warren>
vagrantc: there's both PXE and coreboot + etherboot thincan
10:49
vagrantc: the PXE one is fine
10:49
<ogra>
we highly discourage apt-get dist-upgrade since about two years
10:49
<laga>
ogra: were those failures also fixed in the packages themselves?
10:49
<ogra>
but people dont listen and trash their systems
10:49
<warren>
hmm
10:49
Client MAC address: ArtecGro_00:05:85 (00:18:2d:00:05:85)
10:49
<cliebow>
ogra:righto..
10:49
<ogra>
laga, there is nothing fixable
10:49
<Gadi>
warren: so the PXE-based thin can reports a VCI of Etherboot?
10:50
<warren>
Gadi: yes
10:50
<Gadi>
yuck
10:50
<warren>
Gadi: some Etherboot can do PXE, others can't
10:50
<ogra>
laga, conffile changes, dpkg --configure runs you would need to run manually with certain package combos etc
10:50
<cliebow>
pgra: you strongly urged this last time i borked my system ;-]
10:50
<Gadi>
right - but I thought the PXE_based thincan uses commercial PXE
10:50
<warren>
yes
10:51
<Gadi>
so, its VCI should be PXEClient
10:51
or some such
10:51
<ogra>
laga, definately never ever recommend dist-upgrade to anyone wanting to do a release upgrade
10:51
<laga>
ogra: ah. conffile changes would cause a prompt. nothing an experienced user can't handle ;)
10:51
ogra: ok, i wont
10:51
<warren>
Gadi: commercial PXE is PXEClient
10:51
<cliebow>
so much traffic..i can wait a few days..
10:51
<warren>
Gadi: my problem is some etherboot (like qemu) can do real PXE while this cant.
10:51
<ogra>
laga, there is more, conffile prompts are just a prominent example ...
10:51
<warren>
and both identify themselves as Etherboot-5.4
10:52
<vagrantc>
warren: keep adding numbers ...
10:52
<warren>
?
10:52
<laga>
ogra: i'll just have to believe you then :)
10:52
<vagrantc>
warren: Etherboot-5.4.12 ...
10:52
for example
10:52
<ogra>
cliebow, just make sure to have gutsy-updates enabled and have the latest update-manager installed ...
10:52
<Gadi>
you are burning a new etherboot into the thincan?
10:52
<ogra>
it will show a button to upgrade to hardy by defaut if you have the right version
10:52
<warren>
vagrantc: I did a tcpdump of the thincan with coreboot + etherboot, it is only "Etherboot-5.4"
10:53
<vagrantc>
warren: ok, and the other etherboot machines?
10:53
<warren>
vagrantc: "Etherboot-5.4" as well
10:53
<vagrantc>
for shame.
10:53
<warren>
yeah
10:53
<Gadi>
if you are doing the burning, can you not adjust the VCI?
10:53
<warren>
It might not be possible to do this out of the box now.
10:53
Gadi: I didn't do any burning
10:54
<Blinny>
apt-get is deprecated anyway, right?
10:54
<warren>
I don't know how to burn
10:54
<Gadi>
you are using a new etherboot image, no?
10:54
<vagrantc>
warren: so your goal is ... to get this model
10:54* Gadi is so confused
10:54ray_ has joined #ltsp
10:54
<warren>
Gadi: I'm just trying to boot this thincan with its coreboot + etherboot
10:55
vagrantc: my goal is to make everything possible boot using our default dhcpd.conf
10:55
<Gadi>
and its coreboot+etherboot is doing PXE?
10:55
<warren>
which now seems impossible because there's no way to identify it
10:55
<Gadi>
but reporting Etherboot VCI
10:55
<warren>
Gadi: no
10:55
Gadi: this thincan has coreboot + etherboot and reporting "Etherboot-5.4"
10:55
<Gadi>
so, feed it an etherboot image like every other etherboot
10:55
<warren>
Gadi: the problem is that the "Etherboot-5.4" name is ambiguous
10:55
<vagrantc>
warren: so, you want etherboot that is capable of doing PXE to do PXE, but etherboot that's not capable to do the old-school stuff ?
10:56
<warren>
yes
10:56* vagrantc agrees with gadi
10:56
<vagrantc>
if it's etherboot, give it an etherboot image ...
10:56
<warren>
that means qemu can't test pxelinux anymore
10:56
<Gadi>
why?
10:56
<vagrantc>
not with a default configuration, no.
10:57
<Gadi>
wait a qemu client?
10:57
<vagrantc>
i think it very well may be impossible to distinguish and have a default that works in all cases.
10:57
<Gadi>
I am so confused
10:57
<warren>
qemu reports itself as Etherboot-5.4
10:57
<Gadi>
do this:
10:58
download the bootable iso image for etherboot-PXE from rom-o-matic and boot qemu from cdrom
10:58
that should (maybe) report PXEClient?
10:58
dunno
10:58
only ever tested etherboot booting with qemu
10:59* Gadi knows not how to test PXE-booting with qemu
10:59
<Gadi>
I dont even think it supports it
11:00
but, I bet you could modify the VCI of the rom-o-matic image
11:00
<vagrantc>
warren: without changes to qemu or etherboot code, i don't think you can accomplish what you want.
11:00
<warren>
hmm
11:00
<Gadi>
and in that way, get a PXE-capable bootable image with VCI=PXEClient
11:01
<ogra>
there are etherboot imaes with pxe emu, no ?
11:01
<warren>
Are you saying there's no real machines that report "Etherboot-5.4" that are actually PXE capable?
11:01
<vagrantc>
ogra: that's exactly the problem
11:01
warren: i would wager to say most machines that report Etherboot-5.4 are PXE capable.
11:01
<Gadi>
warren: not that I know of
11:02
and if there were, they should report a VCI of PXEClient
11:02
otherwise they are dup'ing everyone
11:02
<warren>
vagrantc: because most machines are using a real BIOS?
11:02
<vagrantc>
Gadi: no, because they have PXE capability ... but they're still etherboot
11:02
warren: indeed.
11:02
Gadi: they support PXE, NBI, ELF, etc ...
11:02
<Gadi>
big deal, they still need to tell somebody that
11:02
:)
11:02
somehow
11:03
<ogra>
isnt the "runs with netware" sticker supposed to tell you that ? :)
11:03
<Gadi>
so, there must be a flag to tell the server that
11:03
<vagrantc>
Gadi: or there should be
11:03
<Gadi>
otherwise, you just feed them all etherboot images
11:03
<warren>
strange, I don't see the string Etherboot anywhere in qemu
11:03
<vagrantc>
Gadi: i'm thinking we should get etherboot to report it's supported types in the VCI string somehow
11:04
<warren>
vagrantc: except dhcpd doesn't do substring matches
11:04
<Gadi>
vagrantc: that would be great for the future
11:04
<warren>
vagrantc: it matches from the beginning of the string
11:04
<Gadi>
not exactly backward compatible
11:04
<vagrantc>
warren: it's better than sending *nothing*
11:04
<warren>
true
11:04ray_ has quit IRC
11:04
<vagrantc>
you'd have to have multiple lines or something ...
11:04
<Gadi>
dhcpd does do substring matches
11:05
<warren>
of VCI
11:05
?
11:05
<Gadi>
it only starts at the beginning because we set offset to 0
11:05
:)
11:05ray_ has joined #ltsp
11:05
<vagrantc>
Gadi: well, it's not backwards-breaking either, if it lists them after the existing string
11:05
<Gadi>
match if substring (option vendor-class-identifier, 0, 9) = "PXEClient";
11:05
the 0 in that line means "from first char"
11:06
the 9 means nine chars
11:06
you can match as you like
11:06
<warren>
I can't do this by MAC address either
11:06
because the MAC prefix of both the PXE and non-PXE thincans are the same
11:06
<vagrantc>
Gadi: we'd have to standardize on a format, though...
11:06
<Gadi>
yeah, MAC is independent
11:07
<vagrantc>
Gadi: i.e. Etherboot-5.4-ELF-NBI-PXE ... what if it just supports PXE ?
11:07
<Gadi>
vagrantc: there's also a limit of chars
11:07
<vagrantc>
Gadi: how big of a limit?
11:07* Gadi forgets if it is 16 or 32
11:07
<ogra>
i think it was 16
11:07
<vagrantc>
16 would suck.
11:08
<Gadi>
well, I want to standardize for the initramfs, too
11:08
<vagrantc>
seeing as Etherboot-5.4 is 13
11:08
<Gadi>
I send "LTSP" as VCI from initramfs
11:08
<ogra>
to where ?
11:08
<warren>
Does all "Etherboot" support elf?
11:08
<Gadi>
but, remember we talked about "LTSP-i386", etc
11:08
<vagrantc>
Etherboot-5.4-ENP ... Etherboot-5.4-XXP ... ?
11:08
<Gadi>
so we could assign different rootfs's based on arch?
11:08
<ogra>
Gadi, oh, in the second dhcp call you mean
11:09
<Gadi>
ogra: yeah
11:09exodos has quit IRC
11:09
<Gadi>
put that in the back of your head for when/if we do multi-arch support
11:09
:)
11:09
<vagrantc>
warren: older versions only supported NBI, newer versions support ELF, and even newer versions support PXE ... but it's possible to build etherboot images with only some or none of them.
11:10
<warren>
do newer versions support NBI?
11:10
<vagrantc>
warren: i believe default since 5.4.0 was to support all of them ... but i may be mistaken.
11:10
<warren>
I might have to do least common denominator
11:10
<vagrantc>
warren: i bet your thincan doesn't support NBI
11:10
<warren>
I think you're right
11:11
oh, nope, it does NBI
11:11
<vagrantc>
warren: basically, it can support 1-3 of any of them, depending on whoever compiled etherboot.
11:11
<generic>
ogra
11:11
<vagrantc>
warren: so technically, there's no least common denominator, as it's a pick-and-choose.
11:12
warren: all versions have the capability to support NBI, but not all builds will support it.
11:12
<warren>
I could make "Etherboot-5.4" default to ELF
11:12
"Etherboot" fallthrough NBI
11:12
<vagrantc>
that's pretty reasonable.
11:13
<warren>
if that doesn't work for anyone then they need to edit it manually
11:13
<vagrantc>
exactly.
11:13
<ogra>
warren, in th eltsp5 implementations out there already they need to do that anyway ...
11:13
<vagrantc>
warren: and to test pxelinux from qemu, you'll have to edit it manually.
11:38BGomes has quit IRC
11:39viking-ice has quit IRC
11:41mhterres has joined #ltsp
11:42
<warren>
vagrantc: in your bootprompt args you have no "initrd="?
11:43
checking if image is initramfs...it isn't (bad gzip magic numbers); looks like an initrd
11:44Skarmeth has joined #ltsp
11:44
<warren>
and it gets stuck
11:48
[root@newcaprica i386]# file elf-2.6.25-1.fc9.i586.img
11:48
elf-2.6.25-1.fc9.i586.img: ELF 32-bit LSB executable, Intel 80386, version 1 (SYSV), statically linked, corrupted section header size
11:48
hmm
11:50indradg has joined #ltsp
11:51
<generic>
hey guys do i need to install tftpd on ubunrt +ltsp 5
11:51
<ogra>
it is pulled in automatically
11:51
<generic>
though i have install ltsp-server-standalone openssh-server
11:51
sudo ltsp-build-client
11:52
still
11:52
tftp error
11:52
<ogra>
ltsp-server Depends: debconf (>= 0.5) | debconf-2.0, debconf-utils, debootstrap, gettext-base, iproute, lsb-release, nbd-server, openbsd-inetd | inet-superserver, openssh-client | ssh, python (>= 2.4), squashfs-tools, tcpd, tftpd-hpa, update-inetd
11:53
ltsp-server-standlone depends on ltsp-server
11:53Q-FUNK has quit IRC
11:53
<ogra>
generic, are you running an i386 server or amd64 ?
11:53
<generic>
i386
11:54
<ogra>
ls /var/lib/tftpboot/ltsp/i386 ?
11:54
are there files ?
11:54
grep tftp /etc/inetd.conf
11:54
does that return a line for tftp ?
11:55
<generic>
wait
11:55
<vagrantc>
warren: yup, no need for an initrd argument.
11:55
warren: at least on debian and ubuntu
11:55
<warren>
i can't manage to boot anything made with mkelfImage
11:55
i'm trying mkelf-linux now
11:56
<generic>
yes
11:56
they are
11:56
<ogra>
generic, the inetd.conf has a line as well ?
11:57
<generic>
what line
11:57
<vagrantc>
oh, by the way ... tried dnsmasq both as DHCP server and TFTP server, and works quite nicely.
11:58
<ogra>
generic, grep tftp /etc/inetd.conf
11:58
does that return a line for tftp ?
11:58
(one that doesnt start with a # sign)
11:59
<generic>
no
11:59
<ogra>
weird
12:00
dpkg -l tftpd-hpa
12:00
is it installed ? (output should contain ii somewhere)
12:01
<warren>
doh
12:01
mkelfImage is generating a corrupted image
12:01
<ogra>
your file output somewhat indicated that
12:01
"corrupted section header size"
12:01
<warren>
I know
12:02
<Blinny>
That tftp /etc/inetd.conf sounds like the funky postinstall problem again.
12:03
<generic>
its ok
12:03
thin client when boot says tftp can not open connection
12:03
yes installed
12:03
<ogra>
Blinny, except that i had about 100 tests on different platforms and didnt see it anywhere but for ppl that had debian in their sources.lst or other weird stuff yet
12:03
<generic>
this is ined.conf file
12:03
571 stream tcp nowait nobody /usr/sbin/tcpd /usr/sbin/ldminfod
12:03
9572 stream tcp nowait nobody /usr/sbin/tcpd /usr/sbin/nbdswapd
12:03
2000 stream tcp nowait nobody /usr/sbin/tcpd /usr/sbin/nbdrootd /opt/ltsp/images/i386.img
12:03
~
12:04
<ogra>
generic, ok, but you have that line in inetd.conf ?
12:04
no, thats wrong
12:04
<generic>
i make tftp on boot yes too
12:04
in /etc/default
12:04
<ogra>
argh
12:04
dont do that
12:04
<generic>
what wrong
12:04
<warren>
sigh, it doesn't like the image generated by mkelf-linux either
12:04
<ogra>
sudo dpkg-reconfigure tftpd-hpa
12:04
make sure it gets started from inetd
12:04* warren tries the i386 version
12:05
<generic>
ok i did
12:05
<ogra>
good
12:06
just leave all defaults the system sets its been tested and known to work that way :)
12:06
<generic>
oh wait
12:06
filename "/lts/i386/pxelinux.0";
12:06
in dhcp is this
12:07
<ogra>
lts ?
12:07
are you sure ?
12:07
thats wrong
12:07
<generic>
yes
12:07
this is wrong
12:07
right
12:07
<ogra>
needs to be ltsp
12:07
<Blinny>
left
12:07
<ogra>
did you edit the file as well ?
12:08
<generic>
yes
12:08
what it should be
12:08
<ogra>
the default
12:08
as it was installed
12:08
<warren>
ogra: I just tried the mkelfImage binary from edubuntu and it generates the same corrupted image
12:08
so it is likely something with my kernel...
12:08
<ogra>
only edit the IP range in case you use a different static IP for the thin client interface
12:09
warren, to big probably
12:09
<warren>
oh?
12:09
there's size limits for elf images?
12:09
<ogra>
no idea
12:09
just a guess
12:10
i dont really know if there is a size limit, but as i said before, it merges kernel and initramfs
12:10
that can get quite big
12:11
-r--r--r-- 1 root root 6,0M 2007-12-29 01:07 nbi.img-2.6.24-16-386
12:11
generic, the defaults of all apps and config files are set to support booting thin clients out of the box, if you changed anything, change it back
12:11
<warren>
it is possible my initrd is too big
12:12
<generic>
where is default dhcp file
12:12
<ogra>
its /etc/ltsp/dhcpd.conf
12:12
<generic>
oh
12:13
i just delete and make my own
12:13
<ogra>
sudo apt-get install --reinstall ltsp-server might reset it
12:13
<generic>
can u check plz
12:13
<ogra>
(not sure that works for conffiles like dhcpd.conf though)
12:14
<generic>
http://pastebin.com/m1260065b
12:14
here my dhcp file
12:14
<vagrantc>
dnsmasq seems like it was practically designed for LTSP these days.
12:14
<generic>
also i have set a range but clinets are getting ips line 40 , 69 ,
12:15
whats is that
12:16
<ogra>
thats totally messed up
12:16
just try the command i gave above to restore the default file
12:17
<generic>
not restoring
12:17soneyka has quit IRC
12:17
<generic>
no use
12:21fuhviofbi has joined #ltsp
12:22
<ltsppbot>
"ogra" pasted "default ubuntu dhcpd.conf" (22 lines) at http://pastebot.ltsp.org/516
12:22
<ogra>
copy that in place, replace 192.168.0 globally with your network addess
12:22
restart dhcp3-server
12:23
if you changed any other files, change them back as well
12:25
<shogunx>
hey ogra. what is the purpose of the i386.img squashfs image generated by ltsp-build-client?
12:25
<vagrantc>
shogunx: it's your root filesytem.
12:26
<ogra>
well, the readonly part of your root fs :)
12:26
<vagrantc>
well, yeah.
12:26
<ogra>
:)
12:26
<shogunx>
main reason for asking is that i will need to make a few mods to the chroot. one terminal will have an atheros, and act as an access point, among other things.
12:26
<ogra>
shogunx, best is to generate a special chroot for these if you have the space on your server
12:27
have a look at ltsp-build-client --extra-help
12:27
it has various options
12:27
<shogunx>
its a few additional kernel modules (which are there anyway) and an init script i will have to add. thats it.
12:27
<ogra>
if yu made changes in the chroot, run ltsp-update-image -a <chrootname> to regenerate it
12:28
ah, k
12:28
<shogunx>
ahh. thats what i was looking for i think. thanks.
12:28
140mb... that all loads into the client?
12:28
<ray_>
... i was on earlier this morning ... asking about rdesktop and local disk drive access. I actually had some luck getting the /media directory to mount ... but it's pretty bad about the delayed writes to the USB disk.
12:28
<ogra>
if you have only an i386 chroot you can omit -a
12:29
<ray_>
may have to look into the ltsp file system ...
12:29
<shogunx>
omit -a. check.
12:29
still functional with this method with 128 megs of memory on the client, correct?
12:30
still a nbd, just with squashfs also. or does the entire squashfs image load into ram on the client?
12:31
or should i be ordering 256meg sodimms?
12:31
<ogra>
128M shoudl be fine
12:31
the squashfs sits on a network block device
12:31
<shogunx>
excellent. many thanks.
12:31
<ogra>
so it doesnt load into ram
12:31
<shogunx>
i dig it.
12:34soneyka has joined #ltsp
12:35
<warren>
ogra: how big is your initrd?
12:35
the gzip compressed cpio archive
12:36
<ray_>
so ... as i understand it, there's no way right now to use the ltspfs or ltspfsd stuff in the ubuntu repositories by itself? ...
12:36
<ogra>
-rw-r--r-- 1 root root 4,3M 2008-02-26 15:40 initrd.img-2.6.24-16-generic
12:36
<warren>
thx
12:36
my initrd is twice the size
12:37
I'll try one with all the crap ripped out
12:37jammcq has joined #ltsp
12:37vagrantc has quit IRC
12:37
<ogra>
yo jammcq
12:38Blinny has quit IRC
12:39
<ray_>
Does anyone here know anything about using the ltspFS stuff? ...
12:40
<ogra>
yes, but nothing about rdesktop
12:41
<ray_>
is the ltspfs supposed to be used in a client server function? ... like ... the server specifically mounts the clients local devices ON the server itself with fuse? ...
12:43
i guess my ultimate goal is to make the usb device have faster writes to the disk, so that by the time the user is logging out ... their disk should have finished making the writes ...
12:43
<ogra>
if you plug in a block device, an udev rule is executed that add the device with mountpoint to a specal fstab file .... then it triggers the mount through the ssh tunnel we use for X transport on the server side
12:44
in the users session on the server side ltspfsmounter triggers the actual network boot
12:44
s/boot/mount/
12:45
inbetween there is a token exchanged that sits on the X root wondow of the users session to make sure the user is allowed to actually do the mount
12:46
<ray_>
what does the udev rule look like? .. can you do something similar with ivman? ...
12:48
it seems like with block devices ... if you lose the connection, or the connection hicups or something like that ... won't that corrupt the device? ...
12:49
and how does the ltspfs handle a logout situation to correctly end communications with the disk? ... (finish writing and unmount ... etc)
12:55
sorry for asking so many questions ... i guess i'm just looking for a simple solution to the problem of local media ... i guess the short answer is that there really isn't one : )
12:58spectra has joined #ltsp
13:02
<ray_>
hey ogra, i finally found the page in the wiki on ltspfs and fuse ... do you think that it would make any sense to run both ltspFSd and ltspFS on the same box, to take advantage of some of this stuff?
13:03
<ogra>
ray_, i think Gadi does such strange things :)
13:03* ogra -> dinner
13:03
<ray_>
thanks ogra : )
13:13
Gadi ... is there anyway i could get in contact with you about some of the ltspfs stuff? stuff?
13:24
<Gadi>
ray_: hi - sorry in and out ... that is the hackery of which I spoke this morning
13:24
I run ltspfs and ltspfsd on the thin client to take advantage of it for rdp et al
13:25
it requires some trickery (as this is not the intended use)
13:25
:)
13:25
but, it works great
13:25
there are also a few other friends that need tweaking (cdpinger, etc)
13:27
<ogra>
cdpinger gets harder to tweak now that its C
13:28
warren, btw, "AMD64 replaces terms such as “Hammer” (now, “AMD64 technology”) and “x86-64 ISA” (now, “AMD64 ISA”)."
13:28
from the officia release of the amd64 arhitecture
13:28
http://www.amd.com/us-en/Corporate/VirtualPressRoom/0,,51_104_543_10218~69678,00.html
13:28
<warren>
isn't that really old?
13:29
<ogra>
no, i just had a discussion and was taucght some history
13:29
x86_64 was the work title for the development of the arch
13:29
<warren>
mkelfimage doesn't seem to like our kernel/initrd combination
13:29
<ogra>
apparently suse ported on these dev CPUs
13:30
<warren>
We originally named our stuff amd64
13:30
<ogra>
so they used that name and stuck with it instead of moving to the official name
13:30
<ray_>
that would be awesome thought if you could maybe show me how to use both of them on the client for RDP based stuff ...
13:30
<Gadi>
yeah, well, I will try to collect my hacks before hackfest, so perhaps we can review any pieces that might be folded in
13:30
<warren>
but we renamed it to x86_64 after we supported both Intel and AMD
13:30
glibc's name is x86_64 right?
13:30
<Gadi>
ray_: well, keep in mind, I am still on gutsy
13:30
<ogra>
nope
13:31
<Gadi>
and have no intention of moving to hardy for the rootfs any time soon
13:31
<ray_>
that's cool ... i can move back to gutsy if need be ...
13:31
<ogra>
apparently the BSD's Debian, Ubuntu, Gentoo, JDK, Solaris all refer to it as 'amd64', as well as gcc, kernel and lsb
13:31
<Gadi>
and gutsy->hardy changed a few things (to say the least) in rootfs
13:31
<ogra>
Gadi, ?
13:31
what changed ?
13:31* Gadi has limited C kung-foo, so the more we do in C, the more it hurts me to hack :)
13:32
<ray_>
ok ... i actually just upgraded my client this morning ... i'll happily change back ...
13:32
<Gadi>
ogra: well, the python-> C reqwrites, all of the directories...
13:32
:)
13:32
<pscheie>
Gadi, what language(s) do you like to work in?
13:32* ogra didnt make any changes to ltsp rootfs in hardy
13:32
<ogra>
apart from general package updates
13:32
<Gadi>
all the paths got moved around
13:33
that always takes me time to unravel
13:33
:)
13:33
<ogra>
thats warrens fault :P
13:33
<Gadi>
pscheie: I'm a scripter not a programmer
13:33
<ogra>
but thanks to vagrant all the old paths should still work
13:33
<Gadi>
I prefer as many interperative languages I can get my hands on
13:33
<pscheie>
Gadi, ergo?
13:33
<Gadi>
easier for a simple-minded EE like me
13:33
:)
13:34
ergo: perl, php, python, awk, sed, ...
13:34
<ogra>
shell ;)
13:34
<Gadi>
shell
13:34
:)
13:34
C hurts my brain
13:34
<pscheie>
join the club ;-)
13:34
<Gadi>
which is how I know Im not a programmer
13:34
<ray_>
Gadi_ is there anyway you could put together a little howto on getting ltspFS and ltspFSd working for the RDP client? ...
13:35* pscheie hasn't done any C for 10+ years
13:35
<Gadi>
ray_: its so many little tweaks here and there, I would prefer to make a package
13:35
<ray_>
my email is ray.voelker@gmail.com ... basically local disk support is the final piece of this puzzle for me ...
13:35
<Gadi>
costs the same amount of effort
13:35
<ray_>
if you could make it a package ... that would work great too ...
13:36
<Gadi>
for me, if its a super-long howto, better to have a script called "install" ;)
13:36
<ray_>
even better ... : )
13:36
<Gadi>
but, dont put me in ur critical path
13:36
:)
13:37
I have some larger fish to fry before I can get to that
13:37
I can talk about it in generalities, tho if you like
13:37
:)
13:37
<ray_>
yea ... if you would like ... i'll try to keep up ...
13:37
<Gadi>
ok
13:37
<ray_>
maybe i can put some of these hacks together on my own in the mean time ...
13:38
<Gadi>
unfortunately, I'm the only one of this lil groupito who really cares about connecting to rdp, so...
13:38
<ray_>
yea .. that was my stumbling block as well ...
13:38
<Gadi>
actually, there's an easier way to skin this
13:38
if ur open to it
13:38
<ray_>
i was actually just using rdesktop without the ltsp stuff ...
13:39
i just thought that someone here was more familiar with rdesktop and local disk support ... : )
13:39
<Gadi>
you can set up an LTSP server, give every machine a user account, and create a session on the server (ie a ~/.xsession for each user) that runs rdesktop -f -r disk:drives=/media/$USER ....
13:40
this way, rdesktop runs on the *server* and not on the client
13:40
but the localdev stuff that is default works without any mods
13:40
<ray_>
wait ... that kinda melted my brain a little bit .. : )
13:40
<Gadi>
lol
13:41
ok, so ltsp by default handles giving you a directory on the linux server that you log into for each drive on the client
13:41
it creates the dir in: /media/$USER/
13:41
<ogra>
Gadi, that could be an rc script, triggered by a ldm session entry ;)
13:41
<Gadi>
if you redirect that dir thru rdp,
13:42
and run rdp as the only program in your session
13:42
and use ldm's Autologin feature
13:42
then, the users sit at an rdesktop screen at the client
13:42
that is presented by the server
13:42
they are autologged into the server as their own user
13:42
(say user=hostname or some such)
13:43
<ogra>
that has also the advantage of redirecting the load :)
13:43
<Gadi>
and the localdev is cared for
13:43
the only thing that sucks doing it this way is sound
13:43
<ray_>
oh, ok ... in my situation, the RDP server is a windows box ... and i was looking to basically build a client that could make the server aware of the local disks ...
13:43
<Gadi>
because rdesktop sound would need to be redirected thru padsp
13:43
which is worse that if rdesktop is local on the client
13:44
<ray_>
but you're saying that the only way to do that is by running a linux box with rdp server running with ltsp configured ...
13:44
<Gadi>
right you have two servers:
13:44
WTS and LTSP server
13:44
and then ur thin clients
13:44
<ray_>
oh ... but you're saying you could and the LTSP server to the mix ...
13:44
oh .. ok... thats a little more clear ...
13:45
<Gadi>
LTSP boots the thin clients, they log into it, but it just runs rdesktop fullscreen to the WTS
13:45
so the rdp client rus on the LTSP server
13:45
not on the thin client
13:45
it just gets displayed on the thin client
13:45
<ray_>
oh ... i see ...
13:45
<Gadi>
since LTSP cares for localdev thin client to LTSP server
13:46
all you need to do is redirect the dir created on the ltsp server for the devices
13:46
problem solved
13:47* Gadi does all of his hackery bec our products don't work that way (www.thesymbiont.com)
13:47
<ray_>
excuse my lack of knowledge of ltsp ... but will the ltsp server then be able to facilitate a link between clients and the WTS? ...
13:48
<Gadi>
define "facilitate a link"
13:49
<ray_>
i guess i'm just unclear on how the ltsp server works into the equation of rdesktop client ... and WTS ... if your rdesktop client is now talking to the LTSP box ...
13:49
<johnny>
no.. rdesktop client is on the ltsp box
13:49
<Gadi>
right
13:49
<johnny>
talks to the wts
13:50
<ray_>
oh, so your thin clients are booting from code retrieved from the ltsp server ...
13:50
<Gadi>
with LTSP all applications run on the server
13:51
<ray_>
is this basically through an ssh tunnel? ...
13:51
<Gadi>
the thin clients just get enough of an OS to connect to the server
13:51
it can be
13:51
:)
13:51
by default, with ldm, yes it is
13:51
but there is an option to not tunnel the video thru ssh
13:52
(which is MUCH faster)
13:52
(tho less secure)
13:52
(tho MUCH faster)
13:52
:)
13:52
<ray_>
yea ... i'm not sure about the security ... that may be less of a concern ...
13:52
<ogra>
well, dont do homebanking with encryption off :)
13:53
<Gadi>
...on an untrusted network...
13:53
<ogra>
password handling is in any case done via ssh
13:53
and i.e. in the ubuntu default two NIC setup the TC network isnt publically exposed
13:53
<ray_>
and that's just in the connection between each client PC and the LTSP server? ...
13:53
<Gadi>
so little trust these days....
13:54
<ogra>
right
13:54
<ray_>
so each client would have to be individually created on the server ... before they were able to use any local disks? ...
13:55
<ogra>
no
13:56
<Gadi>
its magic
13:56
plain and simple
13:56
<ray_>
: ) ... awesome ...
13:57
<Gadi>
ray_: if you have a server or even ubuntu installed with qemu or some other virtual machine, just try LTSP and use it a bit
13:57
it is easy to set up
13:57
<ray_>
i'll have to look into this some more ... now the packages for ubuntu in gutsy would be a good place to start with this? ... these aren't disk-less clients ... and unfortunatly, i don't have an environment where i can PXE and tftp : (
13:57
<Gadi>
and u'll get an idea of what it does
13:57
why can't you PXE/tftp?
13:57
everyone can PXE/tftp
13:58
even cheney says so
13:58
:)
13:58
<ray_>
i just dont have access to the dhcp servers here ... : )
13:58
<Gadi>
so, dont use em
13:58
<warren>
gPXE can download via HTTP apparently =)
13:58
<ray_>
will it work without? ...
13:58
<Gadi>
do you have access to the switch?
13:58* ogra wonders hos cheney would pronounce PXE/tftp
13:58
<ogra>
*how
13:58
<ray_>
unfortunately no ... i suppose i could work on that ...
13:59
<cliebow>
Gadi: i hate to think..
13:59
<ogra>
warren, scary
13:59
<Gadi>
ray_: if you stick your ltsp server between the clients and the WTS
13:59
<ray_>
so, much of LTSP hinges on the ability to PXE/tftp ? ...
13:59
<Gadi>
then, you can bypass their dhcp server and use your own
14:00
Cheney's DHCP server --> WTS, LTSP server --> thin clients
14:00
<ray_>
oh geez ... running dhcp server internally may be an automatic death sentence ... the university it people at my institution are pretty harsh ... : )
14:00
<cliebow>
The right of the people to be secure in their persons...any one want to donate some dna to the new secure database?
14:01
<Gadi>
ray_: well, if you do it right, they wouldnt even know
14:01
<ray_>
haha .. : ) ... nice ...
14:01
<Gadi>
but, if thats a limitation, check out thinstation
14:01
they make bootable cdroms and such
14:02
that can boot into rdesktop
14:02
or use XFCE desktop
14:02loather-work has quit IRC
14:02
<ray_>
ok ... i'll look into it ... i've been meaning to try and get into the PXE stuff for a while now ..
14:02
<Gadi>
heh
14:02
will be tough in ur university, tho, it seems
14:02
<ray_>
yea ... that was my last resort .. i was trying to stay away from any type of window manager ...
14:03
<Gadi>
ah, the ties that bind...
14:03
<ray_>
thanks Gadi by the way ... you've really been amazingly helpful ...
14:04elisboa has quit IRC
14:04
<ray_>
ok ... i'll have to get back in touch with you later ... i've gotta get to a meeting ...
14:04elisboa has joined #ltsp
14:05Arauto has quit IRC
14:21pscheie has quit IRC
14:21mikkel has quit IRC
14:24loather-work has joined #ltsp
14:25
<warren>
ogra: somebody in #coreboot said there is also http://www.kernel.org/pub/linux/boot/wraplinux/
14:25
tried this?
14:39cliebow has quit IRC
14:41loather-work has quit IRC
14:41nicoAMG has joined #ltsp
14:43indradg is now known as idg|dinner
14:49pscheie has joined #ltsp
14:52fuhviofbi has quit IRC
14:55Q-FUNK has joined #ltsp
14:56pscheie has quit IRC
14:59gonzaloaf_work has quit IRC
15:02K_O-Gnom has quit IRC
15:08sepski has joined #ltsp
15:08loather-work has joined #ltsp
15:12nantes_geek has joined #ltsp
15:23pscheie has joined #ltsp
15:25
<pscheie>
ogra, didn't you say f4 at ubuntu boot gives 'install ltsp bits' option?
15:25
<laga>
pscheie: only with the alternate disk
15:25
<pscheie>
ah, ok
15:26
time to bittorrent that one, then
15:30nantes_geek has quit IRC
15:30TelnetManta has quit IRC
15:31ray_ has quit IRC
15:35loather-work has quit IRC
15:36rjune_ has joined #ltsp
15:36
<rjune_>
Howdy folks
15:36
!G
15:36
<ltspbot>
rjune_: "G" is Gadi!!!!!!!!!!!!!!!!!!!!!!!!
15:38
<rjune_>
is there any documentation on how LTSP 5 works? Just a brief overview as I apparently get to present LTSP to the local LUG
15:39lns has joined #ltsp
15:40gonzaloaf_work has joined #ltsp
15:41
<lns>
Happy Heron day!
15:45idg|dinner is now known as indradg
16:02slidesinger has quit IRC
16:06Guaraldo has left #ltsp
16:13vagrantc has joined #ltsp
16:20jammcq has quit IRC
16:22DonSilver has joined #ltsp
16:25Gadi has left #ltsp
16:31
<laga>
README-DEVELOPMENT-POLICY
16:31
okay, why is that file empty? ;)
16:31
<vagrantc>
because there is no established policy, but we'd like one somday. :)
16:35hersonls has quit IRC
16:37Egyptian[Home] has joined #ltsp
16:41Bengoa has left #ltsp
16:50sepski has quit IRC
16:54DonSilver has quit IRC
16:54nicoAMG has quit IRC
16:54vagrantc has quit IRC
16:55Skarmeth has quit IRC
16:59mhterres has quit IRC
16:59DonSilver has joined #ltsp
17:10
<laga>
ogra: i'm removing the tabs in client/initramfs/scripts/ltsp_nbd if you don't mind
17:16DonSilver has quit IRC
17:28alekibango has joined #ltsp
17:39mccann has quit IRC
17:50staffencasa has quit IRC
18:05BGomes has joined #ltsp
18:29loathing has joined #ltsp
18:34gonzaloaf_work has quit IRC
18:39BGomes has quit IRC
18:47Q-FUNK has quit IRC
19:19TelnetManta has joined #ltsp
20:03J45p3r has joined #ltsp
20:08mccann has joined #ltsp
20:54petre has joined #ltsp
20:58J45p3r has left #ltsp
21:07vagrantc has joined #ltsp
21:09soneyka has quit IRC
21:26
<lns>
Can I ask if there is anything completely different about how ltsp-server works in hardy (vs gutsy) ?
21:27soneyka has joined #ltsp
21:30
<vagrantc>
hrm. i'm reconsidering my whole limiting ltsp architectures for debian thing
21:32
<lns>
vagrantc, does that mean we'll see ltsp for c=64 sometime soon? ;)
21:32
<vagrantc>
http://packages.debian.org/ltsp-client-core
21:33
currently: alpha amd64 arm armel hppa i386 ia64 mips mipsel powerpc s390 sparc
21:34
<lns>
vagrantc, what makes you want to reconsider?
21:34
<vagrantc>
too much beurocracy to drop them.
21:34
<lns>
and what's difficult about maintaining them all?
21:35
<vagrantc>
well, we have no way to test support for most of them.
21:35
<lns>
lack of hardware?
21:35
<vagrantc>
and the buildd infrastructure within debian sometimes stalls progress on ltsp because some obscure architectures don't keep up.
21:36
<lns>
vagrantc, but these obscure archs are keeping up the progress of debian as a distro at that point, right?
21:36
<vagrantc>
some architectures probably make no sense to have as thin clients (s390)
21:36
<lns>
lol
21:36
very true
21:37
i'm sure 95% use one of i386, ppc, amd64
21:37
<vagrantc>
lns: it hits ltsp harder than many, because it goes to the end of the queue for buildds
21:38
<lns>
why end of queue?
21:38
<vagrantc>
because it's priority: extra
21:38
<lns>
ah
21:38
<vagrantc>
i.e. bash, X, linux get higher priority
21:39
<lns>
vagrantc, but if you had your own h/w you could test it yourself right?
21:39
<vagrantc>
basically, 90% of the packages are higher priority...
21:39
<warren>
vagrantc: http://freshmeat.net/projects/wraplinux/?branch_id=72907&release_id=269862 tried wraplinux?
21:39
<vagrantc>
lns: yes, we could test, and even upload when absolutely necessary... but why bother?
21:39
<warren>
vagrantc: very new alternative to mkelfImage written by a kernel developer
21:39
from scratch
21:40
<vagrantc>
warren: just heard of it seconds ago, so haven't tried it, no :)
21:40* lns claps
21:41
<vagrantc>
warren: isn't H. Peter Anvin one of the authors of syslinux/pxelinux ?
21:41
<warren>
I think so
21:42
<vagrantc>
appears to be involved in all sorts of netbooting technologies
21:42
<lns>
vagrantc, bother because it'd get into debian faster and help the 'stalling of progress with ltsp' right?
21:42
<vagrantc>
lns: nope.
21:42
<warren>
wraplinux still doesn't work for the thincan coreboot + etherboot
21:42
but it did work in qemu/Etherboot
21:43
<lns>
i'd love to hunt down some crazy s390 and ship it to you if you could use it to help ltsp move along =)
21:43
<vagrantc>
lns: it won't improve the speed of the buildd's, which is really the primary issue
21:43
<warren>
so that's better than mknbi, mkelf and mkelfImage
21:43
there is something different about our very new kernel...
21:43
<vagrantc>
lns: s390 has never been the problem.
21:43
<lns>
maybe i'm not understanding.. speed of the builds?
21:43
<vagrantc>
lns: they're basically mainframe-like machines ... not really thin-clients.
21:43
<lns>
vagrantc, i know =) I used to support some crazy s390 type mainframe
21:44
<vagrantc>
lns: buildd's are machines that build (nearly) all packages for all architectures on debian.
21:44
<lns>
or at least one that used that protocol to the termemus on the workstations
21:44
vagrantc, oooh ok
21:44
<vagrantc>
lns: that's the main issue.
21:45
<lns>
vagrantc, so where can we obtain one of these buildds for ourselves =p
21:45
<vagrantc>
lns: it's getting official recognition for those buildds
21:45
lns: it's a beurocratic problem, not a technical one.
21:46
<lns>
vagrantc, i see
21:46
<vagrantc>
lns: you're willing to ask a lot of questions...
21:46
and i'm a sucker for answering them... :)
21:46
<lns>
vagrantc, i enjoy talking about ltsp with people other than who i support ;) I learn more about it that way
21:48
i've been a debian zealot for forever, and thin clients are just cool. i love working with them and figuring out how i can make money with them and at the same time give back as much as i can to the community
21:49
right now exposure to linux is the only thing i can think of that i'm really DOING...IANAP by a long shot. I would love to hire some people to contribute and create tools for ltsp5 that i know my clients can use, but i'm a ways off from being able to hire a programmer
21:49elisboa has quit IRC
21:50
<lns>
i'm a good visionary though =p and i always report/keep up on bugs as much as i can
21:51daya has joined #ltsp
21:59petre has quit IRC
22:12BGomes has joined #ltsp
22:17RiXtEr has joined #ltsp
22:30BGomes has quit IRC
22:32spectra has quit IRC
22:51captain_magnus has quit IRC
22:55captain_magnus has joined #ltsp
23:02soneyka has quit IRC
23:11RiXtEr has quit IRC
23:20soneyka has joined #ltsp
23:49subir has joined #ltsp
23:53cpunches has joined #ltsp